VOGONS


Pentium Pro multiplier options

Topic actions

First post, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Does anyone know the full extent of the Pentium Pro's multiplier options? I'm attaching a snippet from Intel's P6 Family of Processors, Hardware Developer’s Manual, September 1998

Pentium_Pro_multipliers.jpg
Filename
Pentium_Pro_multipliers.jpg
File size
340.9 KiB
Views
1702 views
File license
Fair use/fair dealing exception

Does this mean the PPRO's multiplier can be set from 1.5x up to 8x? And does the second table imply that this can be done via software as well? The Intel VS440FX motherboard I'm using has jumper settings (documented and undocumented) for 2.0, 2.5, 3.0, and 3.5. I've heard of other boards having a 4x feature as well.

If it is adjustable in software, does anyone know of some software which can do this?

What's the fastest stable overclock of a Pentium Pro with and without voltage modification?

If software multiplier adjustment is not possible, has anyone modified a VS440FX motherboard for 4.0x or 4.5x operation?

Plan your life wisely, you'll be dead before you know it.

Reply 3 of 25, by red-ray

User metadata
Rank Oldbie
Rank
Oldbie

AFAIK x4 is the maximum that is possible and looking @ http://valid.x86.fr/top-cpu/496e74656c2050656 … 0323333204d487a the record is about 266MHz which supports this. I recall that the VS440FX only supports up to x3.5, there are four combinations that are x 2.0 to x 3.5.

With Socket 8 then the Pentium II Overdrive runs @ x5 (333MHz/300MHz) and only the FSB setting affects this.

As for getting software to change the multiplier on-the-fly then I doubt it, that said it would be possible for a motherboard to have a facility to do this that would be specific to the motherboard, but I have never heard of one. I also suspect after changing it a reboot would be needed for it to take effect.

If you change the jumpers while the system is running does the CPU speed change?

Last edited by red-ray on 2019-10-19, 15:25. Edited 1 time in total.

Reply 4 of 25, by Horun

User metadata
Rank l33t++
Rank
l33t++
feipoa wrote:

Does this mean the PPRO's multiplier can be set from 1.5x up to 8x?

It appears that way. Maybe Intel did that for specific board designs that might use the 30 or 33Mhz clock to feed the cpu and not a 60 or 66mhz clock that most popular boards use. Just a theory..

Hate posting a reply and then have to edit it because it made no sense 😁 First computer was an IBM 3270 workstation with CGA monitor. Stuff: https://archive.org/details/@horun

Reply 5 of 25, by red-ray

User metadata
Rank Oldbie
Rank
Oldbie
feipoa wrote:

Does this mean the PPRO's multiplier can be set from 1.5x up to 8x?

No, it means some P6 CPUs can be set from x1.5 up to x8. I recall this was further extended for P-III (Coppermine) with a Stepping >= 3 to allow higher than x8.

Reply 6 of 25, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

Most likely was implemented for future proofing. Too bad Socket 8 platform died quickly.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 7 of 25, by luckybob

User metadata
Rank l33t
Rank
l33t

socket 8 is *almost* pin compatible with socket 370. The differences being mostly voltages. If you look at 8/370 converters they have no glue logic, save for a bit for voltage regulation.

That said both the overdrive and celerons have locked multipliers. So the OD chip will only run at 5x. I assume the celeron chips will run identically.

Personally, I've ran a couple of the P-Pro chips I own at 4.5x (300mhz) but I had to drop it to 60mhz fsb (270mhz), and it wasn't exactly stable.

It is a mistake to think you can solve any major problems just with potatoes.

Reply 8 of 25, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Sounds like 266 MHz is a realistic maximum. Has anyone checked for undocumented FSB's on this board (VS440FX)? Being an Intel board, it seems unlikely.

So, to go from 3x to 4x, I'd need to flip the polarity of IGNNE# and A20M#? Perhaps with jumpers removed I can realise this with some cabling?

maxtherabbit wrote:

mine run stable at 233MHz with no modification or voltage boost, but that's as high as I was able to get them - 266 was a non-POST

How many CPUs did you try? And which flavour - 256/512/1M?

Plan your life wisely, you'll be dead before you know it.

Reply 9 of 25, by maxtherabbit

User metadata
Rank l33t
Rank
l33t
feipoa wrote:
Sounds like 266 MHz is a realistic maximum. Has anyone checked for undocumented FSB's on this board (VS440FX)? Being an Intel […]
Show full quote

Sounds like 266 MHz is a realistic maximum. Has anyone checked for undocumented FSB's on this board (VS440FX)? Being an Intel board, it seems unlikely.

So, to go from 3x to 4x, I'd need to flip the polarity of IGNNE# and A20M#? Perhaps with jumpers removed I can realise this with some cabling?

maxtherabbit wrote:

mine run stable at 233MHz with no modification or voltage boost, but that's as high as I was able to get them - 266 was a non-POST

How many CPUs did you try? And which flavour - 256/512/1M?

I only have 2 total PPro chips. 200MHz rated, 256kB cache

Both running together in SMP

Reply 10 of 25, by luckybob

User metadata
Rank l33t
Rank
l33t
feipoa wrote:
Sounds like 266 MHz is a realistic maximum. Has anyone checked for undocumented FSB's on this board (VS440FX)? Being an Intel […]
Show full quote

Sounds like 266 MHz is a realistic maximum. Has anyone checked for undocumented FSB's on this board (VS440FX)? Being an Intel board, it seems unlikely.

So, to go from 3x to 4x, I'd need to flip the polarity of IGNNE# and A20M#? Perhaps with jumpers removed I can realise this with some cabling?

maxtherabbit wrote:

mine run stable at 233MHz with no modification or voltage boost, but that's as high as I was able to get them - 266 was a non-POST

How many CPUs did you try? And which flavour - 256/512/1M?

depends on the clock gen on the board. my asus and tyan boards all have the same chip as pentium cpu's of the era. they have 2 jumpers, for a total of 4 settings. 50, 60 & 66mhz are on the datasheet with a "testmode". at least concerning the clockgen i have in front of me: ICS9159C.

233mhz is usually very easy to attain. 266 is uncommon and only a select few ever posted at 300. And this was 100% based on the L2 cache. 256k chips overclocked well, 512k less so, and at 1mb 233 is optimistic.

At least with the batch that I got last year. Naturally a clockgen with more options will obviously be more useful, but as I understand it, there was almost zero demand for >66mhz until SS7/slot1 was a thing. My P2B-DS runs with a newer ICS9250CF-08. and that is a completely different chip.

It is a mistake to think you can solve any major problems just with potatoes.

Reply 11 of 25, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

but as I understand it, there was almost zero demand for >66mhz

75mhz option was quite popular before that.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 12 of 25, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Thanks guys. I'll see what PLL is on the board to see if there is a hidden 75 MHz option. EDIT: Its an ICS9169M.. datecode with a -01 suffix. https://www.idt.com/document/dst/9169-01-datasheet - 50/60/66 MHz options + TestMode freq, whatever freq. that is. Datasheet does mention "Output frequency ranges to 100 MHz (depending on option)".

LuckyBob, what's the chance that any of my 256 or 512 chips will run at 262/266 MHz reliably? I've two of each. If its only 25%, then I won't bother seeing if I can hack the multiplier jumpers to 4x. What about 1M at 233 MHz - 50% chance? Percents at stock voltage.

By the way, I'm running an Adaptec AAR-2400A in RAID 10 with a 128 MB RAM stick. The PPRO seemed like a fitting system for this. It was free and so were the four IDE drives, so I decided to free up some space in the storage bin. I haven't heard of anyone running this RAID controller, so I thought I'd throw it out there.

Plan your life wisely, you'll be dead before you know it.

Reply 15 of 25, by luckybob

User metadata
Rank l33t
Rank
l33t
feipoa wrote:

Thanks guys. I'll see what PLL is on the board to see if there is a hidden 75 MHz option. EDIT: Its an ICS9169M.. datecode with a -01 suffix. https://www.idt.com/document/dst/9169-01-datasheet - 50/60/66 MHz options + TestMode freq, whatever freq. that is. Datasheet does mention "Output frequency ranges to 100 MHz (depending on option)".

LuckyBob, what's the chance that any of my 256 or 512 chips will run at 262/266 MHz reliably? I've two of each. If its only 25%, then I won't bother seeing if I can hack the multiplier jumpers to 4x. What about 1M at 233 MHz - 50% chance? Percents at stock voltage.

By the way, I'm running an Adaptec AAR-2400A in RAID 10 with a 128 MB RAM stick. The PPRO seemed like a fitting system for this. It was free and so were the four IDE drives, so I decided to free up some space in the storage bin. I haven't heard of anyone running this RAID controller, so I thought I'd throw it out there.

anecdotal evidence what it is, I'd wager 90% of 256k chips will do 266 easy. 512@266 is about 66%, and 1m@266 is about 33%. I haven't met a 200/256 that wouldn't do 233, and every 512 I've personally tested had zero issues at 233 (i own like 10x). but I did notice maybe 1 in 4 1M chips did NOT like 233. They all posted, as I recall, but about 25% did not like it and had memory errors. (cache errors)

I do own one of those raid controllers, but I think mine is a 6-drive affair. I do own a 12-drive one that I did use briefly. I still own it, and I have a desire to get 12x 20gb BIGFOOT drives and have a bit of fun. ^.^ (If anyone wants to see this come to pass, let me know via PM)
I love scsi, and I love raid. But once a system has pci 2.1 (pentium and up) I go straight to sata. Its just so fracking easy, and cheap.

It is a mistake to think you can solve any major problems just with potatoes.

Reply 17 of 25, by feipoa

User metadata
Rank l33t++
Rank
l33t++
maxtherabbit wrote:

IIRC when I tried my board with just one processor it would POST at 266 with the 256kB CPU, but it would not with SMP

Refering to 256, 512, or 1024K at 266 MHz? POST at 266 MHz - did it run WinNT4/W2K/Win9x stable enough to run demanding apps?

Plan your life wisely, you'll be dead before you know it.

Reply 18 of 25, by maxtherabbit

User metadata
Rank l33t
Rank
l33t
feipoa wrote:
maxtherabbit wrote:

IIRC when I tried my board with just one processor it would POST at 266 with the 256kB CPU, but it would not with SMP

Refering to 256, 512, or 1024K at 266 MHz? POST at 266 MHz - did it run WinNT4/W2K/Win9x stable enough to run demanding apps?

No it didn't

256kB cache as stated in the post you quoted

Reply 19 of 25, by feipoa

User metadata
Rank l33t++
Rank
l33t++

lol! I had skimmed through the whole thread looking for where you mentioned the cache size and it was right in front of me. So your 256K chip would not run Windows at 266 MHz in single CPU mode. I'll probably hold off on attempting a multiplier hack. Perhaps 1M at 233 MHz is sufficient for this system.

Plan your life wisely, you'll be dead before you know it.