DekkeR
Power Member
youngmin disse:We are not sending out currently, because F@H client does not know the difference between Athlon and A64, and using Athlon for Gromacs is a lot more productive than using Athlon for QMD (And I think the majority of AMDs are still Athlon-XP). When the next client is introduced with advance cpuid code, we will send QMD WUs to AMD64 too.
Quote feita a um responsável pelo projecto, em Stanford. Podem ler a thread toda aqui.
Mais informação sobre o assunto:
http://forum.folding-community.org/viewtopic.php?t=12415
http://forum.folding-community.org/viewtopic.php?p=94916#94916
http://forum.folding-community.org/viewtopic.php?p=99273#99273
http://forum.folding-community.org/viewtopic.php?p=100794#100794
http://forum.folding-community.org/viewtopic.php?t=12415
Para quem não sabe o que são QMD, aqui ficam umas curiosidades:
Project : 1910
Core : QMD
Frames : 100
Credit : 450
-- List of benchmarks --
DekkeR HT2 ->
Min. Time / Frame : 29mn 51s - 217.08 ppd
Avg. Time / Frame : 30mn 43s - 210.96 ppd
DekkeR HT1 ->
Min. Time / Frame : 26mn 25s - 245.29 ppd
Avg. Time / Frame : 26mn 25s - 245.29 ppd
Project : 1912
Core : QMD
Frames : 100
Credit : 450
-- List of benchmarks --
DekkeR HT2 ->
Min. Time / Frame : 01mn 31s - 4272.52 ppd
Avg. Time / Frame : 26mn 22s - 245.76 ppd
DekkeR HT1 ->
Min. Time / Frame : 00mn 50s - 7776.00 ppd
Avg. Time / Frame : 31mn 02s - 208.80 ppd
Benchs de um P4 3.2C com 2x512MB.
O grande senão destas WUs é mesmo a quantidade massiva de RAM que elas consomem, como podem ver: