pasatoiutd

Please add - GPU - AMD RX 6800M 12GB as the one in Asus G15 Advantage Edition (Asus G513QY): https://hwbot.org/submission/4799521_

Please add: Motherboard - Asus G513QY: https://hwbot.org/submission/4799521_

Great run! Congratulations!

Official BIOS? Clockgen seems to freeze the board even for a 1-2 MHz increase of the FSB. I tried 2 CPUs, 2 types of RAM, external VGA etc.

VIA P4XB - VIA P4X266 Chipset, socket 478: http://hothardware.com/migratedcontent/reviews/images/via_p4x266a/board.htm http://valid.x86.fr/e5fzvf [motherboard not detected correctly; I'll add a picture later]

Thank you! Unfortunately, the CPUs seem kind of slow and, as they are the retail versions, the multiplier is locked. I hope I'll manage to raise the FSB somehow.

Around + 0.55 V offset. After ~ 1800 MHz, the scaling is quite poor.

Nice CPU! Did you also try it on air?

Kingston LoVo 1866 CL9 (Elpida Hyper MGH-E) @ ~ 2.0 V

Thank you for warning me. I corrected the submission.

Really Nice! Congratulations!

Nice efficiency. It seems that Windows 7 is faster than Windows XP.

VGPU @ ~ 1.35V, very low ambient (probably around - 10 C).

Boot @ FSB 140, then raised the FSB to 144.

It's a SL4X2 and it should go higher, as I pushed it 3 years ago up to 2.14 GHz (http://valid.canardpc.com/show_oc.php?id=2114877) for validation, but now it's not very cold, so its limit is around 2 GHz. The other Pentium 4 1.4 Ghz Willamette s423 I have (B2 revision) is worse, while the Intel Pentium 4 1.5 Ghz Willamette s423 is between those two.

Anyone knows why the age of the CPU isn't taken into account for my submissions in the first stage of the challenge? They are made with a Pentium 4 1.4 GHz Williamette, using a Socket 423 Asus P4T motherboard.

Yeah! Unfortunately, my efficiency is worse than that of Moonman (http://hwbot.org/submission/2695907_moonman_superpi___32m_athlon_5350_17min_14sec_453ms), so the score can be improved. Maybe Asus will release a ROG motherboard for AM1 and we will see what Kabini is capable of.

*score

Nice scor!

Sorry for the wrongly uploaded screenshot. Here is the correct one: http://img.hwbot.org/u6067/image_id_1306510.jpeg

Because my PSC memory was air cooled and it hit a wall @ 1.825V, I could not decrease tRAS below 26 and tRFC below 74, but, as you can see, all the other subtimings are quite tight.

You have 6-9-6-21-1-4-60-8-4-16-4-6-6, while I have 7-7-6-20-1-4-66-6-4-16-6-4-7. Your CL 6 vs 7) and TRef (60 vs 66) are tighter, but TRCD (9 vs. 7) and tRAS (21 vs. 20) are looser. For the other secondary subtimings there are no important differences (8-4-16-4-6-6 vs. 6-4-16-6-4-7), because tWCL is very important only for PSC. If I have to bet, I would say TRCD is more important than CL for Sandy Bridge. I cannot say I am very happy with the result, as I could not run SPi 32 M @ 106.4 x 47 with the Hypers, so I will bench again on this platform.

Unfortunately for me, it is slow, actually. It is even slower than your PSC-based submission (http://img.hwbot.org/u27054/image_id_1015875.jpeg) and Elpida Hyper is faster than PSC-based memory at that frequency. Using PSC memory and the same setup, I was more than one second a half slower (http://img.hwbot.org/u6067/image_id_1297646.jpeg).