Wren's CPU Misadventures

[wren4777]

For the past while, I've researched and/or discovered some random pieces of previously unknown information about CPUs etc that don't really need separate threads of their own. So, I'm making this thread as a blog of sorts so this information can be documented somewhere

[wren4777]

As some of you have probably seen, I've been trying to track down information on this mysterious SL7HY Pentium 4 for many years now.

http://www.os2museum.com/wp/mystery-netburst/
http://www.os2museum.com/wp/more-about-that-strange-pentium-4/

To summarise, it is (supposedly) a 3.73GHz/1M cache/1066MHz FSB LGA 775 Pentium 4. What makes it interesting is that the sSpec is completely undocumented, and in both motherboard BIOSs and CPU-Z etc, it actually IDs itself as a Qualification Sample 3.73GHz Intel Xeon.

This behaviour is identical to the QDZA ES, what is theorised to be a cancelled Netburst Xeon for LGA 775, along with the QDZC and QEHB: https://www.cpu-world.com/forum/viewtopic.php?t=15723

What is extra special about the QDZA is that it uses the very rare F42 CPUID, which indicates a G0-step Nocona/Prescott core. The only other CPU with this CPUID is supposedly the Intel Celeron SL8S4, but no evidence of that being real has surfaced.

Previously, the only microcode for this CPUID was found in Socket 604 and 478 motherboards. However, recently I discovered a new variation of this microcode in an LGA 775 motherboard, the Asus P5MT-S. Since this is a server/workstation board, it gives further credibility to the theory that Intel cancelled their 775 Netburst Xeon lineup at the very last minute. I have attached this new microcode here, if anyone wants to experiment with running it with their QDZA (rename it from .zip to .bin first).

Since the SL7HY reports itself as one of these Xeons and not a Pentium 4, I think it is most likely really a QDZA that had its IHS markings removed and faked as a P4 in order to be sold.

[xsecret]

Some research:

* SL7HY is not an Intel part (not listed anywhere in Intel DB). It might be a s-spec produced exclusively for a third party or a crafted P/N used by fakers to rebrand an ES.

* QDZA/QDZC/QEHB are genuine Intel parts.

* SL8S4 is listed as µPGA478 Celeron D 310 (Boxed and Tray) but no proof it was ever produced. This said, this particular s-spec has q-spec counterparts (QIGR/Q/X/Y). SL8S2 is the same one with G1-Step.

* There is only 32 entries for CPUID F42 on CPU-Z database, all ES on various LGA775 boards. Earliest one from Nov, 2005.

* Many of them were flagged as P4 EE and used by well known top overclockers like coolaler. It's probably not a coincidence... Probably a special batch for overclockers. They were pretty common back in these days.

[wren4777]

Thank you very much for the information! I understand if you'd rather not share the CPU-Z validations, though if you could I'd be very grateful. If that isn't possible, could you please tell me if there are any other F42 parts with base clocks other than 3.73GHz? This should then prove the existence of QDZC/QEHB, which I haven't been able to do yet.

What you said about the overclockers does make sense. I vaguely recall finding a user or two on overclock.net who had acquired a QDZA from a computer shop where they worked, or something like that.

[xsecret]

Most are 3.73 GHz, but a few are 2.93GHz and even a strange 3.60 GHz

Genuine Intel® CPU 3.60GHz - https://valid.x86.fr/jermxy
Intel Xeon CPU 2.93GHz (ES) - https://valid.x86.fr/120441
Intel Xeon CPU 2.93GHz (ES) - https://valid.x86.fr/685115

[wren4777]

Wow! So that confirms the existence of QEHB then, since that's the 2.93GHz part. Nothing for QDZC yet, but this gives me hope that I'll be able to find one one day That 3.6GHz part is very interesting too. I wonder what it's Qspec is...

Thank you again for the information!

[xsecret]

QDZC is a 3.00GHz part with FSB 667.

[wren4777]

Yes, that 667MHz FSB is what makes me want it so much I do wonder if the lack of validations is because people had trouble booting them on their boards. I had no issue running a 775-modded Xeon 5050 with its 667MHz FSB on my Asus P5K-VM, but I have heard some Gigabyte owners having trouble with it.

[xsecret]

Yep. There is tens of millions of CPU-Z validation since almost 20 years now, so if a CPU isn't there, it's probably extremely rare.