Pentium Pro 200 - 256k cache vs 512k & 1MB cache

Chas. wrote:


I found some info on Amperage requirements for VRMs for the various
PPro 200 CPUs. It may not be completely acurate but its a start.

Some one from Tyan reported that the 1Mb PPro draws 2.5 more amps than
the 256/512K PPros, and would require a higher capacity VRM.

256K should work with 11.2 Amp VRMs
512k should work with 12.4 Amp VRMs
1 MB should work with 15.5 Amp VRMs

Some VRM Specs for PPro CPUs

VXI VRM 073-20674-07 11.2amp
VXI VRM 073-20713-01 12.4amp 5VIN
VXI VRM 073-20714-01 ?
VXI VRM 073-20715-01 14.5amp 5/12VIN 1.8/3.5VOUT
VXI VRM 073-20674-07X 12.4amp
VXI VRM 073-20740-01 16.0amp
Raytheon RCB001 12.4amp
Raytheon RCB006 15.5amp
Canary 225529-001 ?
Corsair SPX560KM ?

The onboard voltage regulator on your board may not be able to support
the 1MB PPros.

Chas.

Great info, I actually have no idea what capacity my VRMs are- I suspect
they're the RCB001's, though, so I'll have to get that straightened out.
I do know Tyan said the board could be used with the 1mb chips, but
that's obviously dependent on the proper hardware.

Thanks for that post.
J.

--
Pull the '-m*ps' for mail.

"NuT CrAcKeR" wrote in message ...
the more cache will make it feel more snappy, and help most applications to
some degree.

However, if you really want to get jiggy with this, try to locate the PII-OD
chips. 333Mhz on the 66Mhz bus, with 512K fullspeed cache. It performs like
a PII-450 (in my opinon), and works in most systems that are dual
compatable.

I upgraded my ALR Evolution Dual 6 workstation (my webserver), and its fine.
It serves up the sites i need like a bat outta hell, and I put an ALR 3drive
Raid cage in there, and an Ultra SCSI Array controller (40MB/s) with 32M
cache...

Its a screamer, and will be for sale soon...

NuTs

snip

I put in the second CPU tonight. Installed the VRM, changed the
"undocumented"
jumper and booted it up. Started right up but didn't detect the 2nd
CPU. Rebooted and bingo, "Second CPU Initialized".

It's a Hemi 6 Pack!

The heatsink that came with the CPU had a bad fan. I pulled a big one
off of an old Athlon Slot A heatsink and fixed the problem.

I also put in an extra 64MB of memory for 224MB total. I found 512MB
for a good price on Ebay and I also picked up 2 matched PPro 200 512k
CPUs for $12! This motherboard uses 1 onboard voltage regulator and 1
VRM. The onboard VR probably can't take the current draw of a PPro 200
1Mb CPU.

Chas.

J Rennert wrote in message ...
Chas. wrote:



Do the PPro 200s with 512k or 1M cache offer much in the way of
performance advantage over the 256k CPUs.

Please no comments about getting a more modern system. I have PCs all
over the house. This is sort of like rebuilding an old car.

Chas.

Heh, I know what you mean- I searched long and hard for a dual Tyan PPro
board and some 1mb chips, then found eight 128mb SIMMs to stuff in there
to max it out. Haven't found time to get it going yet, but it'll run
gentoo for sure. The P6 was/is an underrated chip.

I believe it's generally accepted that you feed webservers clock speed
and DB servers cache. All relative to the age of your chip, anyway.

Good luck with the project, it sounds like fun. Make sure and check to
see if you need an additional VRM- those can sometimes cost more than
the old CPU.
J.

I put in the second CPU tonight. Installed the VRM, changed the
"undocumented"
jumper and booted it up. Started right up but didn't detect the 2nd
CPU. Rebooted and bingo, "Second CPU Initialized".

It's a Hemi 6 Pack!

The heatsink that came with the CPU had a bad fan. I pulled a big one
off of an old Athlon Slot A heatsink and fixed the problem.

I also put in an extra 64MB of memory for 224MB total. I found 512MB
for a good price on Ebay and I also picked up 2 matched PPro 200 512k
CPUs for $12! This motherboard uses 1 onboard voltage regulator and 1
VRM. The onboard VR probably can't take the current draw of a PPro 200
1Mb CPU.

Chas.

"Chas" == Chas writes:


Chas I put in the second CPU tonight. Installed the VRM, changed the
Chas "undocumented" jumper and booted it up. Started right up but
Chas didn't detect the 2nd CPU. Rebooted and bingo, "Second CPU
Chas Initialized".

What is this undocumented jumper? I never changed any jumper when
adding a CPU, and it worked fine for me. Did I miss something all
these years?

Let me know

Thanks

"Chas" == Chas writes:

Chas Spam Me Please wrote in message
Chas ...
"Chas" == Chas writes:


Chas I put in the second CPU tonight. Installed the VRM, changed the
Chas "undocumented" jumper and booted it up. Started right up but
Chas didn't detect the 2nd CPU. Rebooted and bingo, "Second CPU
Chas Initialized".
What is this undocumented jumper? I never changed any jumper when
adding a CPU, and it worked fine for me. Did I miss something all
these years?

Let me know

Thanks

Chas Hmmm, I guess you're right. There is a W10 jumper next to the
Chas top CPU socket on Micronics W6-LI mobos. I read somewhere in an
Chas NG message that this jumper needs to be moved from 1-2 to 2-3
Chas for dual processors. It turns out that the jumper controls
Chas pipelined bus cycles... whatever that is. It's function isn't
Chas described in the less than useless Micron manual. The Micronics
Chas motherboard manual is a little better.

Okay - is it faster with this jumper enabled? Anyone has done any
benchmarks?

Thanks for the info

Okay - is it faster with this jumper enabled? Anyone has done any
benchmarks?

Why do they always speak of "faster"? Are stability and
determinability not positive values? On the other hand, they do not
deserve it better.

--
------- http://grassomusic.de -------

"Grasso" == Grasso writes:

Okay - is it faster with this jumper enabled? Anyone has done any
benchmarks?

Grasso Why do they always speak of "faster"? Are stability and
Grasso determinability not positive values? On the other hand, they
Grasso do not deserve it better.

Just curious. Those are great questions also. Does anyone have the
answer for any of the above questions?

Thanks

On Wed, 13 Aug 2003 00:52:59 -0500, Spam Me Please wrote:

"Grasso" == Grasso writes:

Okay - is it faster with this jumper enabled? Anyone has done any
benchmarks?

Grasso Why do they always speak of "faster"? Are stability and
Grasso determinability not positive values? On the other hand, they
Grasso do not deserve it better.

Just curious. Those are great questions also. Does anyone have the
answer for any of the above questions?

Here's an animation, with 4 CPUs of the PPro GTL+ Bus and memory bus
signalling showing memory transactions:
http://www.warthman.com/ex-ppro.htm... URL originally supplied by Dave
Wang IIRC. Does your system work "properly" in dual mode with the jumper
in either position? I haven't looked at the details of CPU strapping but
I'm wondering if the jumper might be to allow for strapping the CPU to
eliminate the overhead of bus arbitration in single CPU mode - note the
BREQn# signals. Hmmm, not sure what would happen if it was strapped
wrongly for dual mode.

The other possibility would be setting the size of the IOQ - Chas mentioned
something about it possibly having something to do with pipelining of
addresses. Maybe some tests with Stream would show up any differences.

Rgds, George Macdonald

"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??

(George Macdonald) wrote in message ...
On Wed, 13 Aug 2003 00:52:59 -0500, Spam Me Please wrote:

"Grasso" == Grasso writes:

Okay - is it faster with this jumper enabled? Anyone has done any
benchmarks?

Grasso Why do they always speak of "faster"? Are stability and
Grasso determinability not positive values? On the other hand, they
Grasso do not deserve it better.

Just curious. Those are great questions also. Does anyone have the
answer for any of the above questions?

Here's an animation, with 4 CPUs of the PPro GTL+ Bus and memory bus
signalling showing memory transactions:
http://www.warthman.com/ex-ppro.htm... URL originally supplied by Dave
Wang IIRC. Does your system work "properly" in dual mode with the jumper
in either position? I haven't looked at the details of CPU strapping but
I'm wondering if the jumper might be to allow for strapping the CPU to
eliminate the overhead of bus arbitration in single CPU mode - note the
BREQn# signals. Hmmm, not sure what would happen if it was strapped
wrongly for dual mode.

The other possibility would be setting the size of the IOQ - Chas mentioned
something about it possibly having something to do with pipelining of
addresses. Maybe some tests with Stream would show up any differences.

Rgds, George Macdonald

George, Here's the info from the Micronics Manual:

Jumper settings for the Host Bus Agents pipelining mode.
Host Bus Agents are devices connected to the CPU local
bus which include the CPUs and the 440FX chipset. Host
Bus Agents can generate, up to eight at a time, multiple
overlapped (or Pipelined) transactions ( or Bus Cycles).

W10 Jumper Settings - Pipelining Mode Settings
Normal (Default) - Open
Reserved - Closed

Chas.

(Chas.) wrote in message . com...

I was having some minor problems with the W6-LI motherboard. With
512MB of Memory, the screen stayed black for at least 30 seconds
before POST at bootup or when rebooting. If I took out 256MB the
problem went away. Also it counted up all of the memory during reboots
which I guess is standard for server class motherboards. It's new
memory and I tested all of it with MemTest and DocMem.

Anyway after much research on the web, I decided to update the BIOS to
see if it would fix some problems. Catasrophy! The "correct" BIOS
update from the PCMicron site hosed my BIOS.

All is not lost because I picked up a late model Intel PR440FX cheap.
It came with matched PPro 200 CPUs, a VRM, memory, onbaord SCSI and
sound plus lots of extras. I had to jury rig an I/O shield and I put
on new heatsink/fans. I'm running the PPro 200 512k CPUs at 233Mhz.
They're cool to the touch.

It screams with Win98SE (for a single 233 CPU) and last night I loaded
RedHat 9.1 SMP. The Intel board has the more recent AMI BIOS than the
old Phoenix BIOS on the W6-LI. I can configure APM in the BIOS which
gives automatic ATX shutdown in Win98 and that's a feature that I
like. The PR440FX "seems" a lot faster with the PPro 200 512k CPUs at
233 than the W6-LI with PPro 200 256k CPUs at 233.

Chas.