You need to use the new core: Proof!

[Wedo]

Hey,

I just slapped the new core and SSE flag onto an AMD rig and here's the log file that shows how much faster FAH is with these options:

Old core, ASM flag log:

[19:03:42] Protein: p563_BBA5_ext
[19:03:42]
[19:03:43] Writing local files
[19:03:45] Completed 77500 out of 250000 steps (31)
[19:08:20] Writing local files
[19:08:22] Completed 80000 out of 250000 steps (32)
[19:13:02] Writing local files
[19:13:04] Completed 82500 out of 250000 steps (33)


Roughly 4:37 per frame.

New Core, SSE flag log:

[19:18:34] Protein: p563_BBA5_ext
[19:18:34]
[19:18:34] Writing local files
[19:18:34] Completed 82500 out of 250000 steps (33)
[19:18:34] Extra SSE boost OK.
[19:22:20] Writing local files
[19:22:22] Completed 85000 out of 250000 steps (34)
[19:25:56] Writing local files
[19:25:58] Completed 87500 out of 250000 steps (35)

About 3:36 per frame.

This is almost a 25% increase in production just for this protein. Factor in that according to Stanford some proteins don't even use 3DNow even though it's recorded as such.

Get the new Core, change your flag, do it NOW!!!!

Wedo

 

[Planus]

when was this core released, and how do I get it? Just delete the old ones and restart the client?

 

[Fever]

I got a performance increase too, but only about 20 seconds. I suppose it depends on the protein being folded.

Get the core here .

All you need to do is close FAH, replace the old core in your FAH directory, and restart your client.

 

[mdcomp]

I got a performance increase too, but only about 20 seconds. I suppose it depends on the protein being folded.

Get the core here .

All you need to do is close FAH, replace the old core in your FAH directory, and restart your client.

You do have to reinsert the flags correct? Still -forceasm and -advmethods?

Matt

 

[Quailane]

Not if you are already using them.

 

[OSUmaxx]

I got a performance increase too, but only about 20 seconds. I suppose it depends on the protein being folded.

Get the core here .

All you need to do is close FAH, replace the old core in your FAH directory, and restart your client.

Your link doesn't work. Any other ideas? I haven't been keeping up with new folding stuff for a bit now...

 

[wizard james]

how about the intel?....i dont see anything about intel...does it make it go faster on my p4 too?

 

[SavageBasher]

Its only for AMDs. You can use the -forceSSE flag now.

 

[slater3333uk]

Yep Its doubled production of some wu's on my amd rigs.

From about 250ppw without -forceSEE to about 500-600ppw with -forceSEE.

 

[Maxvla]

i've been using forcesse on my xp chips all along with no problems. will this new core improve times even if i've been using the flag already?

 

[Quailane]

I haven't noticed a difference on my stepdad's 1700 pally.

 

[nikhsub1]

i've been using forcesse on my xp chips all along with no problems. will this new core improve times even if i've been using the flag already?

Nothing significant... Wedo's point is that XP's folding with 3DNow! really blow big time in PPD/PPW. We've had to run with 3DNow! only on the corporate machines as we can not have them crash.... The new core eliminates the crashing with SSE so, the real difference is switching to SSE from 3DNow!. I would upgrade ALL your AMD rigs with the new core if you have not and run SSE. We have seen not ONE issue with the new core.

 

[Soja]

The new core meaning beta core w/ the amd sse fix?

 

[nikhsub1]

The new core meaning beta core w/ the amd sse fix?

Yes.

 

[Wedo]

If you use -forceasm flag it defeats the purpose of the new core. You should use -forcesse. It doesn't affect p4 so you can just use -forcesse on all of your rigs to make things easier.

Wedo

 

[Fever]

Your link doesn't work. Any other ideas? I haven't been keeping up with new folding stuff for a bit now...

The link works, they must've been having problems.

The new core shaves about 30 minutes per WU on average I'd say, for people using SSE with previous cores.

 

[GigaHertzAddict]

Wedo, What system are you folding the posted results on? The Dual 2500XP's?

I'm running a 2200XP at 2324MHz with 166MHz bus speed and Memory. The protien I'm folding is: p520_BBA5_pf. And I'm getting 5.88 minutes with the new core. Is this good? I traded my stock barton 2500 for my current setup. Do I do a good thing?

 

[Wedo]

Wedo, What system are you folding the posted results on? The Dual 2500XP's?

I'm running a 2200XP at 2324MHz with 166MHz bus speed and Memory. The protien I'm folding is: p520_BBA5_pf. And I'm getting 5.88 minutes with the new core. Is this good? I traded my stock barton 2500 for my current setup. Do I do a good thing?

Don't worry about frame times as every protein is different and within each protein, each WU is different as well. The system noted above is an XP2400. If your system works for you, that is all that matters. I think an XP running at 2324 is a sweet deal either way.

Oh, and by the way, that would be Dual Bartons .

Wedo

 

[Audioaficionado]

Since I already successfully use forceSSE on my AMD rigs, will I see any improvements with the new core?

 

[nikhsub1]

Since I already successfully use forceSSE on my AMD rigs, will I see any improvements with the new core?

Maybe a 1% improvement is what has been said. The best part is, you will now not have any lockups.