Pentium Pro s-specs

[gg1978]

This is referring to PPro s-specs

SL22L

SL23J

SL23K

So were these only sold outside the USA or do they even exist? I've looked for pictures of these, as i assume they're all boxed "BP" CPU's, and i've never seen a picture of any of them..

The funny thing is, that they're not listed in Intel's PPro s-spec list, which even includes the Q spec ES chips.

http://developer.intel.com/design/archives/processors/pro/qit/index.htm

[JAC]

Some q-specs don't exist on that guide. ... Unless they are fakes. Ask neon_wa about those above???

[Neon_WA]

best one to ask is Genna (gshv) as he keeps a record of where sSpecs were derived from

They are not listed in the documentation I have, but there was 35 revisions of the Pentium Pro sSpec update
and at least 10 of them included an updated sSpec table

All I can say is if they exist then they will have the ICOMP2 marking seeing what stepping they are listed as

[Robev]

[gshv]

Unfortunately, I don't keep a track where each S-spec code comes from, but I sometimes I can tell the source based on whether the S-spec has certain information or not. SL22L, SL23J and SL23K have core stepping and the maximum temperature, so the only source for these is Intel. I probably took them from older version of qit pages, or from processorfinder, or from some version of Pentium Pro specification update.

Gennadiy

[Neon_WA]

sB1 stepping was manufactured from week 45 1996

so earliest sSpec update I have is Rev 19 from June 1997 which doesnt include them

ideally would be good to find
014 - Dec 1996
015 - Jan 1997
016 - Feb 1997

the table in 017 & 018 should be same as in 019

[gg1978]

Are those available for download anywhere? As i'm sure the old versions aren't on Intel's page anymore. Although the one i linked does have both the 1M sspecs, as well as the SL2FJ and SL22Z's, which are also 0619 sB1 stepping. In fact all of the SLxxx are 0619's, except the SL245 and SL247 ones.. Tihe 1M's were the ones that should be the latest chip, though they were the same stepping, just with more cache.

[Neon_WA]

you just need to do a bit off searching for earlier ones.. as most online are Rev 35

Title is "Pentium Pro Processor Specification Update"
Document no is 242689
so 1st rev will be 242689-001 (35th will be 242689-035)

but typically the file will be saved as such
242689-001 >> 24268901.pdf
242689-017 >> 24268917.pdf
242689-035 >> 24268935.pdf

[gg1978]

well i've found -29 and -32 so far. Though in the document it does show how it was updated for each rev.. Whether they just added errata or updated specs for given s-specs, or added new ones.

edit: just found -019, -021 and -027.. -019 is cool because it shows more detailed info on the assorted ES's..

[JAC]

Are there any documented cases of Intel producing a CPU in that era but not actually releasing it? If so, that would suggest this has happened here. I am not talking about FDIV stuff, we all know about that.

[gg1978]

I wouldn't rate that as likely, but i guess it's possible.. The thing is that if they did such a thing, it would have been expensive for them, as the PPro CPGA package can't be remarked easily, at least without removing some of the ceramic itself to obliterate the markings. It might even imply that if they chose not to release them, they'd all be destroyed because of that..

Given that these are production s-specs that would tend to imply a production quantity of CPU's made, if they were made..

I can understand some PPro ES's being unobtainium, or damn close because they were low quantity to begin with, and not officially release for public use.. I mean the Q0815 is the closest thing to a common PPro ES, with all others being rare or nonexistant..

But i can't see that for production s-specs, if they were actually made in production quantities, as i've never even seen a picture of any of these sspecs.

I mean maybe they were released in another country, but it doesn't make much sense as they could just make them under the regular s-specs for production cpus, like SL23L or SL23M's..

[Neon_WA]

typos are common in Intel documentation

but in this case these particular sSpecs may have been listed for a OEM that end up ordering one of the normally listed sSpecs

You can see this in Pentiums.. many of the hard to find sSpecs were only made for one particular OEM.

Back to Pros.. all of the found SL2FJ have been from the same FPO
so likely only a maximum of 10000 was made.. could be less
out of these.. only 5 have been found

[JAC]

These extra sSpecs don't exist in later documentation, correct?

[Neon_WA]

[JAC]

There has to be a reason for that. I wonder how/where the answer can be found. Guessing is subjective