• 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.

Supercharged P2685?

M

musky

[H]ard|DCer of the Year 2012
Joined
Dec 14, 2009
Messages
3,154
So my SR-2 uploaded a P2685 last night and downloaded another P2685. The client immediately crashed (unfortunately, fairly normal.) I restarted the client, then the machine rebooted (not at all normal.) I restarted the client again and kept an RDC connection open so I could keep an eye on it last night. It ran through the night without further incident. Looking at it this morning, this P2685 is running at 14:18/frame and 112.5K ppd. That is faster than the best times I have seen with P2686s and P2692s by a few seconds, and close to a minute/frame faster than the P2685 I just completed last night. I was getting hopeful that Standford had adjusted these to be closer to the better units until I looked at my 970 machine, which got a P2685 a few hours after my SR-2 did and is processing it at about the same rate as the "normal" P2685s. I must have just gotten a sweetheart of a P2685 unit on the SR-2. I am just hoping for no further crashes.
 
The 2686s I have got this week are about a minute per frame faster than the one I captured and was using for benchmarking from a few weeks ago.

I just had one 2686 frame that took over half an hour (usually 12:15 ish) but I have never had the client crash. How stable are you? Do you test for stability with linpack etc or just find a "folding stable" level?

And now that I am folding fulltime, with a massive 5 days experience, I notice the bigadv units do float in render times of about 50 seconds or so across a WU.

Is there a tool that parses a logfile to generate a list of frame times for the whole 100 frames?
 
The HFM Benchmark tool will get you averages on frame time for all of the workunits you have run. It is basically a log parser I believe.

As far a stable, I can't run LinX at 10Gb, but it is stable with LinX at 768Mb or whatever the default is. It is not as stable as yours, but it is still pretty good. I am sure I have some memory issues somewhere.
 
Not sure what you would want them all for, but the benchmark tool in HFM.net may be right up your ally
Posted via [H] Mobile Device
 
From HFM:

Project ID: 2685
Core: GRO-A3
Credit: 8955
Frames: 100


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

Min. Time / Frame : 00:14:10 - 115,500.9 PPD
Avg. Time / Frame : 00:15:07 - 104,785.9 PPD
Cur. Time / Frame : 00:14:11 - 113,137.3 PPD
R3F. Time / Frame : 00:14:12 - 113,004.5 PPD
All Time / Frame : 00:14:16 - 112,476.4 PPD
Eff. Time / Frame : 00:15:35 - 102,973.1 PPD

Note the current and average differential.
 
Yeah they do look quicker this week. For sure.

Thanks I already use HFM - The reason I want to parse the logfile is that I use this machine during the day to work on, and so my averages are completely useless, as they have been messed up by doing other things. But it would be nice to see by how much and when, so I can work out how to best manage folding in the background. And I have to manage it, because my fricking 3D renderer refuses to grab more than 50% CPU if there is a background process no matter what priority, so I am constantly using winAFC to change affinity.

But these occasional half hour frames bug me. I have only seen 2, but I want to know is it me (overclock problem) or just FAH. So if I had a list of frames times I could spot them much quicker. Might have to have my own script together.
 
musky said:
The client immediately crashed (unfortunately, fairly normal.)
Click to expand...

That doesn't sound "folding stable" to me. You may want to back off the OC a bit, just so you don't have to babysit as much. I've never had a crash or a bad WU on my SR-2 (other than the P2682s that were pulled).
 
SazanEyes said:
That doesn't sound "folding stable" to me. You may want to back off the OC a bit, just so you don't have to babysit as much. I've never had a crash or a bad WU on my SR-2 (other than the P2682s that were pulled).
Click to expand...

Does the same thing at stock speeds...
 
Yep, the one I just finished (Run 2, Clone 16, Gen 27) was averaging 13mins:10sec
The one I had on 15th Sept (Run 0, Clone 16, Gen 26) was averaging 12mins:35sec
 
Sigh... I need to upgrade my SR2 to a proper triple ram setup, on the fscked up dual channel set up I am running, I am getting TPF about a minute and half slower than everyone else at similar speeds.

CPU0 = 2x4GB Dual Channel DDR3 @ 1496 MHz (8GB)
CPU1 = 1x1GB Dual Channel DDR3 @ 1496 MHz (2GB)

Dual X5650's running at 4.1GHz;

Code: 
Project ID: 2686
 Core: GRO-A3
 Credit: 8955
 Frames: 100

 Name: W01-CHAOS
 Path: C:\Windows\SysWOW64\
 Number of Frames Observed: 100

 Min. Time / Frame : 00:14:15 - 114,489.2 PPD
 Avg. Time / Frame : 00:14:35 - 110,586.4 PPD

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

 Name: W01-CHAOS
 Number of Frames Observed: 143

 Min. Time / Frame : 00:18:49 - 75,452.4 PPD
 Avg. Time / Frame : 00:19:37 - 70,548.6 PPD

How do these numbers look compared to your guys results?
 
Yeah, it looks like your RAM - also, what MCH strap are you running at?

I use my SR2 as my main rig, so my averages are messed up, but my minimum frame times seen are in my sig. (now they are best frame times seen in a week, so assume my real average is a bit higher. When not using my machine I guesstimate my average to be about 12:25 for 2686 and 12:40 to 13:10 for 2685)



2685 is generally slower than 2686, but seems more variable.
 
may i be worthy said:
Yeah, it looks like your RAM - also, what MCH strap are you running at?

I use my SR2 as my main rig, so my averages are messed up, but my minimum frame times seen are in my sig. (now they are best frame times seen in a week, so assume my real average is a bit higher. When not using my machine I guesstimate my average to be about 12:25 for 2686 and 12:40 to 13:10 for 2685)



2685 is generally slower than 2686, but seems more variable.
Click to expand...

running it at 2:8(1066MHz in BIOS) with a 1333MHz MCH Strap. T1 timing. I wish I had some proper matched triple channel DDR3 for both the CPU's, even if it was just 6 dimms of matched 1GB to test the memory starvation out.

Before I had it running, same speed and strap, but in single channel mode (1x4GB stick for each CPU) and was getting 50k-60k ppd on 2685, so even just going to the Dual Chanel with horrible mismatched ram did wonders for F@H performance. Just wishing I could afford to get some matched sticks to run it in triple channel to get full performance out of it.
 
You might get a fraction more performance lowering the MCH to 1066 and keeping the timings you have set manually, but of course it might not be stable. But man is bigadv memory dependent!

But I would say look no further, you know what the issue is. Hang in there, hopefully soon funds will become available. :)
 
I got another one of these wonderful P2685s on the SR-2 - 14:30/frame, 111.5K ppd. That is 10 seconds/frame faster than the 2686 this thing just ran. I'll take all of these Stanford is willing to send out... :)
 
musky said:
I got another one of these wonderful P2685s on the SR-2 - 14:30/frame, 111.5K ppd. That is 10 seconds/frame faster than the 2686 this thing just ran. I'll take all of these Stanford is willing to send out... :)
Click to expand...

Quick - ebay it before the deadline!

watch.jpg
 
You must log in or register to reply here.
Back
Top