PPro on a BX?

[P2B]

I intercepted some old systems on their way to the dumpster today, and
grabbed a couple of interesting items I hadn't seen before: Asus C-P6S1
Socket-8 to Slot-1 adapters, complete with PPro 200Mhz/1MB L2 CPUs. They
were on Asus 440FX boards, KN97-X IIRC.

I tried them on a few Asus 440BX boards (with FSB and multiplier jumpers
set appropriately), and they POST but freeze after announcing the BIOS
version - no CPU identification is displayed. My POST diagnostic card
says x0D, which is undocumented but IME means "BIOS does not support
CPU" - so I tried a rom.by patched BIOS. It got a little further -
correctly displays the CPU type (but not speed) and POST code x0E before
freezing.

The 440BX datasheet does not *explicitly* state PPro CPUs are supported,
but quotes like "A Pentium® Pro processor-based system with the Intel®
440BX AGPset supports 4 GB of addressable memory space and 64 KB + 3 of
addressable I/O space" strongly suggests they are.

Anyone know if this is a chipset issue, or strictly BIOS? I'd like to
use the processors, but the only spare motherboards available have 440BX
chipsets.

TIA

P2B

 

[David Maynard]

I intercepted some old systems on their way to the dumpster today, and
grabbed a couple of interesting items I hadn't seen before: Asus C-P6S1
Socket-8 to Slot-1 adapters, complete with PPro 200Mhz/1MB L2 CPUs. They
were on Asus 440FX boards, KN97-X IIRC.

I tried them on a few Asus 440BX boards (with FSB and multiplier jumpers
set appropriately), and they POST but freeze after announcing the BIOS
version - no CPU identification is displayed. My POST diagnostic card
says x0D, which is undocumented but IME means "BIOS does not support
CPU" - so I tried a rom.by patched BIOS. It got a little further -
correctly displays the CPU type (but not speed) and POST code x0E before
freezing.

The 440BX datasheet does not *explicitly* state PPro CPUs are supported,
but quotes like "A Pentium® Pro processor-based system with the Intel®
440BX AGPset supports 4 GB of addressable memory space and 64 KB + 3 of
addressable I/O space" strongly suggests they are.

Anyone know if this is a chipset issue, or strictly BIOS? I'd like to
use the processors, but the only spare motherboards available have 440BX
chipsets.

TIA

P2B

I'm not sure which of the 440BX data sheets you've got but

Intel® 440BX AGPset:
82443BX Host Bridge/Controller
Datasheet
April 1998

makes it pretty clear that BX works with the Pentium Pro although it
doesn't support all it's capabilities (such as PCI ECC and > 4 gig memory).

You'd think it would work because a P-II is basically a repackaged (slot
cart), fewer features (removed the 'mission critical' ones), less expensive
(slower cache) Pentium Pro.

Because of that I'd say it's a BIOS problem.

 

[P2B]

I'm not sure which of the 440BX data sheets you've got but

Intel® 440BX AGPset:
82443BX Host Bridge/Controller
Datasheet
April 1998

Yup

makes it pretty clear that BX works with the Pentium Pro although it
doesn't support all it's capabilities (such as PCI ECC and > 4 gig memory).

You'd think it would work because a P-II is basically a repackaged (slot
cart), fewer features (removed the 'mission critical' ones), less expensive
(slower cache) Pentium Pro.

Because of that I'd say it's a BIOS problem.

Must be - since it POSTs :-?

Googling for PPro & 440BX doesn't yield much. I don't know of any BX
boards with 'official' support for PPro.

Has anyone got it to work?

 

[David Maynard]

Yup



Must be - since it POSTs :-?

Googling for PPro & 440BX doesn't yield much. I don't know of any BX
boards with 'official' support for PPro.

Neither do I and it doesn't surprise me because, as much as that spec sheet
says about Pentium Pro, why would anyone want to? The Pro isn't a 100Mhz
FSB processor and BX was specifically made for the 100Mhz FSB P-IIs.

Nearest thing I found to a Pro on a P-II motherboard, besides the Asus that
slot adapter was made for, is this one

http://www.fonck.nl/computer/asuslxpro.htm

(note that he mentions only a particular rev BIOS works)

And then this, sort of, 'PCChips, like, dual socket' (ooo that hurt.. but
then it felt good too <g>) supermicro board with a socket 8 and Slot-1
both. (That'll confuse the heck out of someone with a PPGA celeron)

http://www.supermicro.com/newsroom/pressreleases/1997/press050797.cfm

You hack around in the BIOS. Can't you 'steal' the CPU codes from one of
those and patch it onto the one you want?

Although, I beginning to suspect that it isn't just processor
identification, or even microcode, and that the Pro needs the registers set
up a bit differently that the P-II.

 

[David Maynard]

Neither do I and it doesn't surprise me because, as much as that spec
sheet says about Pentium Pro, why would anyone want to? The Pro isn't a
100Mhz FSB processor and BX was specifically made for the 100Mhz FSB P-IIs.

Nearest thing I found to a Pro on a P-II motherboard, besides the Asus
that slot adapter was made for, is this one

http://www.fonck.nl/computer/asuslxpro.htm

(note that he mentions only a particular rev BIOS works)

And then this, sort of, 'PCChips, like, dual socket' (ooo that hurt..
but then it felt good too <g>) supermicro board with a socket 8 and
Slot-1 both. (That'll confuse the heck out of someone with a PPGA celeron)

http://www.supermicro.com/newsroom/pressreleases/1997/press050797.cfm

You hack around in the BIOS. Can't you 'steal' the CPU codes from one of
those and patch it onto the one you want?

Although, I beginning to suspect that it isn't just processor
identification, or even microcode, and that the Pro needs the registers
set up a bit differently that the P-II.

Well, I found something else that doesn't necessarily resolve anything but
is, nonetheless, rather interesting. This guy's Dell BX motherboard

http://www.psychowire.com/mainboards/slocket.html

reported his new P-III 700 on a slotket as a "Pentium Pro 500."

Apparently there was something in that BIOS that had, even if nothing else,
at least the string "Pentium Pro" in it.

Maybe you should be looking for the oldest BIOS you can find for those
boards in the hopes they might have 'Pro' remnants left.

 

[Roland Scheidegger]

I intercepted some old systems on their way to the dumpster today, and
grabbed a couple of interesting items I hadn't seen before: Asus C-P6S1
Socket-8 to Slot-1 adapters, complete with PPro 200Mhz/1MB L2 CPUs. They
were on Asus 440FX boards, KN97-X IIRC.

Wow, didn't know such adapters exist...

Anyone know if this is a chipset issue, or strictly BIOS? I'd like to
use the processors, but the only spare motherboards available have 440BX
chipsets.

Is it possible the PPro might have some erratas which make it necessary
to have PCB/bios workarounds? I haven't looked at the spec update, but
such errors are not uncommon. And the workarounds might not be present
on the P2B.
And I hope the bios doesn't use some mmx code to initialize it...
Also, google shows some people got it to work in P2L97 boards, but not
with all bios versions (only the oldest one worked?), which probably
indicates it is a bios problem on the p2b too.

Roland

 

[Hamman]

I intercepted some old systems on their way to the dumpster today, and
grabbed a couple of interesting items I hadn't seen before: Asus C-P6S1
Socket-8 to Slot-1 adapters, complete with PPro 200Mhz/1MB L2 CPUs. They
were on Asus 440FX boards, KN97-X IIRC.

I tried them on a few Asus 440BX boards (with FSB and multiplier jumpers
set appropriately), and they POST but freeze after announcing the BIOS
version - no CPU identification is displayed. My POST diagnostic card says
x0D, which is undocumented but IME means "BIOS does not support CPU" - so
I tried a rom.by patched BIOS. It got a little further - correctly
displays the CPU type (but not speed) and POST code x0E before freezing.

The 440BX datasheet does not *explicitly* state PPro CPUs are supported,
but quotes like "A Pentium® Pro processor-based system with the Intel®
440BX AGPset supports 4 GB of addressable memory space and 64 KB + 3 of
addressable I/O space" strongly suggests they are.

Anyone know if this is a chipset issue, or strictly BIOS? I'd like to use
the processors, but the only spare motherboards available have 440BX
chipsets.

TIA

P2B

I wouldnt piss about with them too boards too much, BX chipsets can take
upto 1GHz P3's which is somewhat faster than a PPro

hamman

 

[P2B]

I wouldnt piss about with them too boards too much, BX chipsets can take
upto 1GHz P3's which is somewhat faster than a PPro

hamman

BX can do better than that The board was running a Tualatin P3-S @
1575Mhz (150Mhz FSB) before I pulled it to try the PPro.

I "piss about" for interest's sake. I've explored the high end of BX
capabilities, now curious about the low end...

P2B

http://tipperlinne.com/p2bmod

 

[P2B]

LX chipset... getting warmer

Perhaps not... the 'particular rev' is the very first release for that
board, and he says later versions didn't work :-(

FX chipset - same as the Asus board the adapter was made for.

Sure I can... but it isn't a microcode problem - POST doesn't get
anywhere near that far,

My thinking too - I hoped Bios Patcher would help. Quoting from the manual:

______________________________________
=> Patcher can add support of CPU:

- AMD K6/K6-2/K6-III/K6-2+/K6-III+
- Intel Pentium Pro/Pentium II/Pentium III/Celeron
- AMD K7/K75/Athlon/Duron/Athlon 4/Athlon MP/Athlon XP (tested!)
- Intel Pentium 4/Celeron-478 (tested!)

"Support" means not only names of CPUs (which shows with kernel name and
can be change) but correct init of L2-cache, FSB, Multiplyer, support of
different steppings. There are many thing that patcher makes, and all of
them that the manufacturer didn't make for correct support of CPU.
______________________________________

The patched BIOS is able to identify the CPU, but goes no further :-(

Well, I found something else that doesn't necessarily resolve anything
but is, nonetheless, rather interesting. This guy's Dell BX motherboard

http://www.psychowire.com/mainboards/slocket.html

reported his new P-III 700 on a slotket as a "Pentium Pro 500."

Apparently there was something in that BIOS that had, even if nothing
else, at least the string "Pentium Pro" in it.

Maybe you should be looking for the oldest BIOS you can find for those
boards in the hopes they might have 'Pro' remnants left.

Good suggestion.

The oldest BX BIOS I have archived is 1006 for a P2B-DS (09/15/98), but
the results are identical - freezes at the same point, patched or not :-(

Anyone got a truly *ancient* Asus BX BIOS they'd be willing to send me?

 

[P2B]

Wow, didn't know such adapters exist...

New one on me too - piqued my interest

Very little to them - a Socket 8, 6 filter capacitors and a dozen or so
surface mount caps, nothing else.

Is it possible the PPro might have some erratas which make it necessary
to have PCB/bios workarounds? I haven't looked at the spec update, but
such errors are not uncommon. And the workarounds might not be present
on the P2B.
And I hope the bios doesn't use some mmx code to initialize it...
Also, google shows some people got it to work in P2L97 boards, but not
with all bios versions (only the oldest one worked?), which probably
indicates it is a bios problem on the p2b too.

It's definitely starting to look that way - further research shows the
initial KN97-X BIOS didn't support PPro either, Asus released one to go
with the adapters.

Do you have any ancient Asus BX BIOSes? The oldest I have is 1006
(which, surprisingly, boots a P3-S but gets no further along with the PPro).

P2B

 

[David Maynard]

LX chipset... getting warmer

Yeah. That was part of the reason for mentioning it: Not 'strictly' FX.

Perhaps not... the 'particular rev' is the very first release for that
board, and he says later versions didn't work :-(

Yes, I know: 'old'.

FX chipset - same as the Asus board the adapter was made for.

Right. But both on the same board from a different manufacturer suggested
there's nothing particularly 'special' in the hardware for one vs the other
and that the Asus adapter shouldn't need an 'Asus' board. Not 'proof', but
a suggestion.

Sure I can... but it isn't a microcode problem - POST doesn't get
anywhere near that far,



My thinking too - I hoped Bios Patcher would help. Quoting from the manual:

______________________________________
=> Patcher can add support of CPU:

- AMD K6/K6-2/K6-III/K6-2+/K6-III+
- Intel Pentium Pro/Pentium II/Pentium III/Celeron
- AMD K7/K75/Athlon/Duron/Athlon 4/Athlon MP/Athlon XP (tested!)
- Intel Pentium 4/Celeron-478 (tested!)

"Support" means not only names of CPUs (which shows with kernel name and
can be change) but correct init of L2-cache, FSB, Multiplyer, support of
different steppings. There are many thing that patcher makes, and all of
them that the manufacturer didn't make for correct support of CPU.
______________________________________

The patched BIOS is able to identify the CPU, but goes no further :-(

That does seem like it should work, unless it didn't go in right for some
reason.

Ya know, after all that talking I did about being hardware similar, I
suppose it's possible there's some 'magic' pin/power/signal somewhere, like
FC-PGA vs PPGA, that just didn't get put on later boards. I wonder if
scouring old Intel Slot-1 docs would uncover some rev level change in the
slot=1 wiring from model #1 to the later ones.

Come to think of it, are your BX boards able to supply the Pro's 3.3v Vcore?

 

[~misfit~]

BX can do better than that The board was running a Tualatin P3-S @
1575Mhz (150Mhz FSB) before I pulled it to try the PPro.

I agree. I have a Tui Celeron running at 1.6Ghz on a BX here.

I "piss about" for interest's sake. I've explored the high end of BX
capabilities, now curious about the low end...

Me too. These slockets you've found fascinate me. I've always liked the P
Pro, especially the 1MB, 200Mhz version. The first x86 CPU with on-die,
full-speed ECC cache wasn't it?

Interesting project, keep us informed if you feel so inclined. I'd like to
know how you go.

 

[Rob Stow]

Me too. These slockets you've found fascinate me. I've always liked the P
Pro, especially the 1MB, 200Mhz version. The first x86 CPU with on-die,
full-speed ECC cache wasn't it?

I built and maintained a bunch of PPro workstations
way back in those olde days of yore. Bleeding edge
x86 workstations by the standards of those days

My recollection is that the PPro L2 cache ran at half speed.

My recollection is also that the PPro L2 was not "on-die".
The cpu core and the L2 were separate chips that were
put side by side into a single cpu package.

I would be happy to concede that my ancient
memories are wrong if someone could provide a link
to a reputable source.

 

[Roland Scheidegger]

It's definitely starting to look that way - further research shows the
initial KN97-X BIOS didn't support PPro either, Asus released one to go
with the adapters.

Do you have any ancient Asus BX BIOSes? The oldest I have is 1006
(which, surprisingly, boots a P3-S but gets no further along with the
PPro).

Unfortunately, no. I doubt though it would help, since the p2b boards
are quite a bit newer than the p2L97 boards it seems unlikely asus used
the same code which permits the oldest p2l97 bios to run the ppro in the
initial p2b bios.

Roland

 

[P2B]

I built and maintained a bunch of PPro workstations
way back in those olde days of yore. Bleeding edge
x86 workstations by the standards of those days

My recollection is that the PPro L2 cache ran at half speed.

My recollection is also that the PPro L2 was not "on-die".
The cpu core and the L2 were separate chips that were
put side by side into a single cpu package.

I would be happy to concede that my ancient
memories are wrong if someone could provide a link
to a reputable source.

Not sure if the L2 is on-die, but it definitely runs at full bus speed:

http://processorfinder.intel.com/scripts/details.asp?sSpec=SL259

 

[P2B]

Unfortunately, no. I doubt though it would help, since the p2b boards
are quite a bit newer than the p2L97 boards it seems unlikely asus used
the same code which permits the oldest p2l97 bios to run the ppro in the
initial p2b bios.

Sigh... I expect you are correct (pessimist!), but I (optimistically)
think it's worth a try *if* I can find a very early BX bios.

P2B

 

[P2B]

Yeah. That was part of the reason for mentioning it: Not 'strictly' FX.



Yes, I know: 'old'.



Right. But both on the same board from a different manufacturer
suggested there's nothing particularly 'special' in the hardware for one
vs the other and that the Asus adapter shouldn't need an 'Asus' board.
Not 'proof', but a suggestion.

Agreed - all available information so far points back to BIOS as the
issue :-(

That does seem like it should work, unless it didn't go in right for
some reason.

Ya know, after all that talking I did about being hardware similar, I
suppose it's possible there's some 'magic' pin/power/signal somewhere,
like FC-PGA vs PPGA, that just didn't get put on later boards. I wonder
if scouring old Intel Slot-1 docs would uncover some rev level change in
the slot=1 wiring from model #1 to the later ones.

Come to think of it, are your BX boards able to supply the Pro's 3.3v
Vcore?

Not a problem, they support 1.3v -> 3.5v

 

[Stephan Grossklass]

My recollection is that the PPro L2 cache ran at half speed.

Nope, full speed. That's why it was so expen$ive and Intel preferred to
use half-speed SRAM chips for the PII. Everything on the same die was
first to be found with the 2nd gen Mobile PIIs (.25µ, 256K L2) of which
the well-known Mendocino Celerons apparently were a scaled-down version
(same CPUID at least).

My recollection is also that the PPro L2 was not "on-die".
The cpu core and the L2 were separate chips that were
put side by side into a single cpu package.

AFAIK that's correct.

I've always wanted to have one of these BIG chips (just for looks),
maybe I'll run across one cheap some day.

Stephan

PS: My P2B-D now finally runs 2 667@500Es - that's what I call a longish
upgrade. Since I was unable to find any S370-DLs, I had to have my 6905
Masters modified, then the 2nd proc took a while to arrive, but now it's
running fine. I had to increase the VCore from 1.30 to 1.35 V, otherwise
performance was too low with both procs loaded. (I guess ye olde cA2
CuMine is pretty much at its limits when run at 500 MHz with 1.30 V
core.) Hey, it's still plenty cool enough...

 

[David Maynard]

I agree. I have a Tui Celeron running at 1.6Ghz on a BX here.




Me too. These slockets you've found fascinate me. I've always liked the P
Pro, especially the 1MB, 200Mhz version. The first x86 CPU with on-die,
full-speed ECC cache wasn't it?

It's a second chip on the same chip carrier; not on CPU die. It does run
full speed.

 

[David Maynard]

Not a problem, they support 1.3v -> 3.5v

Interesting, because according to the P-II data sheets, while codes up to
3.5 are defined, only up to 2.8 is 'required' for P-II motherboards.

MIght be worth breaking out the DVM just to make sure it's really doing it.