Sharky Forums Folding Team

except i'm using linux. i'm not sure about the method to add those optimizations. i'll have to look around to see if there's any info on the web for maximizing performance with the linux client. or maybe colossus knows some tricks.

any other place than here to get the betas? :confused: that page is giving me 404 errors.

Quote:

---

Originally posted by sapasion
Save the beta to your Folding@home folder and make a shortcut of it. Cut and Paste the shortcut in your startup folder, Make the following adjustments to the shortcut by right clicking and selecting properties. Then make the highlighted line look like mine

Mind you this is for 3.25beta for 4.00beta instead of -forceasm make it -forceSSE. advmethods will enable you to receive and work on Gromacs Wu's which utilize SSE and have a far better WU points/time ratio than the Tinker WU's like the one you referenced above. forceasm or forceSSE keeps SSE active for AMD cpus. This is not necessary for Intel but it is for AMD cpus which support SSE. Re-boot is also probably necessary. Upon restart you should have a faH-78 core instead. These little tweaks will more than double and perhaps triple your weekly WU points.

---

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> it looks like you just add the -forceasm and -advmethods to the startup command, just like in windows. i have probably about a day and half to go, so if it's twice as fast guess i should kill this WU and start over with a gromacs and gain half a day. :D or should i finish the tinker one first?

Quote:

---

Originally posted by serotone9
any other place than here to get the betas? :confused: that page is giving me 404 errors.

---

yeah i just checked it too, none of the links work :confused:

Just remove the file name and it will give you a directory listing :)

http://www.stanford.edu/group/pandegroup/release/beta/

Quote:

---

Originally posted by Colossus
Just remove the file name and it will give you a directory listing :)

http://www.stanford.edu/group/pandegroup/release/beta/

---

:cool:

Colossus, why are you banned? and if you're banned, how are you posting? :p

Quote:

---

Originally posted by serotone9
:cool:

Colossus, why are you banned? and if you're banned, how are you posting? :p

---

Okay, not everyone was involved in the past inquiries. He's not banned, just playing around. He was real busy and stopped posting for a little bit and changed it. :)

[sarcasm]Congrats on your needing to use multiple pc's to pass me Vert![/sarcasm];) Seriously, you're gonna be another powerhouse on the team...i can't believe schultz is still with me after another day of 24/7 at 2.4...what is his computer on?:confused: but i see we broke into the 800's now, so that's cool. Stealing a line, CONQUER ON!!:p :D

are certain WU's associated with either the Tinker or Gromacs core? all 3 of my rigs have the 3.25 beta running with -advmethods -forceasm, but only 2 of them show the Gromacs core/"extra SSE boost OK" in the log file

the ones that do are currently working on a 254 and a 836, the one that isn't is working on a 638.

Yes here's the Points Listhere

[15:35:46] Protein: p353_unf_305_99
[15:35:46]
[15:35:46] Writing local files
[15:35:46] Extra SSE boost OK.
[15:35:46] Writing local files
[15:35:49] Completed 0 out of 2500000 steps (0%)
[15:43:40] Writing local files
[15:43:42] Completed 25000 out of 2500000 steps (1%)
[15:51:25] CoreStatus = 0 (0)
[15:51:25] Client-core communications error: ERROR 0x0
[15:51:25] Deleting current work unit & continuing...
------------------------------

:eek:

i guess i need to go back to non-optimized settings, or lower my OC a little. although i am only at 11x205 42C, it should be able to handle that no prob. ?

before the error it seems i was doing 1% per 8 minutes = 13.3 hours, which is triple the non-optimized rate. :cool: so maybe i should just reduce the memory timings a little, or crank up the vcore a notch?

Quote:

---

Originally posted by serotone9
[15:35:46] Protein: p353_unf_305_99
[15:35:46]
[15:35:46] Writing local files
[15:35:46] Extra SSE boost OK.
[15:35:46] Writing local files
[15:35:49] Completed 0 out of 2500000 steps (0%)
[15:43:40] Writing local files
[15:43:42] Completed 25000 out of 2500000 steps (1%)
[15:51:25] CoreStatus = 0 (0)
[15:51:25] Client-core communications error: ERROR 0x0
[15:51:25] Deleting current work unit & continuing...
------------------------------

:eek:

i guess i need to go back to non-optimized settings, or lower my OC a little. although i am only at 11x205 42C, it should be able to handle that no prob. ?

before the error it seems i was doing 1% per 8 minutes = 13.3 hours, which is triple the non-optimized rate. :cool: so maybe i should just reduce the memory timings a little, or crank up the vcore a notch?

---

Talk with Saps about it. IIRC, he added some vcore to get it a little more stable.

Quote:

---

Originally posted by serotone9
[15:35:46] Protein: p353_unf_305_99
[15:35:46]
[15:35:46] Writing local files
[15:35:46] Extra SSE boost OK.
[15:35:46] Writing local files
[15:35:49] Completed 0 out of 2500000 steps (0%)
[15:43:40] Writing local files
[15:43:42] Completed 25000 out of 2500000 steps (1%)
[15:51:25] CoreStatus = 0 (0)
[15:51:25] Client-core communications error: ERROR 0x0
[15:51:25] Deleting current work unit & continuing...
------------------------------

:eek:

i guess i need to go back to non-optimized settings, or lower my OC a little. although i am only at 11x205 42C, it should be able to handle that no prob. ?

before the error it seems i was doing 1% per 8 minutes = 13.3 hours, which is triple the non-optimized rate. :cool: so maybe i should just reduce the memory timings a little, or crank up the vcore a notch?

---

Yes I found I needed to add .05v to my OC to make it stable. Folding punishes your cpu like nothing else I've ever used. 3DMark, SuperPi whatever are nothing. There are even warning about folding with an OC'd cpu; especially an OC'd AMD cpu running with SSE. When you think about you're asking your cpu to run full load 24/7. 9 out of 10 says if you add .05v you'll stabilize that OC. And as you found out first hand the gains are worth it

Quote:

---

Originally posted by sapasion
And as you found out first hand the gains are worth it

---

as you predicted. :p

i bumped down to 203x11 and it seems stable, though i've been on my laptop for a while and haven't checked it since 4%. but i lost a minute per step, from 8 to 9. the thing is, i'd just hate to get 1/2 done with a WU and lose it. :mad:

when this one's finished, i'll bump back up and increase the voltage, see how it does. i may even have to retire the trusty AX-7 and invest in an SLK + ceramique. :D

~~~~~~~~~~~~~~~~~~~~~
new slogan suggestion:

We Came, We Crunched, We Conquered!
:D

Quote:

---

Originally posted by serotone9
as you predicted. :p

i bumped down to 203x11 and it seems stable, though i've been on my laptop for a while and haven't checked it since 4%. but i lost a minute per step, from 8 to 9. the thing is, i'd just hate to get 1/2 done with a WU and lose it. :mad:

when this one's finished, i'll bump back up and increase the voltage, see how it does. i may even have to retire the trusty AX-7 and invest in an SLK + ceramique. :D

~~~~~~~~~~~~~~~~~~~~~
new slogan suggestion:

We Came, We Crunched, We Conquered!
:D

---

And that's another thing to keep in mind. I was folding at 2.4Ghz and 1.95v when I started getting problems also my PSU was crapping out which didn't help. Anyway, in hindsight at 1.95v I now realize 2.35Ghz is probably the limit of my stability. As aforementioned I bumped up .05v to 2v and now its quite stable. So my counsel is this. Either bump the vcore and IMO with an AX-7 and a decent fan you can very safely run your Barton at 1.9v. Or drop your OC a few Mhz. Personally, I'd do the vcore up. But if you should to clockdown remember if you drop from say 2.2Ghz to 2.1Ghz that's still just a few % and its better to be stable at a lower speed and always finish WU's than dropping WU's at a higher speed.

alright, i'll buy into the -advmethods thing (though even without the tag i've only been given gromacs so far) on pure faith in sapasion. but does it matter whether i choose to crunch for folding or genome?

Your faith shall be rewarded. Since I went to advmethods I don't believe I've had any tinkers; though some report they still get thru sometimes. Without advmethods you can get Gromacs but you also can get Tinkers. Also its probably best to select Folding@Home as your client type.

but WHY should i select folding over genome? is it just a what-project-do-you-want-to-solve issue, or are there significant computational/scoring ramifications?