Need Help Understanding OC results for 'old' Celery not liking Win2K
I've got an old ABit ZM6 (PPGA360) MoBo and picked up an old Celeron
566 coppermine (cB0) which the board DOES support w/ last couple BIOSs (it's a PPGA board and this is an early FCPGA chip). Previously I had a 366(66) OC'd to 550(100) for years no prob. I dropped in the 566 and w/ 66FSB sure enough it works just fine at 1.5V as advertised. Now here's where I've run into some rather odd behavior given my limited experience w/ OC'ing - I suspect it may just not be OC'able but I'd be curious to understand it's behavior. Is there something special about Win2K at boot time that taxes a CPU perhaps? Basically it will post 850mhz all day long about 1.75V to 1.80V or better. It will boot into DOS(w95) easily as well using a boot disk. Here's the odd behavior: It is a ONE-TIME run session immediately after a BIOS Flash for Win2K specifically. It will never reboot again into Win2K. Sound strange? Let me explain further. Let's say I set the voltage to say 1.85 or 1.90V default (those are good popular voltages for an OC'd SL46T) - to do that I have to reflash the BIOS so the Abit will give me a proper 'default' voltage range. Immediately (and 1 time only) after a given reflash I can then boot it ONE time and ONE TIME ONLY at 850mhz(100FSB). That's IT! It will never reboot, restart or power-off and get back into Win2K at that speed. Doesn't matter what I do to modify the voltage further (I've gone as high as 2V). It will hang on booting into Win2K next go around period. I CAN get back into Win2K if I do a reflash of the BIOS as I would to say modify the default voltage. I am then granted one single full boot into Win2K. BUt I know of know other 'trick' to allow me back in a 2nd time... I CAN also get back into Win2K if I drop the FSB down to 66 for a stock 566 speed (and of course old DOS at 850mhz too). Once in Win2K it's golden at about 1.85V to 1.95V - any of those will yield seemingly long term stability and hours and hours of Prime95 w/ temps as high as 39C (1.95V). I can play games, do nothing, whatever - seems rock solid. But remember I CANNOT get BACK into Win2K a SECOND time if I do a restart or power it down. And also remember that this machine has run for years at 100FSB w/ a 366 OC'd to 550 under the same Win2K setup. (DO I need to do something special in Win2K when upping to a new CPU?) That just doesn't seem consistent w/ my experience w/ OCing. Even though I can trick it into 850mhz under Win2K thru the hassle of a reflash each time - obviously it's not worth the hassle so I'd have to say this one 'apparently' is not OC'able. But I'd like to better understand it's behavior as well as whether or not there is something about Win2K or other BIOS settings etc I should be aware of here. thanks! PS: Also the case reset button no longer functions w/ it set at 100FSB(850mhz) - usually I have to power cycle now too...? |
On Thu, 01 Jul 2004 12:06:05 -0500, pgtr wrote:
I've got an old ABit ZM6 (PPGA360) MoBo and picked up an old Celeron 566 coppermine (cB0) which the board DOES support .... Basically it will post 850mhz all day long about 1.75V to 1.80V or better. It will boot into DOS(w95) easily as well using a boot disk. Here's the odd behavior: It is a ONE-TIME run session immediately after a BIOS Flash for Win2K specifically. I CAN also get back into Win2K if I drop the FSB down to 66 for a stock 566 speed (and of course old DOS at 850mhz too). But remember I CANNOT get BACK into Win2K a SECOND time if I do a restart or power it down. And also remember that this machine has run for years at 100FSB w/ a 366 OC'd to 550 under the same Win2K setup. check for bad caps ... -- Regards, SPAJKY ® & visit my site @ http://www.spajky.vze.com "Tualatin OC-ed / BX-Slot1 / inaudible setup!" E-mail AntiSpam: remove ## |
pgtr wrote in message . ..
I CAN also get back into Win2K if I drop the FSB down to 66 for a stock 566 speed (and of course old DOS at 850mhz too). says it all. PS: Also the case reset button no longer functions w/ it set at 100FSB(850mhz) - usually I have to power cycle now too...? again, says it all. Seems like you been focussing too much on whats not relevant. Something needs slowing down. I've no idea what your speed options are on this combo but hope you can do something better than 566, if not 850. Regards, NT |
Spajky wrote:
On Thu, 01 Jul 2004 12:06:05 -0500, pgtr wrote: I've got an old ABit ZM6 (PPGA360) MoBo and picked up an old Celeron 566 coppermine (cB0) which the board DOES support .... Basically it will post 850mhz all day long about 1.75V to 1.80V or better. It will boot into DOS(w95) easily as well using a boot disk. Here's the odd behavior: It is a ONE-TIME run session immediately after a BIOS Flash for Win2K specifically. Try clearing CMOS and reentering your parameters. I CAN also get back into Win2K if I drop the FSB down to 66 for a stock 566 speed (and of course old DOS at 850mhz too). But remember I CANNOT get BACK into Win2K a SECOND time if I do a restart or power it down. Yes, well, simply saying "it's broke" doesn't provide many clues. What the heck does it DO when it doesn't 'get back into Win2K'? Hang? Crash? reboot? Spew some message at you? And at what point in the shutdown or boot? It runs win2K fine 'one-time'? (whatever the heck 'one-time' means). What was the purpose of the flash and what 'changes' does the update say it makes? That may have a clue. Set your memory timings to the slowest possible and see if that helps. And also remember that this machine has run for years at 100FSB w/ a 366 OC'd to 550 under the same Win2K setup. check for bad caps ... |
On Sun, 04 Jul 2004 18:11:09 -0500, David Maynard
wrote: I CAN also get back into Win2K if I drop the FSB down to 66 for a stock 566 speed (and of course old DOS at 850mhz too). But remember I CANNOT get BACK into Win2K a SECOND time if I do a restart or power it down. Yes, well, simply saying "it's broke" doesn't provide many clues. What the heck does it DO when it doesn't 'get back into Win2K'? Hang? Crash? reboot? Spew some message at you? And at what point in the shutdown or boot? It hangs. It usually hangs in the black 'F8' screen w/ the progress bar at the bottom that says it is starting windows. The progress bar at the bottom usually goes about 75-80% across when it locks up. The case reset switch has no affect at this point either - I have to power it off. It will not log a bootlog file either. It runs win2K fine 'one-time'? (whatever the heck 'one-time' means). Here's what I wrote before in anticipation of your question: It will never reboot again into Win2K. Sound strange? Let me explain further. Let's say I set the voltage to say 1.85 or 1.90V default (those are good popular voltages for an OC'd SL46T) - to do that I have to reflash the BIOS so the Abit will give me a proper 'default' voltage range. Immediately (and 1 time only) after a given reflash I can then boot it ONE time and ONE TIME ONLY at 850mhz(100FSB). That's IT! It will never reboot, restart or power-off and get back into Win2K at that speed. Doesn't matter what I do to modify the voltage further (I've gone as high as 2V). It will hang on booting into Win2K next go around period. I'll attempt to rephrase: It means after flashing the BIOS I can then successfully boot into W2K exactly ONE time. No more and no less. Once into W2K that initial time (after a flash) at 850mhz - runs just fine all day long - torture tests etc... But should I reboot or restart or shut down and later restart it will NOT run W2K again (see above). (until of course I go thru the reflash process). At this point I have to kick the FSB back from 100 to 75. I can boot to a 98 command prompt boot disk at 850mhz (100FSB) as well. Tomorrow I'll throw together 98 setup on a spare HD and try that to see if it can get into a full 98 windows environment. What was the purpose of the flash and what 'changes' does the update say it makes? That may have a clue. The purpose of flashing is to modify the default voltage range from 1.5V +-.2V to a higher default such as 1.7V (+-.2V) or greater. There are no 'changes' per se - I reflash w/ the last release BIOS. Set your memory timings to the slowest possible and see if that helps. Long since set to slowest most conservative settings. Everythign in BIOS is set conservatively. Remember this all worked fine w/ a 100FSB setting w/ a 366 OCd to 550. The 100FSB itself in thoery shouldn't be a problem here I would think...? And also remember that this machine has run for years at 100FSB w/ a 366 OC'd to 550 under the same Win2K setup. check for bad caps ... |
Can you expand on how to check for bad caps?
thanks, On Sun, 04 Jul 2004 10:46:53 +0200, Spajky wrote: On Thu, 01 Jul 2004 12:06:05 -0500, pgtr wrote: I've got an old ABit ZM6 (PPGA360) MoBo and picked up an old Celeron 566 coppermine (cB0) which the board DOES support .... Basically it will post 850mhz all day long about 1.75V to 1.80V or better. It will boot into DOS(w95) easily as well using a boot disk. Here's the odd behavior: It is a ONE-TIME run session immediately after a BIOS Flash for Win2K specifically. I CAN also get back into Win2K if I drop the FSB down to 66 for a stock 566 speed (and of course old DOS at 850mhz too). But remember I CANNOT get BACK into Win2K a SECOND time if I do a restart or power it down. And also remember that this machine has run for years at 100FSB w/ a 366 OC'd to 550 under the same Win2K setup. check for bad caps ... |
|
pgtr wrote:
On Sun, 04 Jul 2004 18:11:09 -0500, David Maynard wrote: I CAN also get back into Win2K if I drop the FSB down to 66 for a stock 566 speed (and of course old DOS at 850mhz too). But remember I CANNOT get BACK into Win2K a SECOND time if I do a restart or power it down. Yes, well, simply saying "it's broke" doesn't provide many clues. What the heck does it DO when it doesn't 'get back into Win2K'? Hang? Crash? reboot? Spew some message at you? And at what point in the shutdown or boot? It hangs. It usually hangs in the black 'F8' screen w/ the progress bar at the bottom that says it is starting windows. The progress bar at the bottom usually goes about 75-80% across when it locks up. The case reset switch has no affect at this point either - I have to power it off. It will not log a bootlog file either. The case reset switch not working is particularly disturbing. Does it work on the 'first' boot? I mean, instead of trying a reboot can you kill it by hitting reset? It runs win2K fine 'one-time'? (whatever the heck 'one-time' means). Here's what I wrote before in anticipation of your question: It will never reboot again into Win2K. Sound strange? Let me explain further. Let's say I set the voltage to say 1.85 or 1.90V default (those are good popular voltages for an OC'd SL46T) - to do that I have to reflash the BIOS so the Abit will give me a proper 'default' voltage range. Immediately (and 1 time only) after a given reflash I can then boot it ONE time and ONE TIME ONLY at 850mhz(100FSB). That's IT! It will never reboot, restart or power-off and get back into Win2K at that speed. Doesn't matter what I do to modify the voltage further (I've gone as high as 2V). It will hang on booting into Win2K next go around period. I'll attempt to rephrase: It means after flashing the BIOS I can then successfully boot into W2K exactly ONE time. No more and no less. Once into W2K that initial time (after a flash) at 850mhz - runs just fine all day long - torture tests etc... But should I reboot or restart or shut down and later restart it will NOT run W2K again (see above). (until of course I go thru the reflash process). At this point I have to kick the FSB back from 100 to 75. I can boot to a 98 command prompt boot disk at 850mhz (100FSB) as well. Tomorrow I'll throw together 98 setup on a spare HD and try that to see if it can get into a full 98 windows environment. What was the purpose of the flash and what 'changes' does the update say it makes? That may have a clue. The purpose of flashing is to modify the default voltage range from 1.5V +-.2V to a higher default such as 1.7V (+-.2V) or greater. There are no 'changes' per se - I reflash w/ the last release BIOS. Yeah. Ok. If you'd mentioned Abit and flashing for Vcore I'd have known what you meant. I've never heard of a problem quite like this but it's as if the motherboard doesn't like the foolie flash. Which Abit motherboard is it? I presume it's one of their socket 370 boards because if you're using a slotket you could adjust Vcore there and not bother with the flash. Well, unless you're using one of the el-cheapo slotkets with no Vcore jumpers. Set your memory timings to the slowest possible and see if that helps. Long since set to slowest most conservative settings. Everythign in BIOS is set conservatively. Remember this all worked fine w/ a 100FSB setting w/ a 366 OCd to 550. The 100FSB itself in thoery shouldn't be a problem here I would think...? Well, it could but, in this case, I doubt that's it since it works 'forever' until you try to reboot. Just because it worked with a 366 OC'd, though, doesn't mean a 566 'must'. On the other hand, that overclock isn't in any way unusual. The only significant difference I can think of, off hand, is the 366 being a ppga but the 566 a COPPERMINE. What stepping is it? The B stepping is socket compatible with ppga but I think there were some differences with the C. I don't have an answer but some thoughts. It seems to be either BIOS or coppermine related and not the overclock, per see, since it runs fine the one time. The reboot failure sounds similar to the reset switch not working since the last thing a reboot does is essentially a reset. Which wouldn't seem to explain not working a second time from a cold start EXCEPT a cold start 'starts' with reset (By reset I mean to include setting up the processor registers). So then the question would be, why does it work the first time at all? And, with that, it starts at 66, so it had a reset, you flash, reboot, set Vcore/FSB, and then run 100. Have you tried powering it off RIGHT after you set 100MHz FSB, but no boot into Windows, to see if it will THEN run the 'one time' from a cold start at 100MHz FSB? Doesn't seem to explain why it would run in DOS, though, unless whatever is 'missing' from the reset is of no consequence to DOS. |
pgtr wrote in message . ..
On 4 Jul 2004 10:36:17 -0700, (N. Thornton) wrote: pgtr wrote in message . .. I CAN also get back into Win2K if I drop the FSB down to 66 for a stock 566 speed (and of course old DOS at 850mhz too). says it all. Can you expand more on what all that says? I'm no overclocking genius but if it all works dandy at one speed and frequently hangs at 30% faster speed... hello? PS: Also the case reset button no longer functions w/ it set at 100FSB(850mhz) - usually I have to power cycle now too...? again, says it all. Seems like you been focussing too much on whats not relevant. Something needs slowing down. I've no idea what your speed options are on this combo but hope you can do something better than 566, if not 850. I've got the thing as conservatively set up as possible. Is a 566 running at 850 conservative? The only real variable here that I'm aware of is the FSB - can you suggest what I appear to be missing? the obvious? The chip is locked at 8.5 multiplier. It runs fine at 66FSB (it's stock speed). It will run fine at 75FSB (a mild OC of 637mhz). It will run fine at 100FSB one time immediately after a BIOS flash. It will also run fine at 850mhz(100FSB) repeatedly if using a boot disk into DOS or Win98 command prompt. (I'm going to check a full 98 setup later). So far the indications are that the COMBINATION of W2K and 850 on this chip are mutually exclusive. Also all the H/W has worked fine in the past w/ Win2K at 100FSB (using a 366 OC'd to 550). Can you offer some suggestions on what is relevant to try? DOS / 98 command prompt is small and simple compared to win98, and thus far more likely to boot than full Win when you have data errors going on. This is to be expected. I dont know I could always be wrong, but it seems quite obvious that youre trying to run it at a speed it cant do. What to do? Run it at a speed it will do! What else? Youve tried the tricks, tweaking the voltage... thats it. Your max speed depends not just on CPU or mobo, but on the pair of them together. If the CPU will run faster in another mobo, you might contemplate using the other mobo. Regards, NT |
On Wed, 07 Jul 2004 02:17:42 -0500, David Maynard
wrote: SNIP It hangs. It usually hangs in the black 'F8' screen w/ the progress bar at the bottom that says it is starting windows. The progress bar at the bottom usually goes about 75-80% across when it locks up. The case reset switch has no affect at this point either - I have to power it off. It will not log a bootlog file either. The case reset switch not working is particularly disturbing. Does it work on the 'first' boot? I mean, instead of trying a reboot can you kill it by hitting reset? Let me expand on that. The 'reset' switch on the case normally works fine and of course is only needed in a dire situation (which is super rare or almost never for me under W2K). If I OC to 850mhz and do the flash trick I can get into W2K. At that point I've never tried the reset swich because there was no need. The PC functioned normally and I could then do a normal restart or shut down thru windows. The NEXT time into W2K it will of course hang usually on the black F8 windows starting phase... Now HERE if I hit the reset button it will blank the screen, you'll here some ticks coming from the HD as you normally would but then it just sits there w/ a black screen indefinately. I simply have to turn the power off and then back on to get it to respond again. I CAN however put a boot disk in and boot (at 850mhz) into a DOS prompt (W98 book disk) and the reset button works there as well as CNTRL ALT DEL. Basically when it hangs on teh 2nd attempt and beyond going into W2K at 850mhz - it REALLY hangs and not even the reset button will wake it up properly - just a power off/on. Yeah. Ok. If you'd mentioned Abit and flashing for Vcore I'd have known what you meant. Yep. I thought I'd mentioned in the orig post it was an ABit ZM6 (similar to BM6 but less memory). You know your mobos! I have the latest flashing they released (SU?) to move the default voltage up... I think the chip is pretty happy at 1.85V which is right in line w/ the many entries in overclockers.com CPU database for OCing the 566 to 850. I've never heard of a problem quite like this but it's as if the motherboard doesn't like the foolie flash. Which Abit motherboard is it? I presume it's one of their socket 370 boards because if you're using a slotket you could adjust Vcore there and not bother with the flash. Well, unless you're using one of the el-cheapo slotkets with no Vcore jumpers. Yes again - you do know those old mobos! The ZM6 w/ the last couple of flashings (QU? and SU?) will support the earliest coppermine CPUs including explicitly the cB0 stepping up to 600mhz - this chip is a 566 cB0 stepping. No slotket or adapter. Set your memory timings to the slowest possible and see if that helps. Long since set to slowest most conservative settings. Everythign in BIOS is set conservatively. Remember this all worked fine w/ a 100FSB setting w/ a 366 OCd to 550. The 100FSB itself in thoery shouldn't be a problem here I would think...? Well, it could but, in this case, I doubt that's it since it works 'forever' until you try to reboot. Indeed. Forever: Several days at a time including extended overnight sessions w/ prime95 torture test. No apparent temp problems or stability issues. It just runs fantastic at 850mhz that ONE time after a flash under W2K. Its the 2nd time and beyond that it hangs... As I said never quite heard of something like this in my limited OCing experience. Strange! I got lucky in stumbling into the fact that it would work the one time after a reflash - if it wasn't for that I might very well have concluded it was not do-able. Just because it worked with a 366 OC'd, though, doesn't mean a 566 'must'. On the other hand, that overclock isn't in any way unusual. No it doesn't. But 366 at 550 tells me that the system is stable and happy w/ a 100FSB. And the 566 CAN run nicely at 850 indefinately that 'one' time after a flash. It can also run at 850 consistently via a DOS prompt boot. Sure seems like the ducks are all in a row for it to OC at 850mhz but throw in W2K and ...? The only significant difference I can think of, off hand, is the 366 being a ppga but the 566 a COPPERMINE. What stepping is it? The B stepping is socket compatible with ppga but I think there were some differences with the C. Yep - see above - it's a cB0 which is supported by ABit ZM6 w/ the last couple of BIOS releases. I don't have an answer but some thoughts. It seems to be either BIOS or coppermine related and not the overclock, per see, since it runs fine the one time. The reboot failure sounds similar to the reset switch not working since the last thing a reboot does is essentially a reset. Which wouldn't seem to explain not working a second time from a cold start EXCEPT a cold start 'starts' with reset (By reset I mean to include setting up the processor registers). So then the question would be, why does it work the first time at all? And, with that, it starts at 66, so it had a reset, you flash, reboot, set Vcore/FSB, and then run 100. Have you tried powering it off RIGHT after you set 100MHz FSB, but no boot into Windows, to see if it will THEN run the 'one time' from a cold start at 100MHz FSB? No I haven't tried that particular combination. What might that tells us one way or the other? To be honest going thru the flashing process is a little tricky and sometimes it doesn't take or preserve the voltages correctly and may take a 2nd or 3rd attempt (I've had problems w/ it reverting from 1.7V to 1.5V default on subsequent flashes so I have to do it twice). Even though I have a UPS we have lots of power fluctuations out here and to be honest I'm just plain nervous about doing any more flashes. ;) Doesn't seem to explain why it would run in DOS, though, unless whatever is 'missing' from the reset is of no consequence to DOS. I'm leaning towards something odd going on w/ W2K that seems to upset the applecart so to speak w/ regards to: BIOS or COppermine or something...? I believe the chip posts consistently at 850. It doesn't have any temperature related stability problems at 850. W2K is happy w/ the chip at 66FSB and 75FSB. There is a setting as I recall in the BIOS that is something like 'Force Update ESCD' - could that have any impact? I'm going to hit the jumper on the mobo and clear the CMOS (may need to reflash to get back to 1.7V default?) and see what that does. I'm also going to grab an extra small HD and put a Win98 install on there and swap out the two large HDs w/ W2K for this one temporarily - I'd like to see if it will boot consistently into a full Win98 at 850... I don't know how much it says that it can boot into a W98 dos prompt at 850 consistently - it is only a DOS prompt as opposed to the full W98. Basically at this point I'm sure the chip will post at 850 consistently. Also that it doesn't seem to have any temperature related stability problems (based on 'successfuly' initial boots after flash). Thanks, |
|
N. Thornton wrote:
pgtr wrote in message . .. On 4 Jul 2004 10:36:17 -0700, (N. Thornton) wrote: pgtr wrote in message . .. I CAN also get back into Win2K if I drop the FSB down to 66 for a stock 566 speed (and of course old DOS at 850mhz too). says it all. Can you expand more on what all that says? I'm no overclocking genius but if it all works dandy at one speed and frequently hangs at 30% faster speed... hello? It means there's a problem. Your assumption that it also means the processor can't handle it isn't necessarily the case, however. And your description is misleading. 'Frequently', with no other description, suggests it's semi 'random', that it 'just happens' some times, but there's a definite, repeatable, known set of circumstances that causes it to hang at a single, known, spot while it runs perfectly fine for days on end under the right circumstances. PS: Also the case reset button no longer functions w/ it set at 100FSB(850mhz) - usually I have to power cycle now too...? again, says it all. Seems like you been focussing too much on whats not relevant. Something needs slowing down. I've no idea what your speed options are on this combo but hope you can do something better than 566, if not 850. I've got the thing as conservatively set up as possible. Is a 566 running at 850 conservative? Depends on how you define 'conservative' and in what context. It isn't 'unusual' for a 566 to run 850, NOT running 850 would be the unusual case, and the CB0 566 I had ran 1020/120MHz FSB (properly torture tested). In that context, 850 does look rather 'conservative'. The only real variable here that I'm aware of is the FSB - can you suggest what I appear to be missing? the obvious? I think the problem here with 'obvious' is you seem to be operating under the theory that if a 'problem' occurs then it's 'broke' so give up (the 'obvious' answer) whereas we're looking at what others have done, and the statistical success rate, and wondering why this one seems to be so oddball compared to what virtually every other 566 can do; then looking at the symptoms to see if we can find out why. You may be right, but it's not the odds. The chip is locked at 8.5 multiplier. It runs fine at 66FSB (it's stock speed). It will run fine at 75FSB (a mild OC of 637mhz). It will run fine at 100FSB one time immediately after a BIOS flash. It will also run fine at 850mhz(100FSB) repeatedly if using a boot disk into DOS or Win98 command prompt. (I'm going to check a full 98 setup later). So far the indications are that the COMBINATION of W2K and 850 on this chip are mutually exclusive. Also all the H/W has worked fine in the past w/ Win2K at 100FSB (using a 366 OC'd to 550). Can you offer some suggestions on what is relevant to try? DOS / 98 command prompt is small and simple compared to win98, and thus far more likely to boot than full Win when you have data errors going on. This is to be expected. I doubt DOS would boot with flat out 'errors'. It isn't as if DOS has 'spare' program material that can get corrupted with no effect. I dont know I could always be wrong, but it seems quite obvious that youre trying to run it at a speed it cant do. What to do? Run it at a speed it will do! What else? Find what the problem is. Youve tried the tricks, tweaking the voltage... thats it. Your max speed depends not just on CPU or mobo, but on the pair of them together. If the CPU will run faster in another mobo, you might contemplate using the other mobo. That's a possibility but I'm more inclined to think it has something to do with the foolie flash. Regards, NT |
pgtr wrote:
On Wed, 07 Jul 2004 02:17:42 -0500, David Maynard wrote: SNIP It hangs. It usually hangs in the black 'F8' screen w/ the progress bar at the bottom that says it is starting windows. The progress bar at the bottom usually goes about 75-80% across when it locks up. The case reset switch has no affect at this point either - I have to power it off. It will not log a bootlog file either. The case reset switch not working is particularly disturbing. Does it work on the 'first' boot? I mean, instead of trying a reboot can you kill it by hitting reset? Let me expand on that. The 'reset' switch on the case normally works fine and of course is only needed in a dire situation (which is super rare or almost never for me under W2K). If I OC to 850mhz and do the flash trick I can get into W2K. At that point I've never tried the reset swich because there was no need. The PC functioned normally and I could then do a normal restart or shut down thru windows. That is what I presumed was the case and is why I asked about trying the reset switch when the system is in the 'working' state. The NEXT time into W2K it will of course hang usually on the black F8 windows starting phase... Now HERE if I hit the reset button it will blank the screen, you'll here some ticks coming from the HD as you normally would but then it just sits there w/ a black screen indefinately. I simply have to turn the power off and then back on to get it to respond again. WELLLlllllll, now, that's a sight different than the original description of "reset switch has no affect (sic)." It IS apparently doing 'something'... it blanks the screen and the hard drive response indicates it got a hard reset. The motherboard apparently isn't 'starting back up', from the reset, though. I CAN however put a boot disk in and boot (at 850mhz) into a DOS prompt (W98 book disk) and the reset button works there as well as CNTRL ALT DEL. It's beginning to sound like a BIOS power management issue. APM, ACPI, nothing, conflicting, confused, I don't know. Of course, DOS couldn't care less about APM/ACPI but Windows 2000 sure as heck does and it's going to be very angry about having the wrong kernel (APM, ACPI, etc) vs what the motherboard claims. Basically when it hangs on teh 2nd attempt and beyond going into W2K at 850mhz - it REALLY hangs and not even the reset button will wake it up properly - just a power off/on. Yes, ok. The rest doesn't cause a proper reboot but, according to your current description, it IS doing something. Yeah. Ok. If you'd mentioned Abit and flashing for Vcore I'd have known what you meant. Yep. I thought I'd mentioned in the orig post it was an ABit ZM6 (similar to BM6 but less memory). You know your mobos! I have the latest flashing they released (SU?) to move the default voltage up... I think the chip is pretty happy at 1.85V which is right in line w/ the many entries in overclockers.com CPU database for OCing the 566 to 850. Yeah, I don't think it's the voltage, per see, or else the symptom would be more random. How does the 566 perform at 850 if you use the max voltage the 'normal' default range allows? I've never heard of a problem quite like this but it's as if the motherboard doesn't like the foolie flash. Which Abit motherboard is it? I presume it's one of their socket 370 boards because if you're using a slotket you could adjust Vcore there and not bother with the flash. Well, unless you're using one of the el-cheapo slotkets with no Vcore jumpers. Yes again - you do know those old mobos! The ZM6 w/ the last couple of flashings (QU? and SU?) will support the earliest coppermine CPUs including explicitly the cB0 stepping up to 600mhz - this chip is a 566 cB0 stepping. No slotket or adapter. Set your memory timings to the slowest possible and see if that helps. Long since set to slowest most conservative settings. Everythign in BIOS is set conservatively. Remember this all worked fine w/ a 100FSB setting w/ a 366 OCd to 550. The 100FSB itself in thoery shouldn't be a problem here I would think...? Well, it could but, in this case, I doubt that's it since it works 'forever' until you try to reboot. Indeed. Forever: Several days at a time including extended overnight sessions w/ prime95 torture test. No apparent temp problems or stability issues. It just runs fantastic at 850mhz that ONE time after a flash under W2K. Its the 2nd time and beyond that it hangs... As I said never quite heard of something like this in my limited OCing experience. Strange! I got lucky in stumbling into the fact that it would work the one time after a reflash - if it wasn't for that I might very well have concluded it was not do-able. Just because it worked with a 366 OC'd, though, doesn't mean a 566 'must'. On the other hand, that overclock isn't in any way unusual. No it doesn't. But 366 at 550 tells me that the system is stable and happy w/ a 100FSB. And the 566 CAN run nicely at 850 indefinately that 'one' time after a flash. It can also run at 850 consistently via a DOS prompt boot. Sure seems like the ducks are all in a row for it to OC at 850mhz but throw in W2K and ...? Which kernel DO you have in win2000? ACPI? 'Standard pc'? The only significant difference I can think of, off hand, is the 366 being a ppga but the 566 a COPPERMINE. What stepping is it? The B stepping is socket compatible with ppga but I think there were some differences with the C. Yep - see above - it's a cB0 which is supported by ABit ZM6 w/ the last couple of BIOS releases. Yes. Well, CB0 should be plug in compatible with ppga anyway. I don't have an answer but some thoughts. It seems to be either BIOS or coppermine related and not the overclock, per see, since it runs fine the one time. The reboot failure sounds similar to the reset switch not working since the last thing a reboot does is essentially a reset. Which wouldn't seem to explain not working a second time from a cold start EXCEPT a cold start 'starts' with reset (By reset I mean to include setting up the processor registers). So then the question would be, why does it work the first time at all? And, with that, it starts at 66, so it had a reset, you flash, reboot, set Vcore/FSB, and then run 100. Have you tried powering it off RIGHT after you set 100MHz FSB, but no boot into Windows, to see if it will THEN run the 'one time' from a cold start at 100MHz FSB? No I haven't tried that particular combination. What might that tells us one way or the other? To be honest going thru the flashing process is a little tricky and sometimes it doesn't take or preserve the voltages correctly and may take a 2nd or 3rd attempt (I've had problems w/ it reverting from 1.7V to 1.5V default on subsequent flashes so I have to do it twice). Even though I have a UPS we have lots of power fluctuations out here and to be honest I'm just plain nervous about doing any more flashes. ;) Well, I was trying to establish whether the reset ever worked 'right' with win2000 but I'm now on the power management tract. What *I* would do, at this stage, since the Vcore range is your big issue, is wire strap the processor pins for a higher Vcore so the BIOS just naturally thinks it should be higher: no flash required. Pulling VID3, that's AJ37, to ground (AK36, Vss, is right next to it) will give you 1.9V default (just about your only choice short of insulating/pulling pins), which you can then lower to the 1.85 you're using. That is, if the default core is 1.5, as I think you said. If it's higher then jumpering AJ37 low will result in a correspondingly higher Vcore above 1.9. Doesn't seem to explain why it would run in DOS, though, unless whatever is 'missing' from the reset is of no consequence to DOS. I'm leaning towards something odd going on w/ W2K that seems to upset the applecart so to speak w/ regards to: BIOS or COppermine or something...? I believe the chip posts consistently at 850. It doesn't have any temperature related stability problems at 850. W2K is happy w/ the chip at 66FSB and 75FSB. There is a setting as I recall in the BIOS that is something like 'Force Update ESCD' - could that have any impact? I'm going to hit the jumper on the mobo and clear the CMOS (may need to reflash to get back to 1.7V default?) and see what that does. Clearing BIOS is a grand idea with all the flashes you've done. I'm also going to grab an extra small HD and put a Win98 install on there and swap out the two large HDs w/ W2K for this one temporarily - I'd like to see if it will boot consistently into a full Win98 at 850... I don't know how much it says that it can boot into a W98 dos prompt at 850 consistently - it is only a DOS prompt as opposed to the full W98. Basically at this point I'm sure the chip will post at 850 consistently. Also that it doesn't seem to have any temperature related stability problems (based on 'successfuly' initial boots after flash). Try the VID jumper so you don't have to flash. |
pgtr wrote in message . ..
On 7 Jul 2004 04:33:41 -0700, (N. Thornton) wrote: I apologise for being a bit rude. I'm no overclocking genius but if it all works dandy at one speed and frequently hangs at 30% faster speed... hello? It hangs specifically at startup under one specific OS in a consistent fashion. THis largely eliminates the typical OC issues of not posting and temperature stability problems and leaves some sort of configuration issue. POST failure, startup hang and instability are all caused by the same thing: data errors. Data errors are what you get with an _unstably_ overclocked system. Not POSTing is merely a more extreme data error problem than hang on boot. Crashes after startup are the same issue again, merely with a lower data error rate. The difference merely lies in when the error occurred: during POST, during startup, or after. The level of errors will determine which. 1 error per 0.1 secs will cause POST fail, 1 error per hour will cause crashes during uptime. This is why DOS will often boot when Win wont: far less work = less chance of error during startup. So the picture leaves us with data errors, not configuration problems. PS: Also the case reset button no longer functions w/ it set at 100FSB(850mhz) - usually I have to power cycle now too...? again, says it all. Seems like you been focussing too much on whats not relevant. Something needs slowing down. I've no idea what your speed options are on this combo but hope you can do something better than 566, if not 850. I've got the thing as conservatively set up as possible. Is a 566 running at 850 conservative? To be more specific I was referring to all BIOS settings - they are as conservative as possible. 850 still isnt conservative. Sorry :) However a 566 at 850 is below average. The CPU database suggests 874 as average. Anything above that up to say 1ghz is aggressive. The 566 is generally regarded as one of the more OCable chips of that ilk much like the 300A back in it's day. 566s may love o/cing, but that doesnt mean every 566 will do way above 566. Inevitably some wont, its always the way. I agree 1G would be aggressive :) The chip is locked at 8.5 multiplier. It runs fine at 66FSB (it's stock speed). It will run fine at 75FSB (a mild OC of 637mhz). It will run fine at 100FSB one time immediately after a BIOS flash. It will also run fine at 850mhz(100FSB) repeatedly if using a boot disk into DOS or Win98 command prompt. (I'm going to check a full 98 setup later). So far the indications are that the COMBINATION of W2K and 850 on this chip are mutually exclusive. Also all the H/W has worked fine in the past w/ Win2K at 100FSB (using a 366 OC'd to 550). Can you offer some suggestions on what is relevant to try? DOS / 98 command prompt is small and simple compared to win98, and thus far more likely to boot than full Win when you have data errors going on. This is to be expected. Can you expand on "data errors" that might prevent W2K from booting and hang? have now done. Hopefully clearly. I dont know I could always be wrong, but it seems quite obvious that youre trying to run it at a speed it cant do. What to do? Run it at a speed it will do! What else? That's not exactly correct - I can reflash the BIOS each time at boot up and it can CONSISTENTLY run at 850 indefinately and be very stable. I am looking for: A) an explanation as to what is special about running W2K immediately after a flash that allows this B) a solution to booting it at 850mhz into W2K w/o having to reflash it each time I know thats a teaser, but a) its not something you can change AFAIK, unles you try older BIOS software versions on the remote offchance one might work. b) the plain stump-like fact is it runs reliably at 566 but not consistently at 850. Youve tried the tricks, tweaking the voltage... thats it. Your max speed depends not just on CPU or mobo, but on the pair of them together. If the CPU will run faster in another mobo, you might contemplate using the other mobo. Could there not be some driver or configuration issues w/ W2K? I guess there one way to find out: put 98se on it, or any other 'as different to 2k as poss' large OS. But if win2k starts up fine at 566 but not at 850, you have a clear answer: data errors at 850. Regards, NT |
On Wed, 07 Jul 2004 17:52:58 -0500, David Maynard
wrote: SNIP WELLLlllllll, now, that's a sight different than the original description of "reset switch has no affect (sic)." It IS apparently doing 'something'... it blanks the screen and the hard drive response indicates it got a hard reset. The motherboard apparently isn't 'starting back up', from the reset, though. Yes BUT remember only if it was at 850. If it's at 637 or 566 it stops and starts fine inclduing the reset button on the case. I CAN however put a boot disk in and boot (at 850mhz) into a DOS prompt (W98 book disk) and the reset button works there as well as CNTRL ALT DEL. It's beginning to sound like a BIOS power management issue. APM, ACPI, nothing, conflicting, confused, I don't know. Of course, DOS couldn't care less about APM/ACPI but Windows 2000 sure as heck does and it's going to be very angry about having the wrong kernel (APM, ACPI, etc) vs what the motherboard claims. Hmmmmmm I might double check and see what BIOS is set at. I've done so many flashes that I've stopped bothered to modify much of anything other than the CPU and maybe the setting for PnP OS. Everything else iincluding those power settings in BIOS are all left at whatever the default is. Basically when it hangs on teh 2nd attempt and beyond going into W2K at 850mhz - it REALLY hangs and not even the reset button will wake it up properly - just a power off/on. Yes, ok. The rest doesn't cause a proper reboot but, according to your current description, it IS doing something. Well it 'tries' to restart and acts like it is, but never gets beyond a blank screen so I power off. Again, only at 850 though not at 566 or 637. Yeah. Ok. If you'd mentioned Abit and flashing for Vcore I'd have known what you meant. Yep. I thought I'd mentioned in the orig post it was an ABit ZM6 (similar to BM6 but less memory). You know your mobos! I have the latest flashing they released (SU?) to move the default voltage up... I think the chip is pretty happy at 1.85V which is right in line w/ the many entries in overclockers.com CPU database for OCing the 566 to 850. Yeah, I don't think it's the voltage, per see, or else the symptom would be more random. How does the 566 perform at 850 if you use the max voltage the 'normal' default range allows? The max teh default allows is 1.5 +-.2 or 1.7V. That just wasn't enough as I recall. It's real happy at about 1.8 to 1.9. I don't recall how it did at 1.75V offhand but I think it might also work there... SNIP Just because it worked with a 366 OC'd, though, doesn't mean a 566 'must'. On the other hand, that overclock isn't in any way unusual. No it doesn't. But 366 at 550 tells me that the system is stable and happy w/ a 100FSB. And the 566 CAN run nicely at 850 indefinately that 'one' time after a flash. It can also run at 850 consistently via a DOS prompt boot. Sure seems like the ducks are all in a row for it to OC at 850mhz but throw in W2K and ...? Which kernel DO you have in win2000? ACPI? 'Standard pc'? YOu would ask. How do I tell? In Control Panel: Power Options it shows an APM tab. In that tab it is NOT checked. Under Device Mgr if I view hidden devices the NT/APM Legacy driver shows itself but w/ a red X - it's disabled. Let me know what to change or check for and I'll definately go for it. No I haven't tried that particular combination. What might that tells us one way or the other? To be honest going thru the flashing process is a little tricky and sometimes it doesn't take or preserve the voltages correctly and may take a 2nd or 3rd attempt (I've had problems w/ it reverting from 1.7V to 1.5V default on subsequent flashes so I have to do it twice). Even though I have a UPS we have lots of power fluctuations out here and to be honest I'm just plain nervous about doing any more flashes. ;) Well, I was trying to establish whether the reset ever worked 'right' with win2000 but I'm now on the power management tract. FWIW the reset does work fine @ 566 and 637 (75fsb) and did w/ the old 366 @ 550(100fsb) as well. What *I* would do, at this stage, since the Vcore range is your big issue, is wire strap the processor pins for a higher Vcore so the BIOS just naturally thinks it should be higher: no flash required. Pulling VID3, that's AJ37, to ground (AK36, Vss, is right next to it) will give you 1.9V default (just about your only choice short of insulating/pulling pins), which you can then lower to the 1.85 you're using. That is, if the default core is 1.5, as I think you said. If it's higher then jumpering AJ37 low will result in a correspondingly higher Vcore above 1.9. So you're saying soldering a wire between AJ37 and AK36? Would this best be done from under the mobo? Here's a link to an interesting article that talks about modifying the same ZM6 mobo for later coppermines - it's where I picked up the flash trick. http://www.3feetunder.com/krick/370mod/ I need to find a better diagram of the celeron as that one is too small/blurry - probably at intel.com I can find a better one. Any other easy/quickee tricks to jump AJ37 and AK36? I don't mind putting something around a pint to insulate it (like a piece of insulation from an old ribbon cable). Bit more of a hassle to pull everything apart and solder from behind - small for me to solder but I think doable for me. Doesn't seem to explain why it would run in DOS, though, unless whatever is 'missing' from the reset is of no consequence to DOS. I'm leaning towards something odd going on w/ W2K that seems to upset the applecart so to speak w/ regards to: BIOS or COppermine or something...? I believe the chip posts consistently at 850. It doesn't have any temperature related stability problems at 850. W2K is happy w/ the chip at 66FSB and 75FSB. There is a setting as I recall in the BIOS that is something like 'Force Update ESCD' - could that have any impact? I'm going to hit the jumper on the mobo and clear the CMOS (may need to reflash to get back to 1.7V default?) and see what that does. Clearing BIOS is a grand idea with all the flashes you've done. Just did it today at lunch and sure enough it kicked me back to 1.5V. SO I figured I'd reflash w/ /cc and have been running 850mhz all afternoon (including a torture test for kicks). I haven't had teh heart to restart it figuring I'll probably lose the 850 yet again... I'm also going to grab an extra small HD and put a Win98 install on there and swap out the two large HDs w/ W2K for this one temporarily - I'd like to see if it will boot consistently into a full Win98 at 850... I don't know how much it says that it can boot into a W98 dos prompt at 850 consistently - it is only a DOS prompt as opposed to the full W98. Basically at this point I'm sure the chip will post at 850 consistently. Also that it doesn't seem to have any temperature related stability problems (based on 'successfuly' initial boots after flash). Try the VID jumper so you don't have to flash. Yeah I understand what you are thinking but I need to think that one thru. FYI the 'default' chips only cycle up to 600(66). I also still want to load up W98 on a spare HD and see what happens. If it works that would isolate the problem as unique to W2K combined w/ the OC'd chip. And I will need to (reluctantly) reboot here shortly and verify it will NOT come up @ 850 a second time in W2K. |
N. Thornton wrote:
pgtr wrote in message . .. On 7 Jul 2004 04:33:41 -0700, (N. Thornton) wrote: I apologise for being a bit rude. I'm no overclocking genius but if it all works dandy at one speed and frequently hangs at 30% faster speed... hello? It hangs specifically at startup under one specific OS in a consistent fashion. THis largely eliminates the typical OC issues of not posting and temperature stability problems and leaves some sort of configuration issue. POST failure, He doesn't HAVE a "post failure." startup hang and instability are all caused by the same thing: data errors. Data errors are what you get with an _unstably_ overclocked system. Simply not true. A bad driver can cause a system hang. The wrong O.S. kernel can cause a system hang. There are LOTS of things that can cause hangs and 'instability' and that have nothing to do with an overclock. Not POSTing is merely a more extreme data error problem than hang on boot. Crashes after startup are the same issue again, merely with a lower data error rate. Again, not true, and for many of the same reasons. "Not posting" is NOT "merely a more extreme data error problem" unless by "data error" you simply mean 'something went wrong'. The difference merely lies in when the error occurred: during POST, during startup, or after. 'The difference' is what one uses to debug the cause of the problem. The level of errors will determine which. 1 error per 0.1 secs will cause POST fail, 1 error per hour will cause crashes during uptime. This is why DOS will often boot when Win wont: far less work = less chance of error during startup. I'd love to see where you came up with those error rates. So the picture leaves us with data errors, not configuration problems. PS: Also the case reset button no longer functions w/ it set at 100FSB(850mhz) - usually I have to power cycle now too...? again, says it all. Seems like you been focussing too much on whats not relevant. Something needs slowing down. I've no idea what your speed options are on this combo but hope you can do something better than 566, if not 850. I've got the thing as conservatively set up as possible. Is a 566 running at 850 conservative? To be more specific I was referring to all BIOS settings - they are as conservative as possible. 850 still isnt conservative. Sorry :) In your opinion. However a 566 at 850 is below average. The CPU database suggests 874 as average. Anything above that up to say 1ghz is aggressive. The 566 is generally regarded as one of the more OCable chips of that ilk much like the 300A back in it's day. 566s may love o/cing, but that doesnt mean every 566 will do way above 566. Inevitably some wont, its always the way. I agree 1G would be aggressive :) True. However, the fact that, on the first boot, it will run perfectly fine till the cows come home suggests there is no problem at all with his 566 running 850. There are no crashes, random reboots (he'd probably love to see one), or any other instability symptoms. There are no, as you put it, 'data errors'. The chip is locked at 8.5 multiplier. It runs fine at 66FSB (it's stock speed). It will run fine at 75FSB (a mild OC of 637mhz). It will run fine at 100FSB one time immediately after a BIOS flash. It will also run fine at 850mhz(100FSB) repeatedly if using a boot disk into DOS or Win98 command prompt. (I'm going to check a full 98 setup later). So far the indications are that the COMBINATION of W2K and 850 on this chip are mutually exclusive. Also all the H/W has worked fine in the past w/ Win2K at 100FSB (using a 366 OC'd to 550). Can you offer some suggestions on what is relevant to try? DOS / 98 command prompt is small and simple compared to win98, and thus far more likely to boot than full Win when you have data errors going on. This is to be expected. Can you expand on "data errors" that might prevent W2K from booting and hang? have now done. Hopefully clearly. I dont know I could always be wrong, but it seems quite obvious that youre trying to run it at a speed it cant do. What to do? Run it at a speed it will do! What else? That's not exactly correct - I can reflash the BIOS each time at boot up and it can CONSISTENTLY run at 850 indefinately and be very stable. I am looking for: A) an explanation as to what is special about running W2K immediately after a flash that allows this B) a solution to booting it at 850mhz into W2K w/o having to reflash it each time I know thats a teaser, but a) its not something you can change AFAIK, unles you try older BIOS software versions on the remote offchance one might work. One certainly can't imagine a 'change' that could be made if one doesn't at least develop a theory as to why it makes a difference. b) the plain stump-like fact is it runs reliably at 566 but not consistently at 850. Except your "plain stump fact" isn't. The "plain stump fact" is it runs perfectly fine at 850, for days on end. But the dern thing doesn't reboot, nor run a second time, for some unknown reason. Whether that's due to the flash, running 100MHz FSB (which is a standard FSB, btw), some oddball BIOS setting, a misconfigured Win2000, bad caps, bad PSU, the overclock, gremlins, bad karma, B.O., or some other as of yet unidentified problem is simply unknown. Hell, for all you know he simply forget to set the dern AGP or PCI divider. Youve tried the tricks, tweaking the voltage... thats it. Your max speed depends not just on CPU or mobo, but on the pair of them together. If the CPU will run faster in another mobo, you might contemplate using the other mobo. Could there not be some driver or configuration issues w/ W2K? I guess there one way to find out: put 98se on it, or any other 'as different to 2k as poss' large OS. But if win2k starts up fine at 566 but not at 850, you have a clear answer: data errors at 850. Not everything that can create problems is a 'data error', unless you mean to include every electrical signal in the system, including power rails, as being some form of 'data'. But I don't think that's the meaning you intend and your 'clear answer' just isn't so. |
Result:
Well it was a step forward.... kinda... After clearing the CMOS via the mobo jumper and reflashing to get my voltage options up I was again in W2K at 850 for the better part of this afternoon. I was reluctant to restart windows because I figured it would fail and I'd be back in at 637mhz. Well wouldn't ya know it, after the CMOS clear, and all that I DID get into W2K again at 850. And again and again via 'restart'. Even ran a torture test for an hour or two while at dinner. But wait! there's a catch. I then powered down the PC and after a few minutes turned it on again. Uh oh... back to square one. It would not get into W2K after a power down after all of that. ------- On the reset switch power front. I did notice the ACPI tab on device driver under the 'computer' tab in the device manager - missed that in my previous post. After a little digging around on the web I realized my mobo (the old ZM6 ABit from late 90s) is not fully ACPI compatible w/ W2K - the telltale sign is it has never fully powered down from W2K like it used to under W98SE. THere is an APM tab in the power options device manager entry that indicates W2K did detect APM when it was installed as well as (limited?) ACPI. I switched the APM driver on and enabled in W2K as well as the BIOS and it now powers down the system from W2K rather than physically pretting the power button. No biggee - just a minor discovery and improvement. But the reset button continues to do little more than blank the screen and cause the drives to click a few times if it follows a previously unsuccessful attempt to load W2K at 850. The reset button seems to be working under other conditions like slower speeds. Anyway I 'think' I can amend the previous condition of one time after flash to many restarts after flash until a power-off... Well it's progress of sorts... -------- I'm going to start off my 98SE setup at last. I'm thinking on that Vid to Vss jumper mentioned earlier. May try safe mode and bootlogging again though neither worked in the past. thanks! On Wed, 07 Jul 2004 17:52:58 -0500, David Maynard wrote: pgtr wrote: On Wed, 07 Jul 2004 02:17:42 -0500, David Maynard wrote: SNIP It hangs. It usually hangs in the black 'F8' screen w/ the progress bar at the bottom that says it is starting windows. The progress bar at the bottom usually goes about 75-80% across when it locks up. The case reset switch has no affect at this point either - I have to power it off. It will not log a bootlog file either. The case reset switch not working is particularly disturbing. Does it work on the 'first' boot? I mean, instead of trying a reboot can you kill it by hitting reset? Let me expand on that. The 'reset' switch on the case normally works fine and of course is only needed in a dire situation (which is super rare or almost never for me under W2K). If I OC to 850mhz and do the flash trick I can get into W2K. At that point I've never tried the reset swich because there was no need. The PC functioned normally and I could then do a normal restart or shut down thru windows. That is what I presumed was the case and is why I asked about trying the reset switch when the system is in the 'working' state. The NEXT time into W2K it will of course hang usually on the black F8 windows starting phase... Now HERE if I hit the reset button it will blank the screen, you'll here some ticks coming from the HD as you normally would but then it just sits there w/ a black screen indefinately. I simply have to turn the power off and then back on to get it to respond again. WELLLlllllll, now, that's a sight different than the original description of "reset switch has no affect (sic)." It IS apparently doing 'something'... it blanks the screen and the hard drive response indicates it got a hard reset. The motherboard apparently isn't 'starting back up', from the reset, though. I CAN however put a boot disk in and boot (at 850mhz) into a DOS prompt (W98 book disk) and the reset button works there as well as CNTRL ALT DEL. It's beginning to sound like a BIOS power management issue. APM, ACPI, nothing, conflicting, confused, I don't know. Of course, DOS couldn't care less about APM/ACPI but Windows 2000 sure as heck does and it's going to be very angry about having the wrong kernel (APM, ACPI, etc) vs what the motherboard claims. Basically when it hangs on teh 2nd attempt and beyond going into W2K at 850mhz - it REALLY hangs and not even the reset button will wake it up properly - just a power off/on. Yes, ok. The rest doesn't cause a proper reboot but, according to your current description, it IS doing something. Yeah. Ok. If you'd mentioned Abit and flashing for Vcore I'd have known what you meant. Yep. I thought I'd mentioned in the orig post it was an ABit ZM6 (similar to BM6 but less memory). You know your mobos! I have the latest flashing they released (SU?) to move the default voltage up... I think the chip is pretty happy at 1.85V which is right in line w/ the many entries in overclockers.com CPU database for OCing the 566 to 850. Yeah, I don't think it's the voltage, per see, or else the symptom would be more random. How does the 566 perform at 850 if you use the max voltage the 'normal' default range allows? I've never heard of a problem quite like this but it's as if the motherboard doesn't like the foolie flash. Which Abit motherboard is it? I presume it's one of their socket 370 boards because if you're using a slotket you could adjust Vcore there and not bother with the flash. Well, unless you're using one of the el-cheapo slotkets with no Vcore jumpers. Yes again - you do know those old mobos! The ZM6 w/ the last couple of flashings (QU? and SU?) will support the earliest coppermine CPUs including explicitly the cB0 stepping up to 600mhz - this chip is a 566 cB0 stepping. No slotket or adapter. Set your memory timings to the slowest possible and see if that helps. Long since set to slowest most conservative settings. Everythign in BIOS is set conservatively. Remember this all worked fine w/ a 100FSB setting w/ a 366 OCd to 550. The 100FSB itself in thoery shouldn't be a problem here I would think...? Well, it could but, in this case, I doubt that's it since it works 'forever' until you try to reboot. Indeed. Forever: Several days at a time including extended overnight sessions w/ prime95 torture test. No apparent temp problems or stability issues. It just runs fantastic at 850mhz that ONE time after a flash under W2K. Its the 2nd time and beyond that it hangs... As I said never quite heard of something like this in my limited OCing experience. Strange! I got lucky in stumbling into the fact that it would work the one time after a reflash - if it wasn't for that I might very well have concluded it was not do-able. Just because it worked with a 366 OC'd, though, doesn't mean a 566 'must'. On the other hand, that overclock isn't in any way unusual. No it doesn't. But 366 at 550 tells me that the system is stable and happy w/ a 100FSB. And the 566 CAN run nicely at 850 indefinately that 'one' time after a flash. It can also run at 850 consistently via a DOS prompt boot. Sure seems like the ducks are all in a row for it to OC at 850mhz but throw in W2K and ...? Which kernel DO you have in win2000? ACPI? 'Standard pc'? The only significant difference I can think of, off hand, is the 366 being a ppga but the 566 a COPPERMINE. What stepping is it? The B stepping is socket compatible with ppga but I think there were some differences with the C. Yep - see above - it's a cB0 which is supported by ABit ZM6 w/ the last couple of BIOS releases. Yes. Well, CB0 should be plug in compatible with ppga anyway. I don't have an answer but some thoughts. It seems to be either BIOS or coppermine related and not the overclock, per see, since it runs fine the one time. The reboot failure sounds similar to the reset switch not working since the last thing a reboot does is essentially a reset. Which wouldn't seem to explain not working a second time from a cold start EXCEPT a cold start 'starts' with reset (By reset I mean to include setting up the processor registers). So then the question would be, why does it work the first time at all? And, with that, it starts at 66, so it had a reset, you flash, reboot, set Vcore/FSB, and then run 100. Have you tried powering it off RIGHT after you set 100MHz FSB, but no boot into Windows, to see if it will THEN run the 'one time' from a cold start at 100MHz FSB? No I haven't tried that particular combination. What might that tells us one way or the other? To be honest going thru the flashing process is a little tricky and sometimes it doesn't take or preserve the voltages correctly and may take a 2nd or 3rd attempt (I've had problems w/ it reverting from 1.7V to 1.5V default on subsequent flashes so I have to do it twice). Even though I have a UPS we have lots of power fluctuations out here and to be honest I'm just plain nervous about doing any more flashes. ;) Well, I was trying to establish whether the reset ever worked 'right' with win2000 but I'm now on the power management tract. What *I* would do, at this stage, since the Vcore range is your big issue, is wire strap the processor pins for a higher Vcore so the BIOS just naturally thinks it should be higher: no flash required. Pulling VID3, that's AJ37, to ground (AK36, Vss, is right next to it) will give you 1.9V default (just about your only choice short of insulating/pulling pins), which you can then lower to the 1.85 you're using. That is, if the default core is 1.5, as I think you said. If it's higher then jumpering AJ37 low will result in a correspondingly higher Vcore above 1.9. Doesn't seem to explain why it would run in DOS, though, unless whatever is 'missing' from the reset is of no consequence to DOS. I'm leaning towards something odd going on w/ W2K that seems to upset the applecart so to speak w/ regards to: BIOS or COppermine or something...? I believe the chip posts consistently at 850. It doesn't have any temperature related stability problems at 850. W2K is happy w/ the chip at 66FSB and 75FSB. There is a setting as I recall in the BIOS that is something like 'Force Update ESCD' - could that have any impact? I'm going to hit the jumper on the mobo and clear the CMOS (may need to reflash to get back to 1.7V default?) and see what that does. Clearing BIOS is a grand idea with all the flashes you've done. I'm also going to grab an extra small HD and put a Win98 install on there and swap out the two large HDs w/ W2K for this one temporarily - I'd like to see if it will boot consistently into a full Win98 at 850... I don't know how much it says that it can boot into a W98 dos prompt at 850 consistently - it is only a DOS prompt as opposed to the full W98. Basically at this point I'm sure the chip will post at 850 consistently. Also that it doesn't seem to have any temperature related stability problems (based on 'successfuly' initial boots after flash). Try the VID jumper so you don't have to flash. |
pgtr wrote: On 7 Jul 2004 04:33:41 -0700, (N. Thornton) wrote: I'm no overclocking genius but if it all works dandy at one speed and frequently hangs at 30% faster speed... hello? It hangs specifically at startup under one specific OS in a consistent fashion. THis largely eliminates the typical OC issues of not posting and temperature stability problems and leaves some sort of configuration issue. PS: Also the case reset button no longer functions w/ it set at 100FSB(850mhz) - usually I have to power cycle now too...? again, says it all. Seems like you been focussing too much on whats not relevant. Something needs slowing down. I've no idea what your speed options are on this combo but hope you can do something better than 566, if not 850. I've got the thing as conservatively set up as possible. Is a 566 running at 850 conservative? To be more specific I was referring to all BIOS settings - they are as conservative as possible. However a 566 at 850 is below average. The CPU database suggests 874 as average. Anything above that up to say 1ghz is aggressive. The 566 is generally regarded as one of the more OCable chips of that ilk much like the 300A back in it's day. The only real variable here that I'm aware of is the FSB - can you suggest what I appear to be missing? the obvious? ok The chip is locked at 8.5 multiplier. It runs fine at 66FSB (it's stock speed). It will run fine at 75FSB (a mild OC of 637mhz). It will run fine at 100FSB one time immediately after a BIOS flash. It will also run fine at 850mhz(100FSB) repeatedly if using a boot disk into DOS or Win98 command prompt. (I'm going to check a full 98 setup later). So far the indications are that the COMBINATION of W2K and 850 on this chip are mutually exclusive. Also all the H/W has worked fine in the past w/ Win2K at 100FSB (using a 366 OC'd to 550). Can you offer some suggestions on what is relevant to try? DOS / 98 command prompt is small and simple compared to win98, and thus far more likely to boot than full Win when you have data errors going on. This is to be expected. Can you expand on "data errors" that might prevent W2K from booting and hang? I dont know I could always be wrong, but it seems quite obvious that youre trying to run it at a speed it cant do. What to do? Run it at a speed it will do! What else? That's not exactly correct - I can reflash the BIOS each time at boot up and it can CONSISTENTLY run at 850 indefinately and be very stable. I am looking for: A) an explanation as to what is special about running W2K immediately after a flash that allows this That's easy - immediately after you flash, the BIOS is in a state which allows W2K to boot at 850 on your specific hardware. This state is lost after you boot W2K, and you have yet to happen on a way to restore it - short of reflashing. Discovering exactly *how* the BIOS state changes and *why* that prevents rebooting is a much more difficult problem, one that is very probably insoluble without sophisticated diagnostic equipment. B) a solution to booting it at 850mhz into W2K w/o having to reflash it each time Finding such a solution, if it exists, is likely a matter of trial and error. I may have missed it, but on reviewing the thread I don't see much evidence of attempts to find a less drastic way to restore the BIOS to it's post-flash state. Have you tried simply entering and exiting the BIOS without saving? Entering the BIOS, changing a trivial parameter, exiting with or without saving? Resetting BIOS to defaults? etc. etc. If there's more than one BIOS revision available which supports your configuration, try these tricks with all suitable versions. Experiments of this nature have yielded positive results for me on several occasions, most notably the ability to run 1GB RAM at 140Mhz FSB on a P2B-DS. This is impossible unless you enter/exit the BIOS on every boot - for reasons unknown and unlikely to be explained, but hey, it works :-) It occurs to me the root of this problem might relate to CPU microcode, given that W2K will apply a CPU microcode update early in the boot process if the BIOS has not already done so, whereas DOS and W98 do not have this functionality. It would be be interesting to know if your BIOS contains the microcode update for the problematic CPU, and what effect removing it (if present) or adding it (if not) has on your symptoms. Finding out if W2K will install at 850, and if so, whether the fresh install will reboot at 850 might also prove to be a worthwhile exercise - or not. Since your issue does not seem to have been encountered and solved previously, I think you're down to trial and error. The trial suggestions above are based on my experience, and with any luck might yield positive results - or at least more information. HTH P2B |
pgtr wrote:
On Wed, 07 Jul 2004 17:52:58 -0500, David Maynard wrote: SNIP WELLLlllllll, now, that's a sight different than the original description of "reset switch has no affect (sic)." It IS apparently doing 'something'... it blanks the screen and the hard drive response indicates it got a hard reset. The motherboard apparently isn't 'starting back up', from the reset, though. Yes BUT remember only if it was at 850. If it's at 637 or 566 it stops and starts fine inclduing the reset button on the case. I am aware of that. And getting to a running 850 always involves a flash procedure. I CAN however put a boot disk in and boot (at 850mhz) into a DOS prompt (W98 book disk) and the reset button works there as well as CNTRL ALT DEL. It's beginning to sound like a BIOS power management issue. APM, ACPI, nothing, conflicting, confused, I don't know. Of course, DOS couldn't care less about APM/ACPI but Windows 2000 sure as heck does and it's going to be very angry about having the wrong kernel (APM, ACPI, etc) vs what the motherboard claims. Hmmmmmm I might double check and see what BIOS is set at. I've done so many flashes that I've stopped bothered to modify much of anything other than the CPU and maybe the setting for PnP OS. Everything else iincluding those power settings in BIOS are all left at whatever the default is. Basically when it hangs on teh 2nd attempt and beyond going into W2K at 850mhz - it REALLY hangs and not even the reset button will wake it up properly - just a power off/on. Yes, ok. The rest doesn't cause a proper reboot but, according to your current description, it IS doing something. Well it 'tries' to restart and acts like it is, but never gets beyond a blank screen so I power off. I know. But doing 'something' vs 'nothing' is potentially useful debug information. Again, only at 850 though not at 566 or 637. Yeah. Ok. If you'd mentioned Abit and flashing for Vcore I'd have known what you meant. Yep. I thought I'd mentioned in the orig post it was an ABit ZM6 (similar to BM6 but less memory). You know your mobos! I have the latest flashing they released (SU?) to move the default voltage up... I think the chip is pretty happy at 1.85V which is right in line w/ the many entries in overclockers.com CPU database for OCing the 566 to 850. Yeah, I don't think it's the voltage, per see, or else the symptom would be more random. How does the 566 perform at 850 if you use the max voltage the 'normal' default range allows? The max teh default allows is 1.5 +-.2 or 1.7V. That just wasn't enough as I recall. It's real happy at about 1.8 to 1.9. I don't recall how it did at 1.75V offhand but I think it might also work there... Well, the POINT is, if it runs AT ALL, try it and see if the dern thing BOOTS past that problem point. SNIP Just because it worked with a 366 OC'd, though, doesn't mean a 566 'must'. On the other hand, that overclock isn't in any way unusual. No it doesn't. But 366 at 550 tells me that the system is stable and happy w/ a 100FSB. And the 566 CAN run nicely at 850 indefinately that 'one' time after a flash. It can also run at 850 consistently via a DOS prompt boot. Sure seems like the ducks are all in a row for it to OC at 850mhz but throw in W2K and ...? Which kernel DO you have in win2000? ACPI? 'Standard pc'? YOu would ask. How do I tell? In Control Panel: Power Options it shows an APM tab. In that tab it is NOT checked. Under Device Mgr if I view hidden devices the NT/APM Legacy driver shows itself but w/ a red X - it's disabled. Let me know what to change or check for and I'll definately go for it. The HAL.DLL will be in system32 but you need to check properties to see what it's original name was. My HAL, for example, says it was originally halaacpi.dll (meaning ACPI Uniprocessor PC). http://www.dewassoc.com/support/win2000/tshoot_hal.htm (note they say to rename it, which is why you need to check properties as Windows install renames it too) No I haven't tried that particular combination. What might that tells us one way or the other? To be honest going thru the flashing process is a little tricky and sometimes it doesn't take or preserve the voltages correctly and may take a 2nd or 3rd attempt (I've had problems w/ it reverting from 1.7V to 1.5V default on subsequent flashes so I have to do it twice). Even though I have a UPS we have lots of power fluctuations out here and to be honest I'm just plain nervous about doing any more flashes. ;) Well, I was trying to establish whether the reset ever worked 'right' with win2000 but I'm now on the power management tract. FWIW the reset does work fine @ 566 and 637 (75fsb) and did w/ the old 366 @ 550(100fsb) as well. I presumed that. The question was whether it worked right during the 'first run', and before a reboot attempt, at 850, but that was when I was going on your description of it doing 'nothing'. Now that I know it DOES 'do something' it's become a moot question. 'Reset' isn't 'dead'; it's the 'startup' (initialization) that's not getting done for some reason. What *I* would do, at this stage, since the Vcore range is your big issue, is wire strap the processor pins for a higher Vcore so the BIOS just naturally thinks it should be higher: no flash required. Pulling VID3, that's AJ37, to ground (AK36, Vss, is right next to it) will give you 1.9V default (just about your only choice short of insulating/pulling pins), which you can then lower to the 1.85 you're using. That is, if the default core is 1.5, as I think you said. If it's higher then jumpering AJ37 low will result in a correspondingly higher Vcore above 1.9. So you're saying soldering a wire between AJ37 and AK36? Would this best be done from under the mobo? If you're going to solder, then yes, on the mobo. I use the 'pin hole in the wire' trick to jumper them on the processor itself: Stripping a single, insulated, strand off a ribbon cable, punching two appropriately spaced holes (which means a semi long piece so each end is maneuverable and it ends up like a loop from one pin to the other) through the insulation (and the center of the wire bundle) with a safety pin (sprung open) and then pushing the wire onto the processor pins. The slight 'tension' to the stretched insulation (since the pin is in there) will keep the wire against the processor pin, if you get the hole in the middle of the wire bundle. Takes me forever, and multiple choice words, to get one 'right' but, once done, it works. Here's a link to an interesting article that talks about modifying the same ZM6 mobo for later coppermines - it's where I picked up the flash trick. http://www.3feetunder.com/krick/370mod/ I need to find a better diagram of the celeron as that one is too small/blurry - probably at intel.com I can find a better one. I've seen the various mod types. I've got two BH6s, an Asus P2B-VM, well, the slotkets, and a Chaintech Via mobo modified for tualatins. With the slotkets I use them to set the voltage. It's the socket 370 Chaintech I use the pin through wire trick on. Did it for a P-III in a ppga socket 370 board too. Any other easy/quickee tricks to jump AJ37 and AK36? Conductive paint pen. Don't try soldering them because, unless you are very good, you'll loose a pin. I don't mind putting something around a pint to insulate it (like a piece of insulation from an old ribbon cable). Bit more of a hassle to pull everything apart and solder from behind - small for me to solder but I think doable for me. I never could get the 'insulation' trick to work and you don't really need to as 1.9v is not so far off as to be a serious problem. Doesn't seem to explain why it would run in DOS, though, unless whatever is 'missing' from the reset is of no consequence to DOS. I'm leaning towards something odd going on w/ W2K that seems to upset the applecart so to speak w/ regards to: BIOS or COppermine or something...? I believe the chip posts consistently at 850. It doesn't have any temperature related stability problems at 850. W2K is happy w/ the chip at 66FSB and 75FSB. There is a setting as I recall in the BIOS that is something like 'Force Update ESCD' - could that have any impact? I'm going to hit the jumper on the mobo and clear the CMOS (may need to reflash to get back to 1.7V default?) and see what that does. Clearing BIOS is a grand idea with all the flashes you've done. Just did it today at lunch and sure enough it kicked me back to 1.5V. I figured it would. SO I figured I'd reflash w/ /cc and have been running 850mhz all afternoon (including a torture test for kicks). I haven't had teh heart to restart it figuring I'll probably lose the 850 yet again... Hehe. Probably. I suppose you could get a UPS and leave it permanently on. Then hope you never have an update that requires a reboot. (good luck) I'm also going to grab an extra small HD and put a Win98 install on there and swap out the two large HDs w/ W2K for this one temporarily - I'd like to see if it will boot consistently into a full Win98 at 850... I don't know how much it says that it can boot into a W98 dos prompt at 850 consistently - it is only a DOS prompt as opposed to the full W98. Basically at this point I'm sure the chip will post at 850 consistently. Also that it doesn't seem to have any temperature related stability problems (based on 'successfuly' initial boots after flash). Try the VID jumper so you don't have to flash. Yeah I understand what you are thinking but I need to think that one thru. FYI the 'default' chips only cycle up to 600(66). The 'default chips' shouldn't matter as you're setting the FSB manually anyway. It's the foolie part of the flash I'm wondering about. I also still want to load up W98 on a spare HD and see what happens. If it works that would isolate the problem as unique to W2K combined w/ the OC'd chip. Yep. A reasonable test. And I will need to (reluctantly) reboot here shortly and verify it will NOT come up @ 850 a second time in W2K. /play violins |
pgtr wrote:
Result: Well it was a step forward.... kinda... After clearing the CMOS via the mobo jumper and reflashing to get my voltage options up I was again in W2K at 850 for the better part of this afternoon. I was reluctant to restart windows because I figured it would fail and I'd be back in at 637mhz. Well wouldn't ya know it, after the CMOS clear, and all that I DID get into W2K again at 850. And again and again via 'restart'. Even ran a torture test for an hour or two while at dinner. But wait! there's a catch. I then powered down the PC and after a few minutes turned it on again. Uh oh... back to square one. It would not get into W2K after a power down after all of that. ------- On the reset switch power front. I did notice the ACPI tab on device driver under the 'computer' tab in the device manager - missed that in my previous post. After a little digging around on the web I realized my mobo (the old ZM6 ABit from late 90s) is not fully ACPI compatible w/ W2K - the telltale sign is it has never fully powered down from W2K That's a pretty dern useful clue. It's ALWAYS had a power management issue. like it used to under W98SE. THere is an APM tab in the power options device manager entry that indicates W2K did detect APM when it was installed as well as (limited?) ACPI. I switched the APM driver on and enabled in W2K as well as the BIOS and it now powers down the system from W2K rather than physically pretting the power button. No biggee - Not hardly. You've got a problem if it's running an ACPI HAL and APM both as that's a one or the other proposition, not together. Since you say ACPI is problematic, I'd suggest you reinstall, selecting Standard PC (F8 during setup), and use APM. You may find, I hope, that the problems go away once ACPI is out of the picture. It'll also clear up any potential hardware config issue. Note, from the article: "Note that you should not attempt to change from an ACPI HAL to a standard HAL or from a standard HAL to a ACPI HAL under any circumstances. Doing so will result in your computer not starting properly or at all. This occurs because the Plug and Play device tree that would be currently loaded is for ACPI, and it does not get reconstructed or revert to a standard HAL Plug and Play device tree." Now, they're talking about not using their 'manual' procedure to change ACPI/Standard HAL; a reinstall is fine, and how one does it. But the reason I point it out is you seem to have some kind of 'hybrid' installation with APM/ACPI mixed together and lord knows what kind of mixed mode ACPI/PnP device tree it's got. just a minor discovery and improvement. But the reset button continues to do little more than blank the screen and cause the drives to click a few times if it follows a previously unsuccessful attempt to load W2K at 850. The reset button seems to be working under other conditions like slower speeds. Anyway I 'think' I can amend the previous condition of one time after flash to many restarts after flash until a power-off... Well it's progress of sorts... I think that is a HELL of a lot of progress and it tends to support my theory of power management, specifically ACPI, being the problem. -------- I'm going to start off my 98SE setup at last. I'm thinking on that Vid to Vss jumper mentioned earlier. Good idea regardless of whether removing ACPI 'fixes' it because then you don't have to worry about some anomoly resetting your Vcore back to 1.5. May try safe mode and bootlogging again though neither worked in the past. thanks! |
On Thu, 08 Jul 2004 01:55:22 -0500, David Maynard
wrote: On the reset switch power front. I did notice the ACPI tab on device driver under the 'computer' tab in the device manager - missed that in my previous post. After a little digging around on the web I realized my mobo (the old ZM6 ABit from late 90s) is not fully ACPI compatible w/ W2K - the telltale sign is it has never fully powered down from W2K That's a pretty dern useful clue. It's ALWAYS had a power management issue. Yes. It was just one of those things that I put off and eventually forgot about when I upgraded to W2K. But the reset button has always worked... like it used to under W98SE. THere is an APM tab in the power options device manager entry that indicates W2K did detect APM when it was installed as well as (limited?) ACPI. I switched the APM driver on and enabled in W2K as well as the BIOS and it now powers down the system from W2K rather than physically pretting the power button. No biggee - Not hardly. You've got a problem if it's running an ACPI HAL and APM both as that's a one or the other proposition, not together. Since you say ACPI is problematic, I'd suggest you reinstall, selecting Standard PC (F8 during setup), and use APM. You may find, I hope, that the problems go away once ACPI is out of the picture. It'll also clear up any potential hardware config issue. Note, from the article: "Note that you should not attempt to change from an ACPI HAL to a standard HAL or from a standard HAL to a ACPI HAL under any circumstances. Doing so will result in your computer not starting properly or at all. This occurs because the Plug and Play device tree that would be currently loaded is for ACPI, and it does not get reconstructed or revert to a standard HAL Plug and Play device tree." Now, they're talking about not using their 'manual' procedure to change ACPI/Standard HAL; a reinstall is fine, and how one does it. But the reason I point it out is you seem to have some kind of 'hybrid' installation with APM/ACPI mixed together and lord knows what kind of mixed mode ACPI/PnP device tree it's got. The power situation causing a non-bootable system is a point well taken. I vaguely recall years ago when I upgraded that I immediately ran into a situation that required the 4 recovery diskettes due to some sort of power config I entered. According to this article: http://is-it-true.org/nt/nt2000/atips/atips42.shtml My system falls into a neutral category (hybrid?). It was not on the APM disable list so the drivers were installed/disabled apparently. It's a simple option under the Control Panel power options to enable legacy APM support and voila - my shotdown correctly powers off! However... It didn't seem happy w/ W98 at 850 either (more on that later). And following the steps in the above article it seems to power down just fine and it's always worked just fine at various non-850 speeds all these years as is otherwise. Based on the 98 scenario I'm not hopeful switch from ACPI to Standard PC is going to make the difference. However you make a compelling argument and I've been thinking about doing the W2K re-install anyway so I'm going to pencil that into the todo list in the next day or so maybe even today. just a minor discovery and improvement. But the reset button continues to do little more than blank the screen and cause the drives to click a few times if it follows a previously unsuccessful attempt to load W2K at 850. The reset button seems to be working under other conditions like slower speeds. Anyway I 'think' I can amend the previous condition of one time after flash to many restarts after flash until a power-off... Well it's progress of sorts... I think that is a HELL of a lot of progress and it tends to support my theory of power management, specifically ACPI, being the problem. You've convinced me it's worth a shot to do the reinstall. thanks again... -------- I'm going to start off my 98SE setup at last. I'm thinking on that Vid to Vss jumper mentioned earlier. Good idea regardless of whether removing ACPI 'fixes' it because then you don't have to worry about some anomoly resetting your Vcore back to 1.5. My hobby of tinkering on old cars for improved performance taught me to do one thing at a time! ;o) I like this trick if it allows me to avoid the foolie flash as it's called. May try safe mode and bootlogging again though neither worked in the past. thanks! |
On Thu, 08 Jul 2004 00:22:34 -0400, P2B wrote:
SNIP I am looking for: A) an explanation as to what is special about running W2K immediately after a flash that allows this That's easy - immediately after you flash, the BIOS is in a state which allows W2K to boot at 850 on your specific hardware. This state is lost after you boot W2K, and you have yet to happen on a way to restore it - short of reflashing. Yes but tell me more about the details of that 'state' ... ;o) Discovering exactly *how* the BIOS state changes and *why* that prevents rebooting is a much more difficult problem, one that is very probably insoluble without sophisticated diagnostic equipment. Indeed. B) a solution to booting it at 850mhz into W2K w/o having to reflash it each time Finding such a solution, if it exists, is likely a matter of trial and error. I may have missed it, but on reviewing the thread I don't see much evidence of attempts to find a less drastic way to restore the BIOS to it's post-flash state. Clearing the CMOS via the mobo jumper DID allow restarts into W2K at 850. HOwever if the system is powered off and back on - it's back to square on. That indicates 'some' progress... Have you tried simply entering and exiting the BIOS without saving? Yes. Entering the BIOS, changing a trivial parameter, exiting with or without saving? Yes. Resetting BIOS to defaults? Yes. etc. etc. If there's more than one BIOS revision available which supports your configuration, try these tricks with all suitable versions. THere is one prior BIOS avail that also supports the early Coppermine cB0 stepping - I believe I tried it once. No joy. But I can't say that I've duplicated every step of the last week or so on that slightly older BIOS. Experiments of this nature have yielded positive results for me on several Well experimentation over the last week to 10 days allowed me to 'discover' I could get it up at 850 under W2K immediately after a flash and based on suggestions here that I could get it up at 850 repeatedly after a CMOS clear as long as I didnt' power it off. occasions, most notably the ability to run 1GB RAM at 140Mhz FSB on a P2B-DS. This is impossible unless you enter/exit the BIOS on every boot - for reasons unknown and unlikely to be explained, but hey, it works :-) It occurs to me the root of this problem might relate to CPU microcode, given that W2K will apply a CPU microcode update early in the boot process if the BIOS has not already done so, whereas DOS and W98 do not have this functionality. It would be be interesting to know if your BIOS contains the microcode update for the problematic CPU, and what effect removing it (if present) or adding it (if not) has on your symptoms. Finding out if W2K will install at 850, and if so, whether the fresh install will reboot at 850 might also prove to be a worthwhile exercise - or not. I did a quickee 98 install last night and it was disappointing at 850. I didn't spend a whole lot of time debugging it but superficially it suggested that the 850 permutation was a probably cause for 98 not being happy as well. Otherwise I'd find that mention of microcode VERY interesting. Since your issue does not seem to have been encountered and solved previously, I think you're down to trial and error. The trial suggestions above are based on my experience, and with any luck might yield positive results - or at least more information. HTH Indeed it does - much appreciated! I'm looking at a W2K reinstall anyway and I can repeat more config scenarios w/ the previos BIOS flash. But the W98 situation concerns me and suggests it may be more of a H/W issue regardless of the OS... I couldn't generate a bootlog.txt in 98 either for some reason... |
pgtr wrote in message . ..
On 7 Jul 2004 17:51:40 -0700, (N. Thornton) wrote: So the picture leaves us with data errors, not configuration problems. It does? I've logged at least 72 hours now under W2K on this particular 566 at 850mhz and it has not locked up or acted in an instable fashion once. Obviously I have to perform the flash trick to get it up at that particular speed but once up it's fine. There's no reason to believe the system couldn't run indefinately and in a stable fashion. Conservative or aggressive - I've logged at least 72 hours now under W2K on this particular 566 at 850mhz and it has not locked up or acted in an instable fashion once. Obviously I have to perform the flash trick to get it up but once up it's fine. In fact I'm posting this message while running at 850. That's not exactly correct - I can reflash the BIOS each time at boot up and it can CONSISTENTLY run at 850 indefinately and be very stable. Again it DOES run consistently at 850mhz for hours or even days even under torture test conditions if I myself am consistent in reflashing it prior to booting. I often leave my system up for weeks, sometimes a month or better. Given the long stints the system is left up for - I could almost justify the pre-boot flash trick as standard procedure. Let me ask you one key question: does it behave itself properly at 566, rebooting happily, but wont reboot at 850? If the answers yes, data errors is the only at all likely explanation. If the answers no, then I either misread or mis-somethinged. So why will it run happily at 850 but not boot at 850 after the first go? guess subsequent boots involve the use of an instruction that the CPU/mobo cant do at the higher speed. This instruction isnt used in runtime. /guess The one question there is all important. I assume you understand exactly how things go wrong when a machine is o/ced past what it can manage stably and reilably? Regards, NT |
|
On Tue, 06 Jul 2004 23:48:40 -0500, pgtr wrote:
Can you expand on how to check for bad caps? check for bad caps ... if they are bulged or leaking like this: http://freeweb.siol.net/jerman55/HP/...ds/badCaps.jpg and check also AGP/PCI divider in bios (if settable, check if also any jumpers on the MoBo for that) try also disabling Acpi in bios & set let bios manage your IRQ table instead of OS & than at default speed CPU reinstall the OS ... -- Regards, SPAJKY ® & visit my site @ http://www.spajky.vze.com "Tualatin OC-ed / BX-Slot1 / inaudible setup!" E-mail AntiSpam: remove ## |
pgtr wrote:
On Thu, 08 Jul 2004 01:55:22 -0500, David Maynard wrote: On the reset switch power front. I did notice the ACPI tab on device driver under the 'computer' tab in the device manager - missed that in my previous post. After a little digging around on the web I realized my mobo (the old ZM6 ABit from late 90s) is not fully ACPI compatible w/ W2K - the telltale sign is it has never fully powered down from W2K That's a pretty dern useful clue. It's ALWAYS had a power management issue. Yes. It was just one of those things that I put off and eventually forgot about when I upgraded to W2K. But the reset button has always worked... like it used to under W98SE. THere is an APM tab in the power options device manager entry that indicates W2K did detect APM when it was installed as well as (limited?) ACPI. I switched the APM driver on and enabled in W2K as well as the BIOS and it now powers down the system from W2K rather than physically pretting the power button. No biggee - Not hardly. You've got a problem if it's running an ACPI HAL and APM both as that's a one or the other proposition, not together. Since you say ACPI is problematic, I'd suggest you reinstall, selecting Standard PC (F8 during setup), and use APM. You may find, I hope, that the problems go away once ACPI is out of the picture. It'll also clear up any potential hardware config issue. Note, from the article: "Note that you should not attempt to change from an ACPI HAL to a standard HAL or from a standard HAL to a ACPI HAL under any circumstances. Doing so will result in your computer not starting properly or at all. This occurs because the Plug and Play device tree that would be currently loaded is for ACPI, and it does not get reconstructed or revert to a standard HAL Plug and Play device tree." Now, they're talking about not using their 'manual' procedure to change ACPI/Standard HAL; a reinstall is fine, and how one does it. But the reason I point it out is you seem to have some kind of 'hybrid' installation with APM/ACPI mixed together and lord knows what kind of mixed mode ACPI/PnP device tree it's got. The power situation causing a non-bootable system is a point well taken. I vaguely recall years ago when I upgraded that I immediately ran into a situation that required the 4 recovery diskettes due to some sort of power config I entered. According to this article: http://is-it-true.org/nt/nt2000/atips/atips42.shtml My system falls into a neutral category (hybrid?). Well, not quite. the 'neutral category' they talk about is the motherboard being not ACPI complaint NOR APM complaint so neither is active. You don't have the situation of neither, you have the supposedly improbable combination of BOTH installed. It was not on the APM disable list so the drivers were installed/disabled apparently. It's a simple option under the Control Panel power options to enable legacy APM support and voila - my shotdown correctly powers off! Except the option shouldn't BE there as it isn't supposed to even 'look' for APM if it's detected and installed ACPI, as in "If ACPI compatibility is not present, W2K installation will attempt to install APM drivers." The operative word is "If." And if it doesn't pass ACPI it isn't supposed to INSTALL an ACPI HAL. However... It didn't seem happy w/ W98 at 850 either (more on that later). And following the steps in the above article it seems to power down just fine and it's always worked just fine at various non-850 speeds all these years as is otherwise. Why it behaves differently at 100MHz FSB, vs lower speeds, is something I don't have a good answer for, unless it's some quirk in their 'not quite right' ACPI BIOS, as in some obscure internal timing parameter they don't alter properly for the higher bus speed. It's almost as if it's ACPI when under 100 MHz FSB but there and above, "surprise, no workie, no ACPI now." Based on the 98 scenario I'm not hopeful switch from ACPI to Standard PC is going to make the difference. However you make a compelling argument and I've been thinking about doing the W2K re-install anyway so I'm going to pencil that into the todo list in the next day or so maybe even today. Well, it's a compelling 'possibility' but not so air tight that I'd bet the house on it ;) just a minor discovery and improvement. But the reset button continues to do little more than blank the screen and cause the drives to click a few times if it follows a previously unsuccessful attempt to load W2K at 850. The reset button seems to be working under other conditions like slower speeds. Anyway I 'think' I can amend the previous condition of one time after flash to many restarts after flash until a power-off... Well it's progress of sorts... I think that is a HELL of a lot of progress and it tends to support my theory of power management, specifically ACPI, being the problem. You've convinced me it's worth a shot to do the reinstall. thanks again... -------- I'm going to start off my 98SE setup at last. I'm thinking on that Vid to Vss jumper mentioned earlier. Good idea regardless of whether removing ACPI 'fixes' it because then you don't have to worry about some anomoly resetting your Vcore back to 1.5. My hobby of tinkering on old cars for improved performance taught me to do one thing at a time! ;o) Excellent point, and quite right. Although, it can be 'singly tested' at the standard 66MHz FSB where everything is 'apparently normal'. I like this trick if it allows me to avoid the foolie flash as it's called. Heck, I just made up the name "foolie flash" as it seemed to capture the gist of it ;) May try safe mode and bootlogging again though neither worked in the past. thanks! |
Spajky wrote:
On Tue, 06 Jul 2004 23:48:40 -0500, pgtr wrote: Can you expand on how to check for bad caps? check for bad caps ... if they are bulged or leaking like this: http://freeweb.siol.net/jerman55/HP/...ds/badCaps.jpg and check also AGP/PCI divider in bios (if settable, check if also any jumpers on the MoBo for that) try also disabling Acpi in bios & set let bios manage your IRQ table instead of OS & than at default speed CPU reinstall the OS ... Spajky has a good point here. Contrary to what one would intuitively think, Microsoft says the BIOS should be set for NOT PnP O.S. |
N. Thornton wrote:
pgtr wrote in message . .. On 7 Jul 2004 17:51:40 -0700, (N. Thornton) wrote: So the picture leaves us with data errors, not configuration problems. It does? I've logged at least 72 hours now under W2K on this particular 566 at 850mhz and it has not locked up or acted in an instable fashion once. Obviously I have to perform the flash trick to get it up at that particular speed but once up it's fine. There's no reason to believe the system couldn't run indefinately and in a stable fashion. Conservative or aggressive - I've logged at least 72 hours now under W2K on this particular 566 at 850mhz and it has not locked up or acted in an instable fashion once. Obviously I have to perform the flash trick to get it up but once up it's fine. In fact I'm posting this message while running at 850. That's not exactly correct - I can reflash the BIOS each time at boot up and it can CONSISTENTLY run at 850 indefinately and be very stable. Again it DOES run consistently at 850mhz for hours or even days even under torture test conditions if I myself am consistent in reflashing it prior to booting. I often leave my system up for weeks, sometimes a month or better. Given the long stints the system is left up for - I could almost justify the pre-boot flash trick as standard procedure. Let me ask you one key question: does it behave itself properly at 566, rebooting happily, but wont reboot at 850? If the answers yes, data errors is the only at all likely explanation. If the answers no, then I either misread or mis-somethinged. One common scenario: System works fine at 66 MHz FSB but fails with a 100MHz FSB overclock. Reason: User forgot to set PCI to 1/3 and AGP to 1/2. Just saying "data errors" is like saying "car broke." It doesn't aid in finding out what's broken. So why will it run happily at 850 but not boot at 850 after the first go? We've actually gotten past the 'reboot' issue and now it's down to a cold start issue. guess subsequent boots involve the use of an instruction that the CPU/mobo cant do at the higher speed. This instruction isnt used in runtime. /guess Not very likely. The one question there is all important. I assume you understand exactly how things go wrong when a machine is o/ced past what it can manage stably and reilably? Yes, and it passes all those things that are the usual suspects. Regards, NT |
David Maynard wrote: Spajky wrote: On Tue, 06 Jul 2004 23:48:40 -0500, pgtr wrote: Can you expand on how to check for bad caps? check for bad caps ... if they are bulged or leaking like this: http://freeweb.siol.net/jerman55/HP/...ds/badCaps.jpg and check also AGP/PCI divider in bios (if settable, check if also any jumpers on the MoBo for that) try also disabling Acpi in bios & set let bios manage your IRQ table instead of OS & than at default speed CPU reinstall the OS ... Spajky has a good point here. Contrary to what one would intuitively think, Microsoft says the BIOS should be set for NOT PnP O.S. Indeed. I suspect most following this thread have known not to set PnP OS = Yes for so long that asking the OP if he did it never occurred to us :-) |
P2B wrote:
David Maynard wrote: Spajky wrote: On Tue, 06 Jul 2004 23:48:40 -0500, pgtr wrote: Can you expand on how to check for bad caps? check for bad caps ... if they are bulged or leaking like this: http://freeweb.siol.net/jerman55/HP/...ds/badCaps.jpg and check also AGP/PCI divider in bios (if settable, check if also any jumpers on the MoBo for that) try also disabling Acpi in bios & set let bios manage your IRQ table instead of OS & than at default speed CPU reinstall the OS ... Spajky has a good point here. Contrary to what one would intuitively think, Microsoft says the BIOS should be set for NOT PnP O.S. Indeed. I suspect most following this thread have known not to set PnP OS = Yes for so long that asking the OP if he did it never occurred to us :-) hehe. Yeah, which is why I, too, never thought to ask. But, hey, it can happen to even the best, especially if one gets sidetracked onto other issues. |
pgtr wrote in message . ..
On 8 Jul 2004 14:36:32 -0700, (N. Thornton) wrote: Let me ask you one key question: does it behave itself properly at 566, rebooting happily, but wont reboot at 850? No After the CMOS clear yesterday I CAN reboot repeatedly and run at 850. excellant! If the answers yes, data errors is the only at all likely explanation. If the answers no, then I either misread or mis-somethinged. So why will it run happily at 850 but not boot at 850 after the first go? guess subsequent boots involve the use of an instruction that the CPU/mobo cant do at the higher speed. This instruction isnt used in runtime. /guess Subsequent boots are fine. Now we are reduced to the power-up scenario negating the ability to run in windows at 850mhz. thats as clear as mud to me. No point me saying a squeak when I dont know whats going on. Regards, NT The one question there is all important. I assume you understand exactly how things go wrong when a machine is o/ced past what it can manage stably and reilably? Givne the system WILL manage itself quite "stably and reilably" even after reboots (but not power offs) what things go wrong w/ this oc? |
Redid W2K w/ std PC instead of ACPI (see below). No joy. Also disabled IRQ Steering in the APM tab for kicks. No joy either. I also tried it w/ just an AGP card and also just an (older) PCI video card - neither combo did anything. Several days back I did pull the one old ISA card I still have (an old creative sound card). This card always worked fine w/ the 366 at 550(100). But I did notice one difference trying to run at 850 that might be worth mentioning: WITH the card in the one ISA slot on the mobo the hang occurs EARLIER - that is in the same black F8 screen w/ teh text based starting windows progress bar on the bottom. WITHOUT the old ISA card and the various changes over the last couple days it generally gets pretty much right to the end of the progress bar and just barely starts the 'white' graphical W2k splash screen but only barely starts the new grpahical starting windows progress bar here. In any event it never gets far enough to write the ntbtlog.txt file. In safe mode it prints a number of drivers on the screen and the last line it prints is the AGP440.SYS line. In a clean boot the next one is normally AUDSTUB.SYS. Anway I can still: * dabble some w/ the BIOS one release prior to my current one * twiddle w/ the jumper to get 1.9V as suggested earlier in the thread - maybe this weekend * strip the other cards and try a newer/better PCI graphics than the one currently in there. * solder up the 2 pins etc per web articles on the back of the mobo or install a cheap FCPGA adapter to see if that mod (both are functionally equivalent) has any affect - plus it opens up the door to much faster chips that cB0 stepping CuMines. I know I tried that prior BIOS at least once - no joy. If it actually worked w/ another PCI card I'd face the problem of having to replace PCI components - it's just not worth it IMHO. But I am intrigued about the jumper to get 1.9V - that would allow me to stop w/ the foolie flash and who knows how it might respond... Maybe the FCPGA CuMine mod too... But after that I think that's pretty much about as far as I'm willing to ride this one... Thanks to everyone especially David M! I'll follow up after that last mod or two. BTW what is better: 550(100) or 637(75) (add'l comments below) On Thu, 08 Jul 2004 21:08:48 -0500, David Maynard wrote: SNIP The power situation causing a non-bootable system is a point well taken. I vaguely recall years ago when I upgraded that I immediately ran into a situation that required the 4 recovery diskettes due to some sort of power config I entered. According to this article: http://is-it-true.org/nt/nt2000/atips/atips42.shtml My system falls into a neutral category (hybrid?). Well, not quite. the 'neutral category' they talk about is the motherboard being not ACPI complaint NOR APM complaint so neither is active. You don't have the situation of neither, you have the supposedly improbable combination of BOTH installed. It was not on the APM disable list so the drivers were installed/disabled apparently. It's a simple option under the Control Panel power options to enable legacy APM support and voila - my shotdown correctly powers off! Except the option shouldn't BE there as it isn't supposed to even 'look' for APM if it's detected and installed ACPI, as in "If ACPI compatibility is not present, W2K installation will attempt to install APM drivers." The operative word is "If." And if it doesn't pass ACPI it isn't supposed to INSTALL an ACPI HAL. Well to be honest I can't say I've had any ACPI problems before except for it not powering down (which I fixed w/o removing ACPI by enabling APM Legacy support) and only recenlty the reset button and then only at 850. Anyway it's gone now - I have Standard PC (APM) at last after doing the W2K install. That was a hassle because I lost SP4 and came up w/ a no dial-up accounts and couldn't create one because the wizard was 'goofy'. WHile I couldn't set up a new dial up account to access the web I COULD finally get it to connect via LAN so used another PC/Modem and went the the ICS dance and go on so I could get SP4 installed again. After SP4 I magically got all my old accounts back and everything starting coming together as I reapplied hot fixes etc... I'm glad I did that exercise and will keep it in mind for the future but it didn't help w/ the 850 thing. Oh and reset works most times (it does seem to hang on rare occasion but 9 out of 10 it works after an 850 attempt now). However... It didn't seem happy w/ W98 at 850 either (more on that later). And following the steps in the above article it seems to power down just fine and it's always worked just fine at various non-850 speeds all these years as is otherwise. Why it behaves differently at 100MHz FSB, vs lower speeds, is something I don't have a good answer for, unless it's some quirk in their 'not quite right' ACPI BIOS, as in some obscure internal timing parameter they don't alter properly for the higher bus speed. It's almost as if it's ACPI when under 100 MHz FSB but there and above, "surprise, no workie, no ACPI now." What's different about a mendocino at 100fsb (what I had in there before a 366 at 550) vs an early coppermine at 100fsb (the current 566 at 850). Based on the 98 scenario I'm not hopeful switch from ACPI to Standard PC is going to make the difference. However you make a compelling argument and I've been thinking about doing the W2K re-install anyway so I'm going to pencil that into the todo list in the next day or so maybe even today. Well, it's a compelling 'possibility' but not so air tight that I'd bet the house on it ;) Well I still have the house but no 850. Just as well to dump ACPI and go w/ APM I guess anyway. SNIP I like this trick if it allows me to avoid the foolie flash as it's called. Heck, I just made up the name "foolie flash" as it seemed to capture the gist of it ;) I like the mmoniker - you get the honors - foolie flash definately describes the point of the /cc to coax it up to a higher voltage range. Thanks |
On Fri, 09 Jul 2004 02:01:57 +0200, Spajky wrote:
On Tue, 06 Jul 2004 23:48:40 -0500, pgtr wrote: Can you expand on how to check for bad caps? check for bad caps ... if they are bulged or leaking like this: http://freeweb.siol.net/jerman55/HP/...ds/badCaps.jpg I know it's not the ultimate check but absolutely no bulging or leaking can be observed. off topic - last time I saw really bad caps was an old '49 AA5 tube radio I restored - it had wax coated paper caps and the electrolytics were loooong gone - some nice new orangedrops and 1 new tube got it going again - those are a bit bigger/easier to work with though! and check also AGP/PCI divider in bios (if settable, check if also any jumpers on the MoBo for that) Yep they are set in BIOS. In fact the BIOS inludes the default setting in teh FSB setting E.g. when I set it to 100 it says: "100 (1/3)" making that one almost idiot proof even for me. The AGP is separate and at 2/3 if I recall. (the only other option is 1/1) try also disabling Acpi in bios & set let bios manage your IRQ table instead of OS & than at default speed CPU reinstall the OS ... ACPI is disabled in BIOS - and now in W2K as well though changing that in W2K to APM was just a 'tad' more involved! ;o) Thanks! |
On Thu, 08 Jul 2004 21:15:00 -0500, David Maynard
wrote: Spajky has a good point here. Contrary to what one would intuitively think, Microsoft says the BIOS should be set for NOT PnP O.S. Yep - I've double checked and it's set to disabled or no or whatever for PnP OS. It has been set to yes or enabled in teh past so maybe it's possible I had a permutation that might have worked but for that. sanity checks for the obvious are always welcome here! thanks |
pgtr wrote:
On Fri, 09 Jul 2004 02:01:57 +0200, Spajky wrote: On Tue, 06 Jul 2004 23:48:40 -0500, pgtr wrote: Can you expand on how to check for bad caps? check for bad caps ... if they are bulged or leaking like this: http://freeweb.siol.net/jerman55/HP/...ds/badCaps.jpg I know it's not the ultimate check but absolutely no bulging or leaking can be observed. off topic - last time I saw really bad caps was an old '49 AA5 tube radio I restored - it had wax coated paper caps and the electrolytics were loooong gone - some nice new orangedrops and 1 new tube got it going again - those are a bit bigger/easier to work with though! I know what you mean about old caps and leaking electrolyte but the 'bulge' on the modern caps, at least in this application, is all it takes. If they are not FLAT on top, they're bad. If you notice, that 'flat' top is radially serrated. It's a pressure relief and if it's bulged up then the electrolyte boiled but, on the ones I saw, there is no obvious electrolyte 'leakage' visible while still installed. Anyway, I believe your visual was correct and that they're ok but it's worth noting for future reference. and check also AGP/PCI divider in bios (if settable, check if also any jumpers on the MoBo for that) Yep they are set in BIOS. In fact the BIOS inludes the default setting in teh FSB setting E.g. when I set it to 100 it says: "100 (1/3)" making that one almost idiot proof even for me. The AGP is separate and at 2/3 if I recall. (the only other option is 1/1) try also disabling Acpi in bios & set let bios manage your IRQ table instead of OS & than at default speed CPU reinstall the OS ... ACPI is disabled in BIOS - and now in W2K as well though changing that in W2K to APM was just a 'tad' more involved! ;o) Thanks! |
All times are GMT +1. The time now is 12:53 PM. |
Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
HardwareBanter.com