More About That Strange Pentium 4

A few years ago I wrote about a strange NetBurst processor with SL7HY S-spec that landed at the OS/2 Museum. After renewed reader interest I pulled it out of the closet and tested the processor again. A collection of miscellaneous notes follows.

CPU-Z today is just as clueless as it was a few years ago (no surprise), thinking it’s a Socket 604 processor and that it’s an engineering sample:

CPU-Z misidentifying the SL7HY processor

The SL7HY processor was briefly tested in an ASUS P5PE-VM board. The BIOS complains at boot-up that it has no microcode for the processor, and clearly shows the brand string as Intel(R) Xeon(TM) CPU 3.73GHz. That gives me confidence that this product string is really what’s burned into the processor.

The P5PE-VM does not support 1066 MHz FSB, but an Intel D975XBX2 (Bad Axe 2) does. In the D975XBX2 board, the SL7HY runs at the rated 3.73 GHz. It requires a sufficiently beefy cooler, but for example the standard Intel LGA775 cooler suitable for Core 2 Extreme processors does a good job.

A kind reader dug up a microcode update for the CPU. Note that the processor signature is 0F42h, which is a signature that is not included in Intel-supplied microcode update packages at all. But the recovered microcode update was tested with Linux and successfully applied; after update, the reported microcode revision changed to 3, clearly indicating that the microcode really was updated. Beyond that, it’s not at all obvious what the updated microcode does. The microcode is from April 21, 2005.

Considering the fact that the SL7HY processor is from late 2004 yielded several thoughts about the timeline and possible classification of the mystery CPU. First of all, it is a relatively early Socket 775 CPU (Socket 775 appeared in 2004), and it is a very early 1066 MHz FSB CPU, available at about the same time as the first chipsets/boards supporting 1066 MHz FSB (late 2004).

The SL7HY must also be one of the earliest Intel processors with NX bit support. NX bit capability also makes it definitely not a Nocona, and it is in fact a Prescott. In general it appears to be a very close relative of the 3.73 GHz Pentium 4 Extreme Edition CPU; the processor features appear to be identical (64-bit, NX, HT, enhanced halt, thermal monitor) and the only significant difference is 1MB L2 cache on the SL7HY vs. 2MB on the 3.73 GHz P4 EE. The SL7HY is also from late 2004 whereas the Prescott P4 EE is from February 2005. It is also noteworthy that the SL7HY CPUID signature is (as mentioned above) 0F42h, while the Prescott EE is 0F43h, hinting at the close relationship.

Note that there was a 3.46 GHz P4 EE with 1066 MHz FSB released in late 2004, but that was not based on the 90nm Prescott core. When the SL7HY shipped, it was possibly the fastest Pentium 4 available, at least for some definition of “available”.

The mystery CPU is also a very close relative of Pentium 4 550 S-spec SL7PZ which was released in November 2004. The SL7PZ ran at 3.4 GHz with 800 MHz FSB but was otherwise the same. If the SL7PZ were changed to run at 1066 instead of 800 MHz FSB and the multiplier reduced from 17 to 14, we’d end up with SL7HY.

Is it real or is it a fake? A reader suggested that someone could have re-lasered the label on the CPU. While that is true, I have difficulty imagining how someone could also change the CPUID brand string to read “3.73 GHz”. In my opinion the CPU is not a fake. In fact the SL7HY S-spec is quite close to other Prescott CPUs from the same era (SL7Z4, SL7E6, SL7K9 etc.).

That still leaves some questions unanswered, and the true answers might never be known:

  • Why does the lid say Pentium 4 but the CPU brand string says Xeon? I really don’t know. It makes no sense.
  • Why is the S-spec not listed anywhere? Probably the SL7HY CPU is an OEM-only “off the books” model, never available through normal channels and therefore never fully documented.

Note that the OEM model theory is not at all outlandish. There are other known “OEM only” S-specs from the same era, for example SL7QB or SL7Q8, and there have been OEM-only Intel CPUs before and after.

Update: It is maddeningly difficult to establish the timeline of which features were available when with Prescott CPUs. The Prescott line-up was, to put it bluntly, a hot mess. There were 32-bit CPUs, there were 64-bit CPUs, there were 32-bit CPUs with NX, 64-bit CPUs with NX, all with a mix of other random features like SpeedStep, enhanced halt states, thermal monitor, etc.

It is well documented what the S-specs were. It’s less well documented which S-spec was sold under what name, and it is even less well documented when a given S-spec became available. Intel’s ark is a mess and does not for example list the model “F” Prescotts which were the first to support AMD64 aka EM64T. Wikipedia does list those, with dead links for reference.

This problem is, of course, not exactly new. Back in 2004-5, people complained how hard it was to figure out which Prescott was what. According to this article, things were so bad that some Pentium 4 CPUs sold under the 540/550/560 processor number designation supported EM64T and others did not. Review articles showed readers how to figure out what a given box contains.

It is virtually certain that Prescotts with 1MB cache and all of HT, EM64T, and NX (same as the mysterious SL7HY) were available in late 2004. The Prescott specification updates (Intel document 302352-xxx) from 2004-2005 would provide good hints what happened when, but they are unfortunately not available… or at least weren’t until recently.

Late Update: Digging through the Prescott specification updates, there are several noteworthy data points:

Spec update 302352-006 is dated September 24, 2004 and it is the first to mention E0-step (0F41h) Prescott models, some of which had EM64T. But that’s not all–some D0-stepping Prescotts (SL7L9/SL7L8/SL7LA) had EM64T too! It is not clear from that spec update which processors might have supported the NX bit, but there are several errata related to NX.

Spec update 302352-008 is from November 2004 and it is the first to list the 3.8 GHz E0-step Prescott.

Spec update 302352-013 is from February 22, 2005 and it is the first to list N0-step (0F43h) Prescott-2M models, including the 3.73 GHz Prescott P4 EE. This spec update also clearly shows which models include C1E (enhanced halt states), TM2 (Thermal Monitor 2), NX, and Enhanced SpeedStep. With the exception of Enhanced SpeedStep, these features are not new in the N0 stepping and existed in some E0 models already.

Spec update 302352-022 is from October 2005 and it is the first to list G1-step (0F49h) Prescotts with 1M cache. Oddity: Errata lists show separate entries for Socket 478 and LGA775 G1-step processors, but no G1-step Socket 478 S-spec is listed.

The next spec update (302352-023) is from November 14, 2005 and it is the first to list R0-step (0F4Ah) Prescott-2Ms with VT-x support (3.6 and 3.8 GHz models).

Spec update 302352-026 from February 2006 lists several new R0-step models without VT-x (e.g. SL8PZ, SL8PY). Those appear to be the final Prescotts, with Smithfield and Presler CPUs now holding the line until the arrival of Core 2 processors.

Later Update: That Celeron D with S-spec SL8S4 is very, very fishy. There is no question that it was listed on Intel’s processorfinder site and claimed to have 0F42h CPUID and G0 stepping, and it’s still listed on ark. But it’s not listed in the Celeron D specification update, and I can’t find any evidence that anyone ever saw a S-spec SL8S4 Celeron. On the one hand, the fact that no photos or CPU-Z screenshots exist doesn’t prove that the S-spec never existed, on the other hand ark is known to list processors that never shipped (like the 4.0 GHz Prescott). Unless someone provably has or had a SL8S4 Celeron, its existence has to be treated as questionable.

The first Celeron D 3xx Sequence spec update which lists Celeron D 310 is Intel document number 302354-011 from August 17, 2005 and it only lists one S-spec (SL8RZ) with 0F41h signature and E0 stepping. Given that the SL8S4 S-spec was apparently listed on Intel’s web site in June 2005, that is quite odd. The situation is the opposite of the SL7HY which clearly exists but is not listed anywhere.

Even Later Update: Digging into the provenance of the 0F42h microcode update was worthwhile. It is included in the BIOS of ASUS P4P800-E Deluxe. The supported processor list for said board includes “Celeron D 310 revG0”, presumably the elusive SL8S4 S-spec, as being supported starting with BIOS version 1008. And sure enough, version 1008 from August 8, 2005 has the 0F42h microcode update, while version 1007 from May 25, 2005 does not.

Now, the P4P800-E is a Socket 478 board, and the SL8S4 Celeron is a Socket 478 CPU. So if that microcode update was intended for the SL8S4, it would actually make sense to have that in a S478 board. What’s more, the same BIOS also contains a microcode update for the G1-step (0F49h) Prescott, and both the 0F42h and 0F49h updates are dated April 21, 2005, and both are revision 3. That suggests the 0F42h and 0F49h processors are extremely close, and the G0/G1 stepping nomenclature makes sense.

What makes much less sense is that the same 0F42h microcode applies to the SL7HY Xeon which is significantly older and has rather different feature set (although that doesn’t mean it can’t be more or less the same silicon). None of this makes a lot of sense.

BIOS Update: After some fortuitous searches it’s now clear why CPU-Z calls the SL7HY processor a G0-stepping Nocona. Here MSI references a “G0 Stepping Nocona Xeon” in a BIOS update for a Socket 604 board. And here is a possible S604 Nocona engineering sample with QFFO S-spec but unknown CPUID signature. The best bit is here, claiming that a Dell Precision 670 (again Socket 604) BIOS included a Nocona G0-step microcode update for a processor with 0F42 CPUID signature. The Dell BIOS appears to store the microcode updates in a custom format, with a header which contains the CPUID signature and a four-letter identifier. For the 0F42 CPUID, the string is 0GCN, or NCG0 when reversed, which almost certainly means Nocona G0 stepping. There are similar strings like NCE0 (Nocona E0), PXA0 (Paxville A0), or IRN0 (Irwindale N0). What this implies is that the ‘Nocona G0’ designation for the 0F42 CPUID almost certainly came from Intel.

Curiously, Wikipedia claims the existence of a released G0-step Nocona Xeon with SL8RW S-spec, clearly based on data from Intel’s ark. But CPU-World disagrees and claims that the SL8RW has 0F49 CPUID signature, making it the G1 stepping. Intel’s actual Specification Update (302402-024) agrees with CPU-World (i.e. SL8RW has a 0F49 signature and G1 stepping), contradicting Intel’s own ark. Now who do you believe, Intel or Intel?

This entry was posted in Intel, PC hardware, PC history, Pentium 4, Undocumented. Bookmark the permalink.

32 Responses to More About That Strange Pentium 4

  1. Richard Wells says:

    Without the provenance of the chip, it will be a challenge to trace.

    However, there were a few announcements in the trades that might be suggestive. The Register had an article about the 1066 MHz FSB and the 3.46 GHz Extreme Edition chip that went with it and a proposed 3.73 GHz Pentium 720 EE which never shipped (nor did any Pentium 4s with 700 series model numbers). That was June 2004. So any Intel roadmap from around that time would show the 1066 FSB plans and possible where the mystery chip fits in.

    My suspicion is that these chips went to one of the retailers that had announced Pentium 4s overclocked to 4 GHz in late 2004. A few thousand halo chips for a boutique vendor to maintain Intel’s performance gaming crown.

  2. Michal Necasek says:

    The thing is that what Intel was promising was a 3.73 GHz Pentium EE with 2MB cache, and that is what they eventually delivered in Feb ’05. But as a stopgap measure the SL7HY chip with 1MB cache available in late 2004 would make sense.

  3. Richard Wells says:

    Intel also promised that the 500 series Pentium 4 would have been moved to the 1066 FSB at the same time as the chipset and Extreme Edition 1066 FSB were announced. I have no way of determining if the chip is 1066 FSB Nocona, Prescott EE with failed cache turned off, or Prescott EE which needed half the cache turned off for thermal reasons not because the cache had errors. The first two options suggest the possibility of other unknown Pentium 4 variants like slower 1066 FSB Noconas or full cache Prescott EE chips reaching 3.73 GHz in 2004.

  4. Michal Necasek says:

    The SL7HY chip is definitely not a Nocona because it has features (like NX bit) that Nocona never had. It’s some kind of a Prescott, but it’s really unclear whether it’s closer to a P4 EE with half the cache or to an earlier 1M Prescott with overclocked FSB.

    I’m now also wondering if it’s also the earliest Xeon-branded CPU for a mainstream desktop socket. In the times of Core 2 and later Intel always released a few Xeon models for the standard socket, but I don’t believe that was the case with any Pentium 4/Pentium D (i.e. Socket 423, 478 or 775).

    Funny thing: Intel’s ark claims that the Prescott P4 EE was released in Q4 2004. Everyone else says it was Feb 2005, and contemporary reviews confirm the Feb ’05 date.

  5. Richard Wells says:

    Nocona upgraded to support NX bit was announced. https://www.theregister.co.uk/2004/07/30/intel_nocona_nx/ That was only slightly after the chip in question was manufactured. 1066 FSB chipsets were announced slightly earlier. Code names and roadmaps were in a bit of flux in the confusion following the cancellation of Tejas.

    The planned server 1066 FSB chipset for 2004/2005 seems to been canceled leaving the chip orphaned. The benchmarks for the 1066 EE chips were not kind and the non-gaming performance was not improved over chips on a slower bus. Only 1 MB of cache should have hurt server benchmarks even more.

    Remember the difference between launch date and availability. Intel, for any highly anticipated but scarce chip, needs several months of production to have sufficient inventory to meet initial demand. Reviews tended to be done after chip is on sale. Intel ARK generally refers only to launch date.

  6. Michal Necasek says:

    Yes, I know that launch and availability dates can be several months off. Doesn’t make things easier.

    Intel was more than a bit lost in 2004, Prescott was just slightly faster than Northwood and Tejas went nowhere. Canceling the 4.0 GHz Prescott was probably the clearest sign that Intel gave up on the clock speed game.

    Another bit of fun was that the first 1066 FSB processor was the 3.46 P4 EE, but that was based on the older Northwood core. And when the Prescott based 3.73 GHz P4 EE was released, the 3.8 GHz Prescott on 800 MHz FSB was already available. With Pentium D Intel got cleverer and only the EE variants had hyper-threading.

    There’s of course not much difference between Nocona and Prescott, basically Nocona is the Socket 604 variant of Prescott (and Irwindale is Prescott-2M). Much like the next microarchitecture was called Merom but the families were Merom, Conroe, Kentsfield, Woodcrest, Tigerton, etc. But you’re absolutely right that there were Nocona CPUs with a very similar or identical feature matrix to the SL7HY, and they were available in 2004.

  7. Zir Blazer says:

    Here I come to drop several hypothesis:

    The mystery of the 0F42h CPUID can be easily explained since it is supposed to be identifying a unique Prescott/Nocona Stepping, which is G0:
    C0 – 0F33h (Prescott)
    D0 – 0F34h (Prescott)
    E0 – 0F41h (Prescott)
    G0 – 0F42h (Prescott?) <— We are here
    N0 – 0F43h (Prescott 2M)
    A0 – 0F44h (Smithfield)
    B0 – 0F47h (Smithfield)
    G1 – 0F49h (Prescott)
    R0 – 0F4Ah (Prescott 2M)

    All the Prescott and Prescott 2M ones but the G0 0F42h appear in the Celeron D, Pentium 4 and Xeon 90nm lines Specification Updates:
    Celeron D 90nm (Order 302354): https://www.intel.com/content/dam/support/us/en/documents/processors/celeron/sb/30235421.pdf
    Pentium 4 90nm (Order 302352): https://www.intel.com/Assets/PDF/specupdate/302352.pdf
    64 Bits Xeon (Order 302402): https://www.intel.com/content/dam/www/public/us/en/documents/specification-updates/xeon-with-800-mhz-system-bus-specification-update.pdf

    Note that officially there was no Celeron D that used a Prescott 2M die with 7/8 Cache L2 disabled (Maybe a heavily die harvested OEM only part?), so it would be weird for the G0 based Celeron D 310 SL8S4 to be a Prescott 2M, as that would make it even more unique. Can the size of the die be determined via x-ray, infrared or something to confirm whenever it fits a Prescott or Prescott 2M size?
    Is possible that the CPUID Model / Stepping was assigned in a first come first served basis, which is why the 0F42h mystery Prescott and 0F43h Prescott 2M are so closely together, then followed by the 0F44h Smithfield. Seems to match a sort of time frame. If anything, the G0 was shortly lived and quickly replaced, which is why there are so few products based on it.

    Note that I think that you're doing an huge mistake by associating codenames with physical dies. CPU-Z and other tools likes to display codenames based on heuristics from available specifications, so I wouldn't really take it seriously if it thinks that it is a Nocona based on an undocumented Model / Stepping.
    Even more important than that, as far that I know, P4 Prescott and Xeon Nocona are the same die (Assuming same Revision), as does P4 Prescott 2M and Xeon Irwindale. Smithfield was two Prescott dies sawed from the wafer together, so there should be a correlation between Prescott revisions like G0/G1 and Smithfield A0/B0 ones. It could be possible that there was a Prescott revision specific for Smithfield, too, and G0 would actually fit there…

    I also wouldn't try to make Processor relationships based on specifications when they're based on different die revisions. As far that I recall, Prescott generation was EXTREMELY convoluted, with a whole bunch of models that had or hadn't a specfic feature, like Hyper Threading, EM64T, NX/XD Bit, and VT-x/Vanderpool. It may be possible that specific features were completely broken on earlier revisions. If anything, I think than this is the only Prescott with 1066 MHz Bus, compared to the other one, the famous P4 EE 3.73 GHz that was a Prescott 2M. Actually, a list about what features were found enabled on a specific die revision would be fun to do, as that confirm that they worked at that point. Did G0 had something enabled than no E0 didn't?

    If anything, there is even a bigger question than the CPUID: Why the hell it has a Xeon string? Giving than it is OEM only, it MAY be possible than it was actually sold as a LGA 775 Xeon. Does anyone knows what system was the SL7HY pulled from? I mean, the S478 P4 with EM64T that you posted are now known to be pulled from already identified specific IBM systems, and they were properly advertised as supporting 64 Bits. If you can identify the source of this mysterious Processor and find than that system was supposed to have a Xeon, then there is your answer about what it is. After all, that the Heatspreader is wrongly marked as a P4 doesn't matter if end users weren't supposed to dissamble the system at all. Software identification tools would make the user happy if it reported as a Xeon.

    Randomly googling around I also found this:
    https://pttdigit.com/hardware/M.1134834917.A.4DF.html
    Mentions 0F42h G0 as a LGA 775 Xeon.
    Should be related to QDZA / JM80547KH1091M, which is a LGA 775 that also has a Xeon string: http://www.cpu-world.com/forum/viewtopic.php?t=15723

  8. Michal Necasek says:

    Moving my comment here, and deleting the comments on the Zen thread (I completely missed that they were “misfiled”). FYI it’s the URLs that likely trigger moderation.

    I don’t see the need to posit the existence of an unknown G0 stepping because I can’t see anything in the SL7HY that an E0 stepping Prescott couldn’t do. But this is Intel, so who knows.

    Unfortunately I have no information about where the SL7HY came from. I couldn’t find mentions of such a CPU back in the day, but 3.73 GHz, 1066 MHz FSB, or ‘Xeon’ aren’t very unique, and exact S-specs or CPUID signatures are usually not found in reviews or advertising.

    The CPUID signatures are assigned freely but not randomly. BIOSes (certainly P4 era BIOSes) use the CPUID signature as a rough guide to which features to look for/enable, and what errata there might be to work around. The assumption is that higher CPUID signatures imply a superset of features within the same family. The 0F42h signature might be a separate stepping, or it might be the same thing as 0F41h but deliberately separate for easier identification. Why an obscure Celeron would have the same CPUID signature is a real mystery, and it’s IMHO not impossible that the signature was reused precisely because the SL7HY was an off-the-books model.

    I’m not assuming that the family names are significant, it’s all Prescott. CPU-Z clearly guesses that ‘Xeon’ in the brand string must mean Socket 604 Nocona, but the socket is definitely misidentified so there’s no need to take the family or stepping identification too seriously. I agree that the brand string is probably more significant than what the lid says, because that’s what users would see.

    The QDZA definitely looks very, very similar, though the fact that it’s apparently newer than the SL7HY is quite strange.

  9. Zir Blazer says:

    The Prescott G0 Stepping is NOT unknow, it is actually acknowledge by Intel as it appeared on Intel Ark predecessor. Some people mentioned it back during 2005:
    https://overclockers.ru/hardnews/show/19257/Celeron_D_310_2.13_GGc_ocherednoj_prizrak_na_steppinge_G0
    https://overclockers.ru/lab/show/19805/Intel_Celeron_D_310_2.13GHz_pochti_idealnyj_processor_dlya_apgrejda_i_razgona

    As far that I know, the CPUID Model / Stepping does not work as a sort of superset from a previous one nor anything like that, it just serves to identify a specific die (But can be used any way that Intel/AMD sees fit, so it is not an absolute rule). This matches how Microcode works, as Intel distributed multiple Microcode versions, each targeting only a specific Model / Stepping.
    Here is an interesing document that is available in Google Cache that covers that: http://webcache.googleusercontent.com/search?q=cache:1K6mafy1dvQJ:ftp://ftp.foundertech.com/server/oldserver/mr100-1100/CPU/Nocona_NDA0204.pdf
    It mentions Nocona, some ES SSpecs and Stepping errata. Since Nocona and Prescott are one and the same, it is reelevant.

    0F30h – Prescott/Nocona A0
    0F31h – Prescott/Nocona B0
    0F32h – Prescott/Nocona B1
    0F37h – Prescott/Nocona C1 (Note that this is out of place cause it is a higher number than 0F33h C0 and 0F34h D0. Go figure…)
    0F34h – Prescott/Nocona D0

    All five Steppings uses different Microcode, as covered in Page 17. Your G0 is unique in that regard, too, as it uses different Microcode than the other ones. I would believe that it is not because Intel was lazy and used that to identify an unique SKU, to me it is actually a physically different die.

    Bonus: Xeon MP 90nm Specification Update (Order 306752) https://www.intel.com/content/dam/support/us/en/documents/processors/xeon/sb/306752_013.pdf
    This is one correspond to the codename Cranford Xeon MP. Intel says that its Steppings are A0 and B0, but their CPUID is 0F41h (Prescott E0) and 0F49h (Prescott G1), respectively…

    If we go with the unique G0 die hypothesis, then the Celeron D is completely logical, too. As for some reason it seems to have had an extremely short run, Intel seems to have decided to completely simplify G0 binning and just go for two models, the 3.73 GHz part, which was a very high end binning based on the fact that the 3.8 GHz Prescotts were always rare, and the slower possible Celeron D 310. Whatever didn’t passed binned or had damaged Cache L2, got sold as the Celeron, and having no intermediated bins simplified the logistics chain for a stopgap product.
    Also, do we know where those G0 Celerons came from, either? Giving the fact that it seems that it wasn’t common for standard BIOSes to have Microcode targetting 0F42h, chances are that they were also found in specific OEM systems, too…

  10. Michal Necasek says:

    There’s one small problem with the articles about G0 stepping Celerons — they’re from late June and August 2005, respectively. The SL7HY is some 8-9 months older than that. However it is good evidence that at least with the Celerons, the 0F42h CPUID meant G0 stepping. Microcode updates for the 0F42h Celeron seem to be as rare as for the 0F42h Xeon!

    Within a single model/family, the stepping normally reflects progression over time. Since features tend to be added much more than removed, the newer steppings are typically supersets of older ones. Of course there are no absolute rules, in the end Intel can assign the stepping numbers and identifiers however they see fit. At the same time they don’t want to complicate the BIOS writers’ lives too much, so Intel does not go too wild.

    I guess the fact that 0F41h/0F49h is Prescott E0/G1 or Cranford A0/B0 tells you that the steppings are just labels, because there’s no reason to think that the dies with the same CPUID weren’t more or less identical.

    Microcode updates are very very specific and a given microcode update is always only for a single CPUID signature, and even then it may apply only to a subset of the processors with that signature. That’s a consequence of the microcode update format.

    It occurred to me that if someone found an OEM BIOS with 0F42h microcode for the Xeon, that would be a pretty good hint what boards it was intended for. All I know that Intel’s own D975XBX2 does not have the microcode, even though it otherwise works with the SL7HY just fine (and it runs at the rated 3.73 GHz). The 0F42h microcode update rev 3 was actually found in a beta BIOS for an ASUS Socket 478 board.

    Here’s something interesting: https://www.intel.com/content/dam/support/us/en/documents/motherboards/server/s5000vsa/sb/d36978010_s5000vsa_tps_r1_9.pdf See the bit about Nocona-T/Dempsey-T vs. Dempsey-J jumper. I would guess that J means socket J aka LGA771, and T means socket T aka LGA775. Which means that Nocona-T would be a LGA775 Xeon. What that jumper does on a board which has two LGA771 sockets I don’t know. This is the only reference to Nocona-T (or Dempsey-T) that I can find.

  11. Richard Wells says:

    There is one OEM BIOS that supported the necessary microcode; it was the BIOS in the system the chip was extracted from. Intel isn’t in the practice of giving away working $1,000 chips.

  12. Zir Blazer says:

    I just remembered that the Win-Raid BIOS mod community has a Microcode repository project. Check here: https://github.com/platomav/CPUMicrocodes/tree/master/Intel

    cpu00F42_plat9D_ver00000003_2005-04-21_PRD_7A20982B.bin

    File has been uploaded two years ago. I wonder if they hold a log of the source from the less common ones, but given than the one you used came from a S478 beta BIOS…

  13. Michal Necasek says:

    Thanks for the link. Not too surprisingly, that microcode update is identical to the one extracted from the ASUS beta bios.

  14. Michal Necasek says:

    And actually it was in a non-beta BIOS for the P4P800-E as well. I updated the post accordingly, this whole thing is just weird.

  15. Michal Necasek says:

    There is a mention of fixes for “Nocona-T” in AMI BIOS release notes (August 2005). There’s also this: “According to Prescott, Nocona, Potomac BWG 0.9, the 267MHz system bus support has been added on Nocona processor or Northwood processors in the LGA775 package.” BWG is the BIOS Writer’s Guide, but the sentence is phrased poorly enough that I’m not sure if it refers to an LGA775 Nocona or not. The LGA775 Northwood with 267MHZ bus was of course the 3.46GHz P4 EE with 1066MHz FSB. There’s also a note that mentions removing Nocona-T microcode in early 2006, implying that Nocona-T never shipped (surprise surprise).

  16. Zir Blazer says:

    I failed to follow you up due to the fact that I couldn’t find the AMI document you mentioned. Related to that, some Intel documents make a reference to “Prescott, Nocona, and Potomac Processor BIOS Writer’s Guide” like the AMI one, but seems to be a NDA only document cause it lacks an Order Number and couldn’t find it googling, either.

    The Nocona-T thing is intriguing. LGA 771 and 775 are physically extremely close, there was even a mod discovered not so many years ago that allowed LGA 771 Xeons to run in LGA 775 Motherboards. I’m sure you’re aware of it: https://www.delidded.com/lga-771-to-775-adapter/#how-adapter-works
    However, since the “T” is supposed to mean LGA 775, the question is why does a LGA 771 Motherboard mentions such Processor at all? Could it be that LGA 775 Processors are physically insertable in LGA 771 Motherboards out of the box? Intel had a similar situation where earlier Socket 603 Processors could be inserted in Socket 604 Motherboards, but not the other way around, so it is not necessarily new.
    What I believe is that the T series were for UP (Uniprocessor) and usable in either LGA 775 or 771 Motherboards. I’m not sure whenever all S604/LGA 771 Xeons of that era were DP or there were UP models only, like Intel did later. Is the only thing that makes sense to me, assuming that the package was physically compatible on the first place.
    I’m curious about what the Processor Select Jumper in the Intel S5000VSA is supposed to do at a physical or Firmware level…

  17. Michal Necasek says:

    The AMI BIOS isn’t public, so not that easy to find. The Prescott etc. BWG is something I’ve never seen myself, only references to it. Would be great to have one day.

    Yes, I’ve seen the tiny adapters that turn a LGA771 Xeon into a LGA775 processor. As far as I can tell, the LGA775 vs. LGA771 distinction is largely artificial, allowing Intel to charge more for dual-socket CPUs. But what the Nocona-T vs. Nocona-J switch does in the S5000VSA board I really don’t know, I can’t make sense of it.

    I don’t know if there were any exceptions but the intended usage was certainly LGA775 for single-socket systems, LGA771 for dual-socket systems, and Socket 604 for four-socket systems, though there were also dual-socket S604 boards (quite common actually) due to the fact that the S604 CPUs were in some ways better than the LGA771 ones. Certainly with any given generation the selection of LGA771 and the corresponding Socket 604 CPUs was different, sometimes vastly different.

    I actually don’t quite get what the point of Nocona-T would have been, or rather how it was supposed to be different from a LGA775 Prescott. Then again Conroe, Wolfdale, etc. Socket 775 Xeons were a thing, and it’s entirely plausible that Nocona-T was the same idea a year or two earlier.

  18. Chris M. says:

    The computer I’m typing this on has a 45nm Wolfdale Xeon E3110. Other then the CPUID results, its just a rebranded Core2Duo E8400. I landed up getting it because it was cheaper than an actual E8400 and it was in stock (there were issues sourcing 45nm chips at the time).

  19. Richard Wells says:

    Socket 771 supported higher TDP than either Socket 775 or Socket 604. I am not sure why the socket TDPs were so much lower than the expected TDPs for Tejas and Jayhawk.

    Xeon 775 gave the workstation market the ECC support plus good I/O that had previously been cobbled together out of modified high end non-server motherboards with server sockets.

    Nocona was a poor choice for most purchasers and trying to force it into the role of a cheap workstation chip was never going to work.

  20. Michal Necasek says:

    The price differences nowadays are interesting, and the difference for a Core 2 Quad Extreme vs. the corresponding LGA771 Xeon is ridiculous (like $200 for C2Q EE vs. $30 for the equivalent Harpertown Xeon). I imagine it’s because the Xeons sold in much larger numbers because corporations don’t mind getting ripped off as much as individuals do.

    I still regularly use old Core 2 laptops and although they’re getting long in the tooth, they’re surprisingly usable 10+ years later. I also have one board with Core 2 QX9770 and that still feels pretty snappy.

  21. Reddington says:

    Interesting find!

    “I still regularly use old Core 2 laptops and although they’re getting long in the tooth, they’re surprisingly usable 10+ years later. I also have one board with Core 2 QX9770 and that still feels pretty snappy.”

    My desktop PC uses an LGA775 Xeon X3380 and its still fast enough for me, admittedly its cranked up to 4GHz on an Asus Maximus II board (going strong for maybe 10 years now). This CPU was an amazing deal, paid a fraction of the price compared to retail cost of Q9650.

  22. Chris M. says:

    I have a laptop with a Core2Duo T5200 with 4GB of RAM still in service. With a $16 SSD, its no slower than many of those Atom machines that are being sold new. In many cases its actually faster! They also have more I/O than modern laptops. You get ExpressCard, Firewire, and multiple video output options. It also has Ethernet and a 56k modem (how quaint)

    Regarding ECC and LGA775 “E” series Xeons, they really weren’t different from their Core2Duo cousins. The board manufacturer was the one who enabled ECC support in the BIOS. My Gigabyte X38 board should have no problem supporting ECC RAM with Core2Duo CPUs, but Gigabyte disabled it in the BIOS. I haven’t done the LGA771 modded route since this board (GA-EX38-DS4) was known to be troublesome with those CPUs.

  23. Michal Necasek says:

    I have a GA-X38-DS4 board too. Seems like a nice board, optionally overclocked to support 1600 MHz FSB processors.

    Sometimes I use a ThinkPad W500, it can do ExpressCard, CardBus (probably PCMCIA too), USB 2.0, FireWire, has an Ethernet port… definitely more flexible than modern laptops. I have for example old audio interfaces that use CardBus or controllers or FireWire, and an old audio interface can be worth as much as a new laptop and a new audio interface would probably be only infinitesimally better. It’s much easier to just use the old laptop.

    The Intel D975XDX2 board supports ECC and the documentation does not say anything about ECC only working with specific CPU models. The memory controller is in the chipset (MCH) so I don’t see any technical reason why ECC should depend on the CPU. With the Nehalem generation CPUs and all 64-bit AMDs that’s of course different.

    I’ve not tried the LGA771 mods either, I have dual-socket LGA771 boards for that kind of thing 🙂 A dual Harpertown board with 16GB RAM or so is still not bad.

  24. Zir Blazer says:

    (Delete my previous TWO comment, cause there was a wrong link and I added further info)

    Glad that you’re getting close. Note that there are SEVERAL parts marked as Nocona “G0?” in CPU-World, which I got merely by trying the next and previous letters to see what I find:

    QFFM 3.6 GHz – http://www.cpu-world.com/sspec/QF/QFFM.html
    QFFN 3.4 GHz – http://www.cpu-world.com/sspec/QF/QFFN.html
    QFFO 3.2 GHz – http://www.cpu-world.com/sspec/QF/QFFO.html
    QFFP 3 GHz – http://www.cpu-world.com/sspec/QF/QFFP.html
    QFFQ 2.8 GHz – http://www.cpu-world.com/sspec/QF/QFFQ.html

    I don’t know whenever someone has one, googled a few but not all. I find it strange that CPU-World list parts that no one seems to have, yet they never added your SL7HY, as there is physical proof.

    Also note that by doing the same thing about looking for the next/previous letter, you get something like this:

    SL7HY – Your mysterious Xeon
    SL7HZ – Saw a legit mention about it here, as “SL7HZ CPU”: https://www.hamiltone.co.uk/coolers.htm
    It list a “6954470300 M/B”. Googled that and got here: https://www.jaguarforum.com/showthread.php?t=12913
    Seems to be a Packard Bell Motherboard of some sorts, perhaps a renamed OEM version of the ASUS P5S800-VM/S?

    I also got here: http://www.rf-china.com/n49/cpu.asp
    …and here: https://www.hamiltone.co.uk/CPUs.htm
    Note a lot of S604 Xeons that begin with SL7H, with Nocona D0 according to CPU-World and Nocona D1 according to the other two…

    SL7HF 2.8 GHz – http://www.cpu-world.com/sspec/SL/SL7HF.html
    SL7HG 3 GHz – http://www.cpu-world.com/sspec/SL/SL7HG.html
    SL7HH 3.2 GHz – http://www.cpu-world.com/sspec/SL/SL7HH.html
    SL7HI – ?
    SL7HJ 3.4 GHz – http://www.cpu-world.com/sspec/SL/SL7HJ.html
    SL7HK 3.6 GHz – http://www.cpu-world.com/sspec/SL/SL7HK.html
    SL7HL – ?
    SL7HM – ?

    SL7HU – 2.6 GHz Prestonia, should be way older http://www.cpu-world.com/sspec/SL/SL7HU.html
    SL7HV – 2.8 GHz S604 Nocona according to the list in the other links. However, it is a Gallatin (Related to Prestonia) according to this list: http://www.cpu-world.com/forum/viewtopic.php?t=15404
    SL7HW – ?
    SL7HX – 2.8 GHz “Dell Processor Kit” https://www.sourcingstreamlined.com/it-hardware/rfq/dell/sl7hx/
    SL7HY – Your Xeon
    SL7HZ – Some LGA 775 Pentium 4?

    May I suggest to ask in CPU-World forums if some collector has these Processors or can provide more info?

  25. Michal Necasek says:

    Would be nice to know where the QFFx S-specs came from, since no one seems to have actual CPUs. It could well nave originated in an Intel NDA document, but that’s just a guess.

    The Dell SL7HX, if it’s not a typo, sounds like another off-the-books S-spec, and clearly the OEM would be Dell in that case.

    The existence of the other SL7Hx S-specs from the same era is what makes me think that SL7HY is a real S-spec assigned by Intel and not something that someone just made up.

  26. Haru Jayasekara says:

    Hi, it’s been a while but I have some updates.

    – The first thing is that I have acquired two Socket 604 QFFN Engineering Samples. These are some of the ones marked as “G0?” in the CPU-World Database.

    – Next is that I have contacted with someone who has access to an Intel database of Q-specs and S-specs. I got them to check for QDZA (the Engineering Sample with identical specifications to the SL7HY), QFFN, and for SL7HY itself. The results were interesting: QDZA is listed as Dempsey-G0, QFFN is indeed Nocona-G0, and SL7HY is nowhere to be found.

    – Now that the existence of Nocona-G0 is confirmed, I started poking around in the BIOSs of several Socket 604 motherboards. The BIOS for the MSI board you linked didn’t have the correct 0F42 microcode, but after a bit of searching, I found some boards that did: the Asus NCLV series, and the Asus NCT-D. What’s also interesting is that this microcode is Revision 2, compared to the Revision 3 one I pulled from the Asus Socket 478 motherboard and gave you.

    – Unfortunately, all of my Socket 604 motherboards are made by Intel, and from my brief examination of the BIOS versions still available on the internet (Intel, in their infinite wisdom, removed downloads for all of their older motherboards), they do not have the 0F42 microcode. So, I will try and acquire one of the aforementioned boards to test the QFFNs for NX-bit support.

    In conclusion, I’m of the opinion that the SL7HY is just a re-lasered QDZA Engineering Sample (though technically speaking, based on the CPU-Z string, it’s a Qualification Sample). I believe this because there have been no other SL7HYs located, apart from yours and the one sold on eBay shown in the CPU-World post you linked to in your previous article. In addition, both of these CPUs have an identical lot code (L422B956); their product string reports as an Engineering Sample, regardless of microcode; there is an unusually large gap between “SL7HY” and “MALAY” on the IHS’ text, not present on any other 775 Pentium 4s of a similar production date I have seen; and CPU counterfeiting with fake S-specs and model numbers is not unheard of (for example, my friend recently acquired an “i7-8700S” with S-spec SR3QH – both of these details are fake, and the CPU is in fact just a common QN8H, a Coffee Lake Engineering Sample).

    I also think that, in turn, as a CPU-World member said on the thread you linked to in your previous article, the QDZA was Intel’s aborted attempt to make a Netburst Xeon for Socket 775. This is supported by the CPU appearing as a Dempsey in the Intel database, despite the first true Dempsey chips shipping on LGA 771 in 2006, based on the 65nm Pentium D, while the QDZA was made in 2004 (as you said in your earlier post, it most likely defaults to “Nocona” in CPU-Z since there have not been enough of these samples analysed to give it its own identity). The product string backs this up, and so does the existence of the QDZC, an as-yet unlocated 775 Xeon sample with a 667MHz FSB. Despite Socket 775 never having any 667MHz FSB chips released for it, I have confirmed that it still works perfectly, by running a Xeon 5050 with an LGA 771 to 775 mod sticker on my Asus P5K-VM (my CPU-Z validation b1c5hj confirms this).

    In short, I am of the opinion that the SL7HY does not exist as a CPU ever made by Intel, and is just a QDZA ES someone re-lasered to sell as a Pentium 4.

  27. Haru Jayasekara says:

    One more thing, QGSZ, which I own and have confirmed to be A0 Dempsey and made in 2005, is for LGA 775, unlike Dempsey B0 and greater which was made for 771.

  28. Michal Necasek says:

    You could be right, although re-lasering the markings for essentially a one-off seems like an awful amount of effort for… what exactly?

    Does the Intel S-spec database list other known off-the-books models, like the 64-bit capable Socket 478 CPUs? If yes then that would be a convincing argument that SL7HY is not an Intel S-spec.

  29. Haru Jayasekara says:

    Yes, I’m pretty sure SL7QB/SL7Q8 (the 64-bit models) are in the database as well. There’s even some really obscure stuff like Socket 775 Northwoods, which most likely were never even actually made.

    As for relasered CPUs – there’s at least two SL7HYs in the wild, the one you have and the one from the CPU-World thread. We know they’re different because the CPU-World one has a deeper dent on the left hand side of the IHS. I think there were definitely more QDZAs that were relasered into SL7HY; it’s very common to see images of trays full of near-new ES/QS chips on Chinese websites like Taobao and Xianyu. A lot of the QS chips get relasered as retail versions (this happened a lot with the Coffee Lake Xeon E CPUs) to look more marketable and to avoid issues with customs clearance. So, I think it’s quite likely that more QDZAs were relasered to make them more sellable – even if the product string reads Xeon, more people would prefer a chip whose performance they know (a 3.73GHz Pentium 4) than some random Intel Confidential chip.

  30. Haru Jayasekara says:

    Another minor update: I’ve acquired a QDZA. I’ll check how it reports in various ID tools + Linux and compare it to the SL7HY.

  31. Kamil J. Dudek says:

    I love this story and the ongoing investigation 🙂

  32. Michal Necasek says:

    Cool, that should be very helpful. Curious to know what you find out!

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.