One Steep Ramp Up

Congrats on your PPD.

Someone finally surpassed my 300k PPD record! You nearly doubled it! Keep it up

ROC
 
Last edited:
So this is the ROC that i've heard so much about?

How about coming and joining our team as we make our final stand against that other team? For old times sake? :)
 
Arrived. built. etc. (L5640 ES)

I am sure I am doing something wrong. But at least it is running for now. As this is a new board for me any pointers apprciated.

SR-2.jpg


http://lh6.ggpht.com/_H6G8c7wgQmo/TKxXW-eaDFI/AAAAAAAAAHQ/xzQoSQk8f9A/SR-2.jpg
 
Last edited:
Needs more Vcore, You have room up to 1.325 with no issue, 1.35 is intel max, and some have run it to 1.4.
(I think these numbers are correct)


Try 1.3 and see what that gets you.
 
1. Shut down the GPU client to get a baseline reference. I realize that is a beast of a card, but a suspect it still isn't going to gain you much for ppd on that machine.

2. Pick a performance metric. I use LinX pass times at a particular memory setting. MIBW uses frame times on a saved-off WU (I would love to hear some more details on this). This will tell you if more clock speed is really giving you a performance increase. You may be surprised at the results.

3. Since your Vtt/IOH/Memory voltages appear to be fine for 200 bclk, the only things you need to mess with is bclk and Vcore. Back down to a safe bclk (180 or so) and run your performance metric. Increase the bclk and re-run the metric. If the metric proves that you are actually faster, repeat the bclk increase and metric run. If not, try a little more Vcore at the new bclk and re-run the metric. If your results mirror mine, you will hit a point around 195 bclk where any increase does not yield better performance. If you can figure out how to get over this, please let us all know.

As a reference, here are my frame times on that 6069 unit at 3564 MHz:

Project ID: 6069
Core: GRO-A3
Credit: 481
Frames: 100


Name: cowgirl
Path: \\COWGIRL\fah\
Number of Frames Observed: 300

Min. Time / Frame : 00:01:13 - 69,521.1 PPD
Avg. Time / Frame : 00:01:16 - 65,445.6 PPD
 
Baselines so far. Currently on the P6058. P6059 was with the GPU running part of the time. GPU is off as the moment.

Project ID: 6058
Core: GRO-A3
Credit: 481
Frames: 100

Name: cpu smp-24
Path: C:\Program Files (x86)\FAH\smp\
Number of Frames Observed: 17

Min. Time / Frame : 00:01:30 - 50,785.2 PPD
Avg. Time / Frame : 00:01:30 - 50,785.2 PPD


Project ID: 6059
Core: GRO-A3
Credit: 481
Frames: 100

Name: cpu smp-24
Path: C:\Program Files (x86)\FAH\smp\
Number of Frames Observed: 183

Min. Time / Frame : 00:01:11 - 72,479.2 PPD
Avg. Time / Frame : 00:02:32 - 23,138.5 PPD

Project ID: 6071
Core: GRO-A3
Credit: 481
Frames: 100


Name: cpu smp-24
Path: C:\Program Files (x86)\FAH\smp\
Number of Frames Observed: 100

Min. Time / Frame : 00:01:12 - 70,974.5 PPD
Avg. Time / Frame : 00:01:12 - 70,974.5 PPD


Project ID: 10632
Core: OPENMMGPU
Credit: 611
Frames: 100

Name: gpu 0
Path: C:\Program Files (x86)\FAH\gpu-0\
Number of Frames Observed: 263

Min. Time / Frame : 00:00:33 - 15,997.1 PPD
Avg. Time / Frame : 00:01:02 - 8,514.6 PPD
 
Is this at 3.6 still, and did you shut off NUMA?

For comparison (this would be before I shut off NUMA):

Project ID: 6058
Core: GRO-A3
Credit: 481
Frames: 100


Name: cowgirl
Path: \\COWGIRL\fah\
Number of Frames Observed: 100

Min. Time / Frame : 00:01:11 - 72,479.2 PPD
Avg. Time / Frame : 00:01:14 - 68,116.7 PPD
 
Did not shut off NUMA yet. That is the next thing. Well,,,, before trying to apply more power.
(Life get's in the way of Folding)
 
Last edited:
[H]ecklerKoc[H];1036267253 said:
Well,,,, before trying to apply more power.
Click to expand...

mentally pictured you hooking jumper cables up to the board laughing like a deranged lunatic the whole time...
 
NUMA will ahve to wait. logged back on and found 1 day will a P2684 is done. Don't want to risk killing off the bigadv. with this short

Current TPF on the P2684 is 20:36 with a PPD of 65,564.7.

Also dropped in a Zotac 460 during the last rebood cycle. Started it running for while too. (yep, from the deal thread)
Current TPF on the companion GPU P10957 is 1:07 with a PPD of 11,928.4.
Well it was until I I stopped the GPU folder.

Think I like the Paliit 465 better.
 
Last edited:
[H]ecklerKoc[H];1036266166 said:
Baselines so far. Currently on the P6058. P6059 was with the GPU running part of the time. GPU is off as the moment.
...
Click to expand...

Your numbers look very close to mine, and I'm at the same overclock. The GPU was killing some of your times, but if you can manage a larger overall PPD it might be worth running one.

My times before disabling NUMA:

P6059
Min. Time / Frame : 00:01:10 - 74,037.9 PPD
Avg. Time / Frame : 00:01:11 - 72,479.2 PPD

P6071
Min. Time / Frame : 00:01:07 - 79,065.8 PPD
Avg. Time / Frame : 00:01:07 - 79,065.8 PPD

P2684
Min. Time / Frame : 00:18:15 - 78,993.7 PPD
Avg. Time / Frame : 00:18:47 - 75,653.3 PPD

and currently...
P6702
Min. Time / Frame : 00:03:01 - 47,640.9 PPD
Avg. Time / Frame : 00:03:04 - 46,480.5 PPD
:p :( :mad:

My GTX 295s see a little over 5K PPD per P10632. That's a horrible WU.
 
SazanEyes said:
My GTX 295s see a little over 5K PPD per P10632. That's a horrible WU.
Click to expand...
Never seen that WU in my systems and I have a combination of GPU2 and GPU3 clients. All I get are P10109 WUs and they produce anywhere between 4300 and 5100 PPD per GPU on G92 architecture depending on OC. Your card is not seeing anywhere near its full potential if my G92 clients are producing similar results.
 
With P10109, I get from 6600 to 6800 PPD per GPU on the GTX 295s. It's a much better WU. With the best WUs I get around 7600 PPD. I run all my GPUs at standard clock.
 
While waiting for the bigadv to finish here's a little SR-2 Skull Pron for you.

P1100422.JPG


More here if you like:
SR-2
 
Final on the current Project ID: 2684 bench run.
TPF: 00:19:26
71,889.6 PPD

There was some GPU interference in the beginninghttp://blog.tropo.com/2010/10/08/asterisk-tropo-and-a-single-adhearsion-dialplan/ of the run.

Swapping NUMA off now.


Start of the next WU after turning NUMA off:
P6068
TPF: 1:29
PPD: 51,522.9
 
Last edited:
Project ID: 2686

BIOS Setting for NUMA Off


Project ID: 2684
Core: GRO-A3
Credit: 8955
Frames: 100


Name: cpu smp-24
Path: C:\Program Files (x86)\FAH\smp\
Number of Frames Observed: 21

Min. Time / Frame : 00:19:01 - 74,265.2 PPD
Avg. Time / Frame : 00:19:07 - 73,683.2 PPD
Cur. Time / Frame : 00:19:12 - 73,065.0 PPD
R3F. Time / Frame : 00:19:12 - 73,065.0 PPD
All Time / Frame : 00:19:07 - 73,383.5 PPD
Eff. Time / Frame : 00:19:32 - 71,818.2 PPD





Core: GRO-A3
Credit: 8955
Frames: 100


Name: cpu smp-24
Path: C:\Program Files (x86)\FAH\smp\
Number of Frames Observed: 100

Min. Time / Frame : 00:13:33 - 123,474.7 PPD
Avg. Time / Frame : 00:14:50 - 107,802.5 PPD



Project ID: 6070
Core: GRO-A3
Credit: 481
Frames: 100


Name: cpu smp-24
Path: C:\Program Files (x86)\FAH\smp\
Number of Frames Observed: 100

Min. Time / Frame : 00:01:09 - 75,653.2 PPD
Avg. Time / Frame : 00:01:09 - 75,653.2 PPD





CPUs Installed: L5640 ES
BIOS Version: 49
CPU Clock: 3600 MHz
CPU Multiplier: 18.0x
CPU FSB: 200 MHZ
Memory Clock 1000 MHz
Memory Speed: DDR3-2000
Memory Timings: 6-9-6-24 CR1
DRAM FSB Ratio: 5:1
Memory Installed: G.Skill F3-16000CL6T-6GBPID (2GBx3)
2 kits, 6 DIMMs, 12GB total
BIOS Memory Voltage Setting: 1.65 V

 
HK, I'm curious what's the highest OC that you were able to attain on the GTX 295 (shader and core)?
 
You must log in or register to reply here.
Back
Top