• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

402blownstroker's 4P Build

sc0tty8 said:
402, how many boxes do you have folding? I see you are on my threat list.
Click to expand...

I have the follding systems folding:
  • 4P - 6166HE
  • SR-2 5645 @ 3.6Ghz
  • Dual 8389
  • Dual 5440 + GTX460 + GTS450
  • HpCloud - (12) 2 vCPU + (2) 4 vCPU
  • T61P
 
Well with 'the kraken' installed, a 6903 WU has gone from 19m45s TPS to 18m20s. That is a nice improvement of 7%. With the quad channel ram and 'the kraken', it is an improvement of 28% over the original configuration :D Later this morning I will flashing the new OC BIOS and applying a 10% OC. I will post improvements with that later.
 
402, what are your RAM timings?
I'm asking because I have the same system (except I use Ubuntu 10.10) and my 6903 TPF is more than a minute better than what you said. Depending on the WU (although still 6903), my TPF varies from 16:55 to 17:20.
You could also try Ubuntu 10.10 to see if it's a significant factor.
 
I can try ubuntu here. Are you running the custom kernel listed in musky's install guide?
 
I smell a case of DLB (Dynamic Load Balancing) not getting engaged...
 
I'm running the stock Kernel but I did use Musky's guide, to the letter actually.
My memory is at 1333 CL7 and DLB engages most of the time, don't know what's triggering it though...
The TPF for 6901 is also better than I expected, at 7:45.
I can't wait to see what these chips can do when overclocked :D
 
With a 10% OC through the new BIOS/CMOS, TPF are 17m30s. About 50s improvement ..... a little depressing.
 
CPU utilization seems off really and could be the answer. However, I have no idea why that would happen. I'd suggest trying Musky's guide with a clean install and see if it helps.
 
With the only real difference is a 2.6 vs a 3.2 kernel, I can not see that Ubuntu 10.10 will make much difference. Unless the FAH cores are taking advantage of 2.6 aspect that is not the same in 3.2. Dumber things have been programmed..... Windozes offers too many examples to list :D
 
One thing I must state is that going from 23m23s to 17m30s is a direct result of this forum and the great people here. An improvement of 33% is no small thing!!! Just want the 'AOL' affect to kick in .... ME TOO!
 
17.5m TPF looks a bit off (I was expecting 16.5m or less)...

Is that with or without DLB?

Have you had a chance to go through performance checklist by any chance?

Also check OC BIOS FAQ entry:
Q: Why am I not getting as much of a performance boost as I expected?
Click to expand...

Please run this (to verify application of Kraken) and post the output (4600% sounds suspicious):
Code: 
ps -Leotid=,psr=,pcpu=,comm= | grep thekraken
and this as well (to determine if there are CPU consumers other than FahCore):
Code: 
ps -eotid=,psr=,pcpu=,comm= | sort -nrk3,3 | head -10
and this (to verify the OC):
Code: 
dmesg | grep -o Detected.*
 
Sorry for the delay in info, but princess pretty pretty was demanding a bit of her man servant :D

Keep in mind this is a Fedora 16 x86_64 system.... just don't know any better:

ps -Leotid=,psr=,pcpu=,comm= | grep thekraken
Click to expand...

2369 33 0.0 thekraken-FahCo
2436 0 96.5 thekraken-FahCo
2437 37 0.0 thekraken-FahCo
2438 40 0.0 thekraken-FahCo
2444 1 98.4 thekraken-FahCo
2445 2 98.5 thekraken-FahCo
2446 3 98.6 thekraken-FahCo
2447 4 98.4 thekraken-FahCo
2448 5 98.3 thekraken-FahCo
2449 6 98.0 thekraken-FahCo
2450 7 97.9 thekraken-FahCo
2451 8 98.0 thekraken-FahCo
2452 9 98.2 thekraken-FahCo
2453 10 98.1 thekraken-FahCo
2454 11 98.3 thekraken-FahCo
2455 12 98.2 thekraken-FahCo
2456 13 98.2 thekraken-FahCo
2457 14 98.4 thekraken-FahCo
2458 15 98.2 thekraken-FahCo
2459 16 98.4 thekraken-FahCo
2460 17 98.4 thekraken-FahCo
2461 18 98.1 thekraken-FahCo
2462 19 98.2 thekraken-FahCo
2463 20 98.1 thekraken-FahCo
2464 21 98.1 thekraken-FahCo
2465 22 98.0 thekraken-FahCo
2466 23 98.1 thekraken-FahCo
2467 24 98.5 thekraken-FahCo
2468 25 97.9 thekraken-FahCo
2469 26 98.0 thekraken-FahCo
2470 27 98.1 thekraken-FahCo
2471 28 98.1 thekraken-FahCo
2472 29 98.1 thekraken-FahCo
2473 30 98.2 thekraken-FahCo
2474 31 98.2 thekraken-FahCo
2475 32 98.2 thekraken-FahCo
2476 33 97.9 thekraken-FahCo
2477 34 98.0 thekraken-FahCo
2478 35 98.2 thekraken-FahCo
2479 36 98.1 thekraken-FahCo
2480 37 98.3 thekraken-FahCo
2481 38 98.3 thekraken-FahCo
2482 39 98.1 thekraken-FahCo
2483 40 98.2 thekraken-FahCo
2484 41 98.2 thekraken-FahCo
2485 42 98.1 thekraken-FahCo
2486 43 98.1 thekraken-FahCo
2487 44 98.1 thekraken-FahCo
2488 45 97.8 thekraken-FahCo
2489 46 97.9 thekraken-FahCo
2490 47 97.9 thekraken-FahCo

ps -eotid=,psr=,pcpu=,comm= | sort -nrk3,3 | head -10
Click to expand...
2369 33 4707 thekraken-FahCo
11231 0 6.1 firefox
2426 9 2.1 gkrellm
2089 0 0.7 konsole
1737 9 0.5 X
99 23 0.0 ksoftirqd/23
98 23 0.0 kworker/23:0
97 23 0.0 migration/23
96 22 0.0 watchdog/22
95 22 0.0 ksoftirqd/22

firefox is not displaying porn in any way... it is not built to do that :D


dmesg | grep -o Detected.*
Click to expand...
None, nothing, zip, and no dreaded NullPointer stack trace :D
 
I hope thay can get this straightened out for you. 1.98 Ghz at 17m30s is a wee bit off, my 1.9 Ghz with no ram timing adjustments and no run through the optimization check list is averaging 16:38 (16:18 last 3) across 80 frames on a 6903.

You should be atleast a minute faster, probly more since you have nearly 100mhz advantage and ram speeds over myself right now.

All this talk of seeing you go 33% faster so far, and still not being done, is really making me wanna get a SM board so I can OC. I'm curious to see how fast I can get these chips to, with a 20% OC putting me at nearly 2.3 Ghz, and optimized ram timings, it just gets me all giddy thinking how much more PPD I could be doing lol

But if anyone can get your board producing more, its these guys here
 
All right, The Kraken looks good (CPUs 0-47 at 90+% usage).

Get rid of firefox (unconditionally) an gkrellm (at least for the sake of experiment).
6% CPU eaten by FF is definitely to blame for some performance degradation.

Can you go through musky's checklist and check HT retries as well (per FAQ), please?

The fact of dmesg buffer losing initial boot messages is interesting.
What do:
Code: 
dmesg | tail -10
grep MHz /proc/cpuinfo  | uniq
return?
 
With a little free time today, I did some tests and it definitely looks like a bad DIMM. memtest always generates errors in the 13.1GB to 13.5GB. So it looks like it is just one DIMM. I went and order another set, but the question is..... does anyone know which slot corresponds to 13GB with 2GB DIMMs installed? I tried the IRC channel, but a couple of 2 year olds where one filling up things with a lot of BS.
 
Good find, Mister.

13.1-13.5 should be the seventh DIMM (round up to next even GB and divide by 2).
If I remember enumeration rules correctly it's going to be third DIMM of CPU2.

Don't take my word for it tho; remove it and re-run memtest; see if error's still there...
or test it individually...

h8qgi-f-bare-402-dimms.jpg
 
Thanks tear. I have a new pair of DIMMs on the way from Newegg. Until they arrive, I have found a setting of 214 with the OC utility does not piss off that DIMM as both memtest and ht-retries.sh run cleanly.

On a serious note, need to get the IRC crew a couple cases of beer and tomales or a hire a stripper. Which ever will go further :D
 
Damn! The new DIMMs came today, I swapped out the 7th one, reflashed to 220, and I am getting memory errors again in the 13.1GB to 13.5GB range. 1 slot tested..... 15 more to go.
 
Up and running strong at 220 :D

Turns out the DIMM in slot '6' was the bad one for the 13.1ish GB to 13.5ish GB range. memtest ran error free over the entire memory range for 2 hours and ran error free for 1 hour in the 12GB to 14GB range.

I am going to let it run at 220 for a couple of WUs and then see how much I can bump it up. Also get the 212+ installed too.
 
You must log in or register to reply here.
Back
Top