Optomized S5 SSE3

Crunch3r
Crunch3r
Joined: 22 Jan 05
Posts: 90
Credit: 30237616
RAC: 0

RE: S5T0001 - eliminated

Message 38876 in response to message 38874

Quote:

S5T0001

- eliminated double jumps

Quite unusual behaviour when you look at the progress.
Seems to "jump" sometimes 0.004 to 0.005% per/sec. ahead.

nice work Akos ;)

IndianaX
IndianaX
Joined: 4 May 06
Posts: 8
Credit: 169171983
RAC: 0

I have installed patch 000

I have installed patch 000 and 001 on this PC:
http://einsteinathome.org/host/628724/tasks
18 Jun 2006 10:33:09 UTC is the last without patch (~27200 seconds per WU).
I have only big WU's so the first results will come in 5 hours.

Have a N.I.C.E. day!

Akos Fekete
Akos Fekete
Joined: 13 Nov 05
Posts: 561
Credit: 4527270
RAC: 0

RE: RE: S5T0001 -

Message 38878 in response to message 38875

Quote:
Quote:

S5T0001

- eliminated double jumps


so should we stop again and update now or wait?

No! I would like to be sure that S5T0000 is correct. I need at least 10-20 valid results without invalid. But other people can do the test with S5T0001...
If S5T0001 is invalid then i can jump back to S5T0000...

Akos Fekete
Akos Fekete
Joined: 13 Nov 05
Posts: 561
Credit: 4527270
RAC: 0

S5T0002 - eliminated

Message 38879 in response to message 38874

S5T0002

- eliminated double jumps
- reduced amount of FPU macro ops (hot loop size: 4831)

Pepperammi
Pepperammi
Joined: 20 Feb 05
Posts: 131
Credit: 437943
RAC: 0

RE: RE: RE: S5T0001 -

Message 38880 in response to message 38878

Quote:
Quote:
Quote:

S5T0001

- eliminated double jumps


so should we stop again and update now or wait?
No! I would like to be sure that S5T0000 is correct. I need at least 10-20 valid results without invalid. But other people can do the test with S5T0001...
If S5T0001 is invalid then i can jump back to S5T0000...


No prob. Ill stay with S5T0000. Add my other machine when i can.
I've suspended my other projects for now to get you these long results a bit quicker.

Mats Nilsson
Mats Nilsson
Joined: 10 Dec 05
Posts: 94
Credit: 15011147
RAC: 0

Should one install all

Should one install all patches or it´s enough too take the last one, for the moment S5T0002

Crunch3r
Crunch3r
Joined: 22 Jan 05
Posts: 90
Credit: 30237616
RAC: 0

RE: Should one install all

Message 38882 in response to message 38881

Quote:
Should one install all patches or it´s enough too take the last one, for the moment S5T0002

Well i think applying the last patch should be enough, because all patches focus on the same issue (SSE hot loop).

Anyhow to be on the safe side i'd say use all patches ;)

ca_grufti
ca_grufti
Joined: 9 Feb 05
Posts: 53
Credit: 4309237
RAC: 0

Running with S5T0002 on

Running with S5T0002 on Athlon 4800+. No problems so far. Waiting for the first results.

Pepperammi
Pepperammi
Joined: 20 Feb 05
Posts: 131
Credit: 437943
RAC: 0

RE: Should one install all

Message 38884 in response to message 38878

Quote:
Should one install all patches or it´s enough too take the last one, for the moment S5T0002

Quote:
Quote:
so should we stop again and update now or wait?
No! I would like to be sure that S5T0000 is correct. I need at least 10-20 valid results without invalid. But other people can do the test with S5T0001...
If S5T0001 is invalid then i can jump back to S5T0000...


LiborA
LiborA
Joined: 8 Dec 05
Posts: 74
Credit: 337135
RAC: 0

I have maybe a stupid

I have maybe a stupid question: are these patches only for CPU with SSE3 support, or I can test it on SSE/SSE2 CPU too?

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.