U41.xx Observation Thread
1 May 2006 19:50:00 UTC
Topic 191158
(moderation:
U41 uses SSE3 truncation to improve the calculation performance.
SSE3 compatibles: Opteron, Turion, Athlon 64 and X2 from E3 stepping, Celeron/Pentium D, Pentium 4 (Prescott), Xeon (from Nocona), Via C7, Transmeta Efficeon TM88xx
U41.xx Observation Thread
)
Just to let know, amd64 3200+ NewCastle (Socket 754) doesn't have SSE3 instructions.
ALL GLORY TO THE HYPNOTOAD!
Do You Dare?
Potrebujete pomoc?
RE: SSE3 compatibles:
)
Opteron starting with Stepping E! Those before do *not* have SSE3, just SSE2
RE: Just to let know, amd64
)
Same for my Amd64 3200+ Winchester (Socket 939).
I need now to setup my X2 quickly !
[
But my AMD 64 3000+ Socket
)
But my AMD 64 3000+ Socket 939 (Venice Core) has SSE3 .. hihi :)
RE: U41 uses SSE3
)
Pentium 4 3,2 GHz (Prescott) needs 0.8 s per WU.
@All Prescott User, be careful,please check your CPU, older one have Problems with SSE3 instruction Code
My Boinc (DocMaboul) say : MMX,SSE,SSE2 instructions.
http://einsteinathome.org/task/27460557
I use now the S41.06 and this Albert runs excellent on my Pentium 4 hosts.
Thank You for Your Good Work, Mr. Akos Fekete.
I hope you get sometimes your Donation for this work to optimze the Albert, and you will earn lot of money.
Greetings from Germany to Hungary
Chris
*Die Signatur befindet sich aus technischen Gründen auf der Rückseite dieses Beitrages!*
RE: RE: U41 uses SSE3
)
I take it you switched back to S41.06 then? U41.01 was producing many errors on this Host
RE: I take it you switched
)
You´re right 11 WUs killed, I use HT so the WUs dies twice at time. But 0.8 s/Wu is very fast, I needed some seconds to Kill Boinc and change Albert back to Version S41.06.
Now the host run and is crunching normally and fast.
Greets
Chris
*Die Signatur befindet sich aus technischen Gründen auf der Rückseite dieses Beitrages!*
I already got 4 invalid
)
I already got 4 invalid results with my Athlon X2. That seems to a bit to much.
I'll watch it, and eventually switch back to S41.06.
But nevertheless, it's great work you do, akos!
cu,
Michael
[edit]2 more invalid results, that means 100%. I switch back.[/edit]
Running U41 on Intel D 820
)
Running U41 on Intel D 820 and it's progress is very slow
Short WU with S4106 15:54, U41 @40% after 13 minutes, projected ETA is 32 mins.
AMD64 3700+ (San Diego) is
)
AMD64 3700+ (San Diego) is finishing WUs ~35 min compared to ~44 min with D41.12
No results back yet. How does everyone else get valid/invalid results seemingly within minutes? It always takes me hours or days to get results.