how high cpu voltage do I dare (3500+)

On Mon, 15 Oct 2007 22:08:03 -0400, Paul wrote:

Kris wrote:
On 10/15/2007 7:17:19 PM, "Kris" wrote:
On 10/15/2007 2:03:15 AM, Paul wrote:
Kris wrote:
On 10/14/2007 11:24:25 PM, Paul wrote:
Kris wrote:
On 10/13/2007 11:37:58 PM, "Kris" wrote:
system in sig.

I have been running this cpu at 220 since I got it (a year ago?), but I
never played with the voltages, and I didn't know whether it was the cpu
or memory or what that was limiting it (I couldn't go over 223 or so).

For some reason I started playing with voltages and nvidia system tune
that came with 163.71 (nforce 6.86) and relatively easily determined that
it was the cpu (venice core as I recall) so I upped cpu, and also memory
cpu orig 1.440 now 1.540 mem orig 2.60 now 2.70
pcie orig 1.5 now 1.60

I use speedfan 4.33 - cpu in cpu gets to 56C, cpu on mobo gets to 49C,
idle is 35 and 40 respectively. case abmient is about 32 or so.

cpu was retail and I use the hsf that came with it as it has always run
very cool. I do not let the mobo control the hsf speed. As I recall, I set
the bios to shut the mobo down if cpu on mobo reaches 56C.

cpu 240 2640mhz --------- EDITED
HT (4x) 960
memory 480 (rather surprised this goes!)
pcie still at 100 (2500mhz) so next I will try playing with that and see.

But the question is, how high can I push the voltage on the cpu and be
relatively safe?

Case is Ultra with 120mm in in front, 120mm out in back, 80mm in on side
over cpu sorta, 350watt Ultra (600 on way) with 120mm fan.

tia!!!

I backed the cpu voltage to 1.520 which I see speedfan lists as 1.55 and
dropped cpu speed to 235 to get 3DMARK05 to not hang.

I then played around with the pci-e clock going up from 100 to 120 and got a
huge increase in 3DMARK05 from 3648 to 3666. wow. LOL. I don't understand why
it doesn't have a bigger effect, but..... That had no effect on stability.

For AMD's idea of the max value, try this doc. Note that this document
never got updated, as newer S939 processors came out, but presumably it is
still accurate. I'm not really crazy about AMD's skills at documentation.
I prefer a set of docs for each unique creation (one for 90nm, one for
65nm etc).

http://www.amd.com/us-en/assets/cont...docs/31411.pdf

On page 53, the value for VDD is the Vcore voltage.

"VDD supply voltage relative to VSS ¡V0.3 V to 1.65 V"

So don't go over 1.65V. Officially speaking.
Ah! thanks for that one!

Since the Vcore regulator on some boards, overvolts a bit at idle, I might
choose 1.60V as the highest BIOS setting for Vcore. That leaves 0.05V of
margin, in case the regulator chip is running on the high side.
I notice that though I have the bios set to 1.520, cpu-z says 1.552, so I am
guessing that this is what you are talking about here - off by 0.032 . But
0.05 sounds like a very good approximation!

For other overclocking info, you should try looking at some threads
here. They have a search feature, so you can narrow your search down
a bit.

http://www.xtremesystems.org/forums/forumdisplay.php?fX

For example, the thread on "Opteron 185" will give you some
ideas as to the range of conditions those people were using
with their gear. There are probably a few other threads
there, where S939 processor get mentioned.

http://www.xtremesystems.org/forums/...ead.php?t1814

Paul

Neither one of those links worked for me, but I will definitely poke around
that site.

Thanks!

:-) Compare what you just posted for those two links, to my original post.

Some characters in the links are corrupted (in your posting). "fX" got
changed to "fX". "t1814" became "t1814".

You might want to verify the stability of your overclock. Orthos should
be able to run for hours, without stopping and reporting an error, if
everything is stable. You can get Orthos here. If it stops in 30 seconds
and reports some kind of error, you've got some adjusting to do on
your motherboard settings.

http://sp2004.fre3.com/beta/beta2.htm

Paul



I just ran it for 5 1/2 minutes with no problems and stopped it out of
boredom. the can got up to 59C which does not make me particularly happy
(vcore=1.54, mult=11,fsb=235)

I widened the hole in the lucite on the side of my case and now have a
120mm fan blowing at the cpu, but I don't think it helped much (rel to
80mm)

I discovered that this abit does not allow incremental adjustment of
memory clocking - only DDR400, DDR433, DDR466, etc. I also found that when
set to 433, it would run, but sandra showed it to be clocked at 167 or
something like that instead of the 216 I would have expected.

This is all a product of boredom. I expect that in a few months I'll
spring for something like a Gigabyte GA-P35-DS3L and an E2160, or
somesuch, which can actually be overclocked!.

Tonight I may try lowering the mult and raising the fsb and testing the
mem that way. max HTT is about 330 so 313.
max CPU is about 240 so probably 230 or so.
The mem's spd is 2.5 4 4 8 400ddr.

Probably time to pull the Zalman off the Shuttle an35n and apply it to
this mobo...


Yeah, stuck the Zalman on there - an old CNPS7000A, all copper. I'm sure
putting fresh AS3 in there didn't hurt. Board wouldn't power so had to take
it all apart and put an extra layer of cardboard as the mobo's resisters were
touching the Zalman metal bottom plate. Ran Orthos for 10 min with no probs
and temp barely touched 50, vs 59 with the other hsf.


Can you access these links OK now ?

http://www.xtremesystems.org/forums/...splay.php?f=58
http://www.xtremesystems.org/forums/...d.php?t=131814

Paul

posted via 40tude Dialog:
(I had composed this on NMP:
no - both give me a "No Thread specified. If you followed a valid link,
please notify the administrator" message.

Using firefox. I get the same result if I use opera.

My reader is NewsMan Pro. Maybe it's doing something to the characters?

Okay, I see... I googled it and what my reader displays as fX is actually
f=58 (f equals 58)
and t box 1814 is actually t=131814 (t equals 131814)

Gotta be my reader.


Darned if I can see how...

Okay, 40tude Dialog does it correctly so I'll ask the author. )

so I just set up dialog again and I'll p[ost this to see if it works.

In the meantime, I checked one of those links and immediately came up with
a venice thread
http://www.xtremesystems.org/forums/...ad.php?t=59405

I think I have a low end chip - I only get 2588 with 1.54v and the first
person I looked at has 2750 with 1.40v . That explains a lot. Bed now.
I'll look at the links tomorrow. THANKS!

Kris wrote:
On Mon, 15 Oct 2007 22:08:03 -0400, Paul wrote:

Kris wrote:
On 10/15/2007 7:17:19 PM, "Kris" wrote:
On 10/15/2007 2:03:15 AM, Paul wrote:
Kris wrote:
On 10/14/2007 11:24:25 PM, Paul wrote:
Kris wrote:
On 10/13/2007 11:37:58 PM, "Kris" wrote:
system in sig.

I have been running this cpu at 220 since I got it (a year ago?), but I
never played with the voltages, and I didn't know whether it was the cpu
or memory or what that was limiting it (I couldn't go over 223 or so).

For some reason I started playing with voltages and nvidia system tune
that came with 163.71 (nforce 6.86) and relatively easily determined that
it was the cpu (venice core as I recall) so I upped cpu, and also memory
cpu orig 1.440 now 1.540 mem orig 2.60 now 2.70
pcie orig 1.5 now 1.60

I use speedfan 4.33 - cpu in cpu gets to 56C, cpu on mobo gets to 49C,
idle is 35 and 40 respectively. case abmient is about 32 or so.

cpu was retail and I use the hsf that came with it as it has always run
very cool. I do not let the mobo control the hsf speed. As I recall, I set
the bios to shut the mobo down if cpu on mobo reaches 56C.

cpu 240 2640mhz --------- EDITED
HT (4x) 960
memory 480 (rather surprised this goes!)
pcie still at 100 (2500mhz) so next I will try playing with that and see.

But the question is, how high can I push the voltage on the cpu and be
relatively safe?

Case is Ultra with 120mm in in front, 120mm out in back, 80mm in on side
over cpu sorta, 350watt Ultra (600 on way) with 120mm fan.

tia!!!

I backed the cpu voltage to 1.520 which I see speedfan lists as 1.55 and
dropped cpu speed to 235 to get 3DMARK05 to not hang.

I then played around with the pci-e clock going up from 100 to 120 and got a
huge increase in 3DMARK05 from 3648 to 3666. wow. LOL. I don't understand why
it doesn't have a bigger effect, but..... That had no effect on stability.

For AMD's idea of the max value, try this doc. Note that this document
never got updated, as newer S939 processors came out, but presumably it is
still accurate. I'm not really crazy about AMD's skills at documentation.
I prefer a set of docs for each unique creation (one for 90nm, one for
65nm etc).

http://www.amd.com/us-en/assets/cont...docs/31411.pdf

On page 53, the value for VDD is the Vcore voltage.

"VDD supply voltage relative to VSS ¡V0.3 V to 1.65 V"

So don't go over 1.65V. Officially speaking.
Ah! thanks for that one!

Since the Vcore regulator on some boards, overvolts a bit at idle, I might
choose 1.60V as the highest BIOS setting for Vcore. That leaves 0.05V of
margin, in case the regulator chip is running on the high side.
I notice that though I have the bios set to 1.520, cpu-z says 1.552, so I am
guessing that this is what you are talking about here - off by 0.032 . But
0.05 sounds like a very good approximation!

For other overclocking info, you should try looking at some threads
here. They have a search feature, so you can narrow your search down
a bit.

http://www.xtremesystems.org/forums/forumdisplay.php?fX

For example, the thread on "Opteron 185" will give you some
ideas as to the range of conditions those people were using
with their gear. There are probably a few other threads
there, where S939 processor get mentioned.

http://www.xtremesystems.org/forums/...ead.php?t1814

Paul

Neither one of those links worked for me, but I will definitely poke around
that site.

Thanks!

:-) Compare what you just posted for those two links, to my original post.

Some characters in the links are corrupted (in your posting). "fX" got
changed to "fX". "t1814" became "t1814".

You might want to verify the stability of your overclock. Orthos should
be able to run for hours, without stopping and reporting an error, if
everything is stable. You can get Orthos here. If it stops in 30 seconds
and reports some kind of error, you've got some adjusting to do on
your motherboard settings.

http://sp2004.fre3.com/beta/beta2.htm

Paul



I just ran it for 5 1/2 minutes with no problems and stopped it out of
boredom. the can got up to 59C which does not make me particularly happy
(vcore=1.54, mult=11,fsb=235)

I widened the hole in the lucite on the side of my case and now have a
120mm fan blowing at the cpu, but I don't think it helped much (rel to
80mm)

I discovered that this abit does not allow incremental adjustment of
memory clocking - only DDR400, DDR433, DDR466, etc. I also found that when
set to 433, it would run, but sandra showed it to be clocked at 167 or
something like that instead of the 216 I would have expected.

This is all a product of boredom. I expect that in a few months I'll
spring for something like a Gigabyte GA-P35-DS3L and an E2160, or
somesuch, which can actually be overclocked!.

Tonight I may try lowering the mult and raising the fsb and testing the
mem that way. max HTT is about 330 so 313.
max CPU is about 240 so probably 230 or so.
The mem's spd is 2.5 4 4 8 400ddr.

Probably time to pull the Zalman off the Shuttle an35n and apply it to
this mobo...

Yeah, stuck the Zalman on there - an old CNPS7000A, all copper. I'm sure
putting fresh AS3 in there didn't hurt. Board wouldn't power so had to take
it all apart and put an extra layer of cardboard as the mobo's resisters were
touching the Zalman metal bottom plate. Ran Orthos for 10 min with no probs
and temp barely touched 50, vs 59 with the other hsf.

Can you access these links OK now ?

http://www.xtremesystems.org/forums/...splay.php?f=58
http://www.xtremesystems.org/forums/...d.php?t=131814

Paul

posted via 40tude Dialog:
(I had composed this on NMP:
no - both give me a "No Thread specified. If you followed a valid link,
please notify the administrator" message.

Using firefox. I get the same result if I use opera.

My reader is NewsMan Pro. Maybe it's doing something to the characters?

Okay, I see... I googled it and what my reader displays as fX is actually
f=58 (f equals 58)
and t box 1814 is actually t=131814 (t equals 131814)

Gotta be my reader.


Darned if I can see how...

Okay, 40tude Dialog does it correctly so I'll ask the author. )

so I just set up dialog again and I'll p[ost this to see if it works.

In the meantime, I checked one of those links and immediately came up with
a venice thread
http://www.xtremesystems.org/forums/...ad.php?t=59405

I think I have a low end chip - I only get 2588 with 1.54v and the first
person I looked at has 2750 with 1.40v . That explains a lot. Bed now.
I'll look at the links tomorrow. THANKS!

The header on your current message shows:

Content-Type: text/plain; charset="big5"
Content-Transfer_Encoding: 8bit

Maybe that has something to do with it ?

A message header from one of my messages has:

Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Transfer_Encoding: 8bit

and I don't know why it would be using 8bit.
Some of my other messages appear with the
header set to 7bit. So that implies some
non-printable characters could be floating around
in this thread.

Paul

On Tue, 16 Oct 2007 07:46:43 -0400, Paul wrote:

Kris wrote:
On Mon, 15 Oct 2007 22:08:03 -0400, Paul wrote:

Kris wrote:
On 10/15/2007 7:17:19 PM, "Kris" wrote:
On 10/15/2007 2:03:15 AM, Paul wrote:
Kris wrote:
On 10/14/2007 11:24:25 PM, Paul wrote:
Kris wrote:
On 10/13/2007 11:37:58 PM, "Kris" wrote:
system in sig.

I have been running this cpu at 220 since I got it (a year ago?), but I
never played with the voltages, and I didn't know whether it was the cpu
or memory or what that was limiting it (I couldn't go over 223 or so).

For some reason I started playing with voltages and nvidia system tune
that came with 163.71 (nforce 6.86) and relatively easily determined that
it was the cpu (venice core as I recall) so I upped cpu, and also memory
cpu orig 1.440 now 1.540 mem orig 2.60 now 2.70
pcie orig 1.5 now 1.60

I use speedfan 4.33 - cpu in cpu gets to 56C, cpu on mobo gets to 49C,
idle is 35 and 40 respectively. case abmient is about 32 or so.

cpu was retail and I use the hsf that came with it as it has always run
very cool. I do not let the mobo control the hsf speed. As I recall, I set
the bios to shut the mobo down if cpu on mobo reaches 56C.

cpu 240 2640mhz --------- EDITED
HT (4x) 960
memory 480 (rather surprised this goes!)
pcie still at 100 (2500mhz) so next I will try playing with that and see.

But the question is, how high can I push the voltage on the cpu and be
relatively safe?

Case is Ultra with 120mm in in front, 120mm out in back, 80mm in on side
over cpu sorta, 350watt Ultra (600 on way) with 120mm fan.

tia!!!

I backed the cpu voltage to 1.520 which I see speedfan lists as 1.55 and
dropped cpu speed to 235 to get 3DMARK05 to not hang.

I then played around with the pci-e clock going up from 100 to 120 and got a
huge increase in 3DMARK05 from 3648 to 3666. wow. LOL. I don't understand why
it doesn't have a bigger effect, but..... That had no effect on stability.

For AMD's idea of the max value, try this doc. Note that this document
never got updated, as newer S939 processors came out, but presumably it is
still accurate. I'm not really crazy about AMD's skills at documentation.
I prefer a set of docs for each unique creation (one for 90nm, one for
65nm etc).

http://www.amd.com/us-en/assets/cont...docs/31411.pdf

On page 53, the value for VDD is the Vcore voltage.

"VDD supply voltage relative to VSS ¡V0.3 V to 1.65 V"

So don't go over 1.65V. Officially speaking.
Ah! thanks for that one!

Since the Vcore regulator on some boards, overvolts a bit at idle, I might
choose 1.60V as the highest BIOS setting for Vcore. That leaves 0.05V of
margin, in case the regulator chip is running on the high side.
I notice that though I have the bios set to 1.520, cpu-z says 1.552, so I am
guessing that this is what you are talking about here - off by 0.032 . But
0.05 sounds like a very good approximation!

For other overclocking info, you should try looking at some threads
here. They have a search feature, so you can narrow your search down
a bit.

http://www.xtremesystems.org/forums/forumdisplay.php?fX

For example, the thread on "Opteron 185" will give you some
ideas as to the range of conditions those people were using
with their gear. There are probably a few other threads
there, where S939 processor get mentioned.

http://www.xtremesystems.org/forums/...ead.php?t1814

Paul

Neither one of those links worked for me, but I will definitely poke around
that site.

Thanks!

:-) Compare what you just posted for those two links, to my original post.

Some characters in the links are corrupted (in your posting). "fX" got
changed to "fX". "t1814" became "t1814".

You might want to verify the stability of your overclock. Orthos should
be able to run for hours, without stopping and reporting an error, if
everything is stable. You can get Orthos here. If it stops in 30 seconds
and reports some kind of error, you've got some adjusting to do on
your motherboard settings.

http://sp2004.fre3.com/beta/beta2.htm

Paul



I just ran it for 5 1/2 minutes with no problems and stopped it out of
boredom. the can got up to 59C which does not make me particularly happy
(vcore=1.54, mult=11,fsb=235)

I widened the hole in the lucite on the side of my case and now have a
120mm fan blowing at the cpu, but I don't think it helped much (rel to
80mm)

I discovered that this abit does not allow incremental adjustment of
memory clocking - only DDR400, DDR433, DDR466, etc. I also found that when
set to 433, it would run, but sandra showed it to be clocked at 167 or
something like that instead of the 216 I would have expected.

This is all a product of boredom. I expect that in a few months I'll
spring for something like a Gigabyte GA-P35-DS3L and an E2160, or
somesuch, which can actually be overclocked!.

Tonight I may try lowering the mult and raising the fsb and testing the
mem that way. max HTT is about 330 so 313.
max CPU is about 240 so probably 230 or so.
The mem's spd is 2.5 4 4 8 400ddr.

Probably time to pull the Zalman off the Shuttle an35n and apply it to
this mobo...

Yeah, stuck the Zalman on there - an old CNPS7000A, all copper. I'm sure
putting fresh AS3 in there didn't hurt. Board wouldn't power so had to take
it all apart and put an extra layer of cardboard as the mobo's resisters were
touching the Zalman metal bottom plate. Ran Orthos for 10 min with no probs
and temp barely touched 50, vs 59 with the other hsf.

Can you access these links OK now ?

http://www.xtremesystems.org/forums/...splay.php?f=58
http://www.xtremesystems.org/forums/...d.php?t=131814

Paul

posted via 40tude Dialog:
(I had composed this on NMP:
no - both give me a "No Thread specified. If you followed a valid link,
please notify the administrator" message.

Using firefox. I get the same result if I use opera.

My reader is NewsMan Pro. Maybe it's doing something to the characters?

Okay, I see... I googled it and what my reader displays as fX is actually
f=58 (f equals 58)
and t box 1814 is actually t=131814 (t equals 131814)

Gotta be my reader.


Darned if I can see how...

Okay, 40tude Dialog does it correctly so I'll ask the author. )

so I just set up dialog again and I'll p[ost this to see if it works.

In the meantime, I checked one of those links and immediately came up with
a venice thread
http://www.xtremesystems.org/forums/...ad.php?t=59405

I think I have a low end chip - I only get 2588 with 1.54v and the first
person I looked at has 2750 with 1.40v . That explains a lot. Bed now.
I'll look at the links tomorrow. THANKS!

The header on your current message shows:

Content-Type: text/plain; charset="big5"
Content-Transfer_Encoding: 8bit

Maybe that has something to do with it ?

A message header from one of my messages has:

Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Transfer_Encoding: 8bit

and I don't know why it would be using 8bit.
Some of my other messages appear with the
header set to 7bit. So that implies some
non-printable characters could be floating around
in this thread.

Paul

yep - the author responded to me this morning and said it would be
corrected in the next beta, which come along fairly quickly, which is good
as I much prefer that program for reading these little newsgroups!

That forum(s) is a wealth of info and will also be a help in looking for a
new board and cpu. I had long written this setup off but out of boredom
played with the voltages and doubled my paltry 4600 3DMARK03 score to 8600,
but that is definitely as far as it will go. Unless of course I get an
8600GTS, which is on the want list... )
One link I got there pointed me to
http://forums.anandtech.com/messagev...VIEWTMP=Linear
over at anandtech which turned out to be a super back-to-basics guide for
me for separating out the components. I didn't learn anything that I
didn't know before (ie my cpu is the limit) but it will help for next board
as it is a very quick approach and only bothered me with 3 or 4 reset-CMOSs
- not bad at all. I am ablt to run the memory faster (245) and run the cpu
at 10.5 instead of 11, but I gain no throughput advantage.

I put the old Zalman 7000CU on it yesterday and that cut the cpu sardine
can temp by 8-9 deg C. But I'm already at vcore=1.5 so no more!

The next setup will be my first Intel and from the looks of it it should be
very easy. I need speed like a hole in the head - it's all a toy value and
just plain fun. I always keep the last pile of components in case I blow
something!

Thanks!

"Kris" wrote in message
...
system in sig.

I have been running this cpu at 220 since I got it (a year ago?), but I
never
played with the voltages, and I didn't know whether it was the cpu or
memory
or what that was limiting it (I couldn't go over 223 or so).

I have a 2.2 3500+ Venice on an nForce 4 BFG (chaintec) board. I'm running
2 dual channel mem modules (400mhz). and it has been a big disappointment
as far as the overclockability goes. My previous Thunderbird 1.4
overclocked like a champ, even without any voltage increases. I had that
thing jacked up beyond belief and it ran wonderfully for years. This 3500+
is a dud as far as my knowledge can take it.

Regardless of what I do, I cant get it to run with stability beyond 206.
With or without any voltage changes. This is the best I can get with
stability. Given the voltages that you reported, I don't think it will
benefit your performance to increase any of it further. Even if you jacked
it up to 1.65, you wont get much practical help from it.

Here is the best I can get out of mine with no voltage changes. The
stability is perfect.

HTT 200 to 206
HT BUS 1000 to 1030
CPU Core 2200 to 2266
Mem Bus 400 to 412
PCIE Bus 2500 to 2850
PCI Bus 33 to 33
GPU Core 565 to 617
GPU Mem Speed 715 to 825.

The largest performance mark increases are to my 3d graphics. My HTT wont
tolerate anything over 206 or it starts to flake out. With or without
voltage changes. This system that I built is beyond a novice like me.

Good luck and keep a fire extinguisher handy

Michael

On Tue, 16 Oct 2007 23:07:48 -0400, Michael wrote:

"Kris" wrote in message
...
system in sig.

I have been running this cpu at 220 since I got it (a year ago?), but I
never
played with the voltages, and I didn't know whether it was the cpu or
memory
or what that was limiting it (I couldn't go over 223 or so).

I have a 2.2 3500+ Venice on an nForce 4 BFG (chaintec) board. I'm running
2 dual channel mem modules (400mhz). and it has been a big disappointment
as far as the overclockability goes. My previous Thunderbird 1.4
overclocked like a champ, even without any voltage increases. I had that
thing jacked up beyond belief and it ran wonderfully for years. This 3500+
is a dud as far as my knowledge can take it.

Regardless of what I do, I cant get it to run with stability beyond 206.
With or without any voltage changes. This is the best I can get with
stability. Given the voltages that you reported, I don't think it will
benefit your performance to increase any of it further. Even if you jacked
it up to 1.65, you wont get much practical help from it.

Here is the best I can get out of mine with no voltage changes. The
stability is perfect.

HTT 200 to 206
HT BUS 1000 to 1030
CPU Core 2200 to 2266
Mem Bus 400 to 412
PCIE Bus 2500 to 2850
PCI Bus 33 to 33
GPU Core 565 to 617
GPU Mem Speed 715 to 825.

The largest performance mark increases are to my 3d graphics. My HTT wont
tolerate anything over 206 or it starts to flake out. With or without
voltage changes. This system that I built is beyond a novice like me.

Good luck and keep a fire extinguisher handy

Michael

take a look at
http://forums.anandtech.com/messagev...VIEWTMP=Linear
to see what is holding it down... It simply may just be another pathetic
Venice.like mine!.. I'm sure that happens with any chip. All one can do
is try to pick a mobo and chip that has had good experience with ocing with
folks and go for it, but what you get is anyone's guess. The GPU increases
look good. I can't do anything with mine as it came already OC'ed at
500/1000, as will the new one at 700/2100. I also think the days of AMD
really OCing well are long gone - time past when we would cut and paste
traces on the chip. no more! I actually had this thing running at 2700mhz
last night, but it just is not stable. I can get it to 238 with vcore of
1.62 which yields 1.65 all giving me a huge 2618mhz, and that is stable,
but I had to put the Zalman 7000CU on the CPU and a little fan pointing
straight at the memory to do that. I wouldn't want to bet my life on it!!!
I think I was running 235 at 1.52 vcore, so that's a lot of additional
volts for very little gain in speed.

I had my heart set on the Gigabyte p35-ds3l board and a dual-core chip.
Then I thought I should at least allow for the possibility of SLI. But
then I went for the 8600gts and I'm not likely to pop another 150 just to
get sli - better to pop for an 8800 a few years from now...
--
-Kris
ABIT KN8 Ultra(NForce4Ultra),
Ath64 3500+ Venice,
2x512MB Geil DDR400 pc3200 dual chan,
Leadtek PX6600GT TDH 128mb,
HTT/FSB max 313, CPU (pathetic) max 235 at vcore 1.54, mem max about 245,
Ultra 600W PS (in mail - tomorrrow),
MSI NX8600GTS T2D256E-OC(in mail - Fri),
Leadtek WinFast TV2000 XP Expert TV card,
XP Pro SP 2++++

On Wed, 17 Oct 2007 21:09:15 -0400, Kris wrote:

On Tue, 16 Oct 2007 23:07:48 -0400, Michael wrote:

"Kris" wrote in message
...
system in sig.

I have been running this cpu at 220 since I got it (a year ago?), but I
never
played with the voltages, and I didn't know whether it was the cpu or
memory
or what that was limiting it (I couldn't go over 223 or so).

I have a 2.2 3500+ Venice on an nForce 4 BFG (chaintec) board. I'm running
2 dual channel mem modules (400mhz). and it has been a big disappointment
as far as the overclockability goes. My previous Thunderbird 1.4
overclocked like a champ, even without any voltage increases. I had that
thing jacked up beyond belief and it ran wonderfully for years. This 3500+
is a dud as far as my knowledge can take it.

Regardless of what I do, I cant get it to run with stability beyond 206.
With or without any voltage changes. This is the best I can get with
stability. Given the voltages that you reported, I don't think it will
benefit your performance to increase any of it further. Even if you jacked
it up to 1.65, you wont get much practical help from it.

Here is the best I can get out of mine with no voltage changes. The
stability is perfect.

HTT 200 to 206
HT BUS 1000 to 1030
CPU Core 2200 to 2266
Mem Bus 400 to 412
PCIE Bus 2500 to 2850
PCI Bus 33 to 33
GPU Core 565 to 617
GPU Mem Speed 715 to 825.

The largest performance mark increases are to my 3d graphics. My HTT wont
tolerate anything over 206 or it starts to flake out. With or without
voltage changes. This system that I built is beyond a novice like me.

Good luck and keep a fire extinguisher handy

Michael

take a look at
http://forums.anandtech.com/messagev...VIEWTMP=Linear
to see what is holding it down... It simply may just be another pathetic
Venice.like mine!.. I'm sure that happens with any chip. All one can do
is try to pick a mobo and chip that has had good experience with ocing with
folks and go for it, but what you get is anyone's guess. The GPU increases
look good. I can't do anything with mine as it came already OC'ed at
500/1000, as will the new one at 700/2100. I also think the days of AMD
really OCing well are long gone - time past when we would cut and paste
traces on the chip. no more! I actually had this thing running at 2700mhz
last night, but it just is not stable. I can get it to 238 with vcore of
1.62 which yields 1.65 all giving me a huge 2618mhz, and that is stable,
but I had to put the Zalman 7000CU on the CPU and a little fan pointing
straight at the memory to do that. I wouldn't want to bet my life on it!!!
I think I was running 235 at 1.52 vcore, so that's a lot of additional
volts for very little gain in speed.

I had my heart set on the Gigabyte p35-ds3l board and a dual-core chip.
Then I thought I should at least allow for the possibility of SLI. But
then I went for the 8600gts and I'm not likely to pop another 150 just to
get sli - better to pop for an 8800 a few years from now...

Well, you can never tell... The vid card works and has nice colors. Happy
with that! The Ultra600W ps wouldn't post the ABIT mobo, and so I got out
my old Shuttle AN35N and it wouldn't post that either. Very strange as all
the voltages seemed to be okay. Wierd. Ultra gave me an RMA# so I'll try
to ship that back tomorrow. They had a record of my having just bought it
from Tigerdirect. Everything is doing fine with the little 350W ps, so go
figure!
--
-Kris
ABIT KN8 Ultra(NForce4Ultra),
Ath64 3500+ Venice,
2x512MB Geil DDR400 pc3200 dual chan,
MSI NX8600GTS T2D256E-OC,
HTT/FSB max 313, CPU max 235 at vcore 1.54, mem max about 245,
Ultra 350W PS,
Leadtek WinFast TV2000 XP Expert TV card,
XP Pro SP 2++++