VOGONS


First post, by andrea

User metadata
Rank Member
Rank
Member

Hello to all,
I was wondering, has anyone ever tried a Cedar Mill P4 in an early, supposedly Prescott only, 775 motherboard?.
I was thinking on upgrading an ASUS PTGD1-LA, or an HP Puffer2, depending on what you want to call it, with the i915P Chipset. Both the ICs on the board and the CPU date from around the 30th week of 2004 so I get the feeling this has to have been a launch system for the new and fancy world of LGA775 and PCI Express.

The latest bios available, sp26760.exe, from Jan. 2006 does include the CPUID F62 for B1-step Cedars (the only production stepping then, from my understanding). But CPUID alone is useless without further, more lower level, BIOS and HW support, which I don't know how to check.

To the best of my knowledge between 2004 and 2006 there were 3 versions of the 775 socket:
1. Prescott only*
2. Prescott, Cedar Mill, Smithfield (lol) and Presler
3. Prescott, Cedar Mill, Smithfield, Presler and Conroe+

But is 1 truly Prescott only or is it rather single core only due to the crippling of the i915 chipset?

If you search along the Interwebs you'll see I'm not the first to ask this question. The answers you'll find are all along "It shouldn't work", "Just get a 2M 6x1 Prescott", "lol why r u upgrading, just get a 5K$+ (whatever was the highest end thing at the time) system". But noone seems to actually have tried it.

I get the feeling the answer to the above is probably on the russian-speaking part of the Internet, as they are quite good in finding workarounds for unsupported upgrades (such as Prescotts in Northwood motherboards, or Conroe cpus in "type 2" motherboards), but as good as google translate is, if I can't write a good search query the answers will be equally bad.

*** EDIT 24/4/21 16.00
I've found this: https://forum.ixbt.com/topic.cgi?id=4:90088. Google translate tells me that a Cedar Mill worked on a P5GD1 (officially unsupported) with only a microcode error on boot (easily solvable) and some speedstep funkyness (probably lack-of-ucode releated). Can someone who understand Russian confirm if Google did translate right?

Granted, it's a different board, also not from a prebuilt, but both that and mine are ASUS board with i915 chipset and a BIOS based on AMI version 8. Furthermore a OEM board only has one requirement: to be cheap. So I don't see them reinventing the wheel for no reason
I've found a P4 641 for 5€ shipped from china so might as well order one and try.

Reply 1 of 5, by Ydee

User metadata
Rank Member
Rank
Member

I cannot say, how it will be in this case, but I myself operate Pressler D935 on Gigabyte 8I865GME-775-RH rev.1.x, which is not listed in supported CPU and everything works without problems. In the line to the Russian forum, the user describes the full functionality of Cedar Mill on the P5GD1 board, although it is also not listed as supported. The only problems are CPU identification when loading the microcode and then in the Asus AiBooster is jumping the CPU multiplier and frequency, in Everest everything is fine, also in the Asus PCProbe. Test in Prime95 9 hours OK.

Reply 4 of 5, by Ydee

User metadata
Rank Member
Rank
Member
andrea wrote on 2021-04-26, 19:00:

(Not gonna lie, paying for a P4 in 2021 feels weird)

Really, it looks a little perverse 😀
But seriously - I also think it will work without a problem, and Cedar Mill (and Pressler), thanks to 65nm technology, at least raised a bit of the flag Pentium 4. There was nothing they could do about NetBurst's architecture, and they had no chance against Intel Core, but at least they had corrected the poor reputation of Prescott's "heaters." I was an AMD fan in the days of LGA775, so I only got Pressler of interest for 4 euros to a board that was free of charge.