Announcing project 5902 (Core 14) on advmethods

[ihaque]

Project 5902 just went up on advmethods. It's similar to P5900, but with some tweaks that should improve WU stability. Like P5900, this project runs on core 14. P5902 is a duplicate of P5760, the protein L folding simulations, intended to validate Core 14. P5902 runs are 4x longer than P5900 runs, and points have been bumped in proportion.

Please note, the behavior of this core is different from what you may expect. The following behaviors are NORMAL, and the points credit takes it into consideration:

• Varying time per frame
• Low GPU temperatures
• Fluctuating CPU usage (somewhat higher than GPU core 11).

WU's are worth 1680 points, with a preferred deadline of 3 days and final deadline of 4 days.

[Magic Michael]

Got it on both my 8800GTs (wine/Linux). Fahmon freezes when trying to load the project data, does anybody else experience this too ? I'm curious about ppd for these little s*****s.

[tbk-aracthebold]

Hi,

I just got one it took 2 tries to get fahmon to load the data. Showing 3162 ppd with 12h21m left to fold. I'm using a 9600 gt top

thanks

[soa_rru]

8800gt

Project : 5902
Core : Unknown
Frames : 100
Credit : 1680


-- Nvidia GPU --

Min. Time / Frame : 5mn 42s - 4244.21 ppd
Avg. Time / Frame : 6mn 16s - 3860.43 ppd
Cur. Time / Frame : 6mn 41s - 3619.75 ppd
R3F. Time / Frame : 6mn 41s - 3619.75 ppd
Eff. Time / Frame : 6mn 30s - 3721.85 ppd

[Gadreel]

GTX295

Project : 5902
Credit : 1680
-- Neo GPU2 --
Min. Time / Frame : 3mn 14s - 7482.06 ppd
Avg. Time / Frame : 3mn 15s - 7443.69 ppd
Cur. Time / Frame : 3mn 17s - 7368.12 ppd
R3F. Time / Frame : 3mn 15s - 7443.69 ppd
Eff. Time / Frame : 3mn 20s - 7257.60 ppd

[jorgelrod]

Got it on my two clients(295GTX) Fahmon estimates 5 hours per WU. Not bad for 1680 pp/wu

Project : 5902
Core : Unknown
Frames : 100
Credit : 1680


-- GPU#1 --

Min. Time / Frame : 2mn 52s - 8439.07 ppd
Avg. Time / Frame : 2mn 54s - 8342.07 ppd
Cur. Time / Frame : 3mn 03s - 7931.80 ppd
R3F. Time / Frame : 3mn 03s - 7931.80 ppd
Eff. Time / Frame : 3mn 00s - 8064.00 ppd


-- GPU#2 --

Min. Time / Frame : 3mn 17s - 7368.12 ppd
Avg. Time / Frame : 3mn 19s - 7294.07 ppd
Cur. Time / Frame : 3mn 25s - 7080.59 ppd
R3F. Time / Frame : 3mn 21s - 7221.49 ppd
Eff. Time / Frame : 3mn 21s - 7221.49 ppd

Jorge

[Slash_2CPU]

9800GT 65nm 633MHz core, 918MHz RAM, 1836MHz shaders:
This card is running at PCIe 1.1 x4. Seems x4 is still plenty adequate for the incoming WU's.

Min. Time / Frame : 5mn 29s - 4411.91 ppd
Avg. Time / Frame : 5mn 32s - 4372.05 ppd
Cur. Time / Frame : 5mn 36s - 4320.00 ppd
R3F. Time / Frame : 5mn 32s - 4372.05 ppd
Eff. Time / Frame : 5mn 40s - 4269.18 ppd

[Birdman86]

I haven't had any issues with core 14 so far, I've completed over 20 project 5900 WU's and now got one of project 5902. I just wonder why the 5902 WU seems to be slower.
I've got Q6600 CPU @ 2.4 GHz, 2x GeForce 8800 GT's (GPU0: 600/1651/900 MHz, GPU1: 650/1692/900 MHz), x64 Vista with 4 GB ram.

Project : 5900
Core : Unknown
Frames : 100
Credit : 420

-- GF8800GT gpu0 --
Min. Time / Frame : 41s - 8850.73 ppd
Avg. Time / Frame : 51s - 7115.29 ppd
Cur. Time / Frame : 1mn 02s - 5852.90 ppd
R3F. Time / Frame : 1mn 01s - 5948.85 ppd
Eff. Time / Frame : 1mn 08s - 5336.47 ppd

-- GF8800GT gpu1 --
Min. Time / Frame : 51s - 7115.29 ppd
Avg. Time / Frame : 54s - 6720.00 ppd
Cur. Time / Frame : 58s - 6256.55 ppd
R3F. Time / Frame : 58s - 6256.55 ppd
Eff. Time / Frame : 1mn 12s - 5040.00 ppd


Project : 5902
Core : Unknown
Frames : 100
Credit : 1680

-- GF8800GT gpu0 --
Min. Time / Frame : 5mn 23s - 4493.87 ppd
Avg. Time / Frame : 5mn 41s - 4256.66 ppd
Cur. Time / Frame : 5mn 44s - 4219.53 ppd
R3F. Time / Frame : 5mn 37s - 4307.18 ppd
Eff. Time / Frame : 5mn 48s - 4171.03 ppd

-- GF8800GT gpu1 --
Min. Time / Frame : 3mn 24s - 7115.29 ppd
Avg. Time / Frame : 4mn 27s - 5436.40 ppd
Cur. Time / Frame : 3mn 58s - 6098.82 ppd
R3F. Time / Frame : 3mn 56s - 6150.51 ppd
Eff. Time / Frame : 4mn 37s - 5240.14 ppd

Are the stability tweaks supposed to cause lower PPD?

Edit: Seems that the WU is accelerating now with 37% done, and my other GPU also got one of those and is at 15%. Maybe I just posted my PPD too early.

[Eveofwar]

On my 9800GT, pretty good. About the same as the 5900 projects.

Project : 5902
Core : Unknown
Frames : 100
Credit : 1680


-- MAIN-GPU --

Min. Time / Frame : 5mn 40s - 4269.18 ppd
Avg. Time / Frame : 5mn 45s - 4207.30 ppd
Cur. Time / Frame : 5mn 56s - 4077.30 ppd
R3F. Time / Frame : 5mn 49s - 4159.08 ppd
Eff. Time / Frame : 5mn 52s - 4123.64 ppd

[MichaelB]

Project : 5902
Core : Unknown
Frames : 100
Credit : 1680


-- 9800GTX+ --

Min. Time / Frame : 4mn 30s - 5376.00 ppd
Avg. Time / Frame : 4mn 39s - 5202.58 ppd
Cur. Time / Frame : 4mn 30s - 5376.00 ppd
R3F. Time / Frame : 4mn 37s - 5240.14 ppd
Eff. Time / Frame : 4mn 40s - 5184.00 ppd

-- Quad 8800GT --

Min. Time / Frame : 5mn 24s - 4480.00 ppd
Avg. Time / Frame : 5mn 56s - 4077.30 ppd
Cur. Time / Frame : 5mn 41s - 4256.66 ppd
R3F. Time / Frame : 5mn 37s - 4307.18 ppd
Eff. Time / Frame : 5mn 51s - 4135.38 ppd

Can't see any difference in performance between 5900 and 5902

[^w^ing]

Damn lags, I dont mind the lags with core 11 where only web browsing is somewhat crippled and some other minor slow-down effects (but most importantly, video playback wasnt stuttering at all), and I can easily avoid any lags by turning off hardware acceleration, but with core 14 and these WUs, video stutters even with turned off hw acceleration.

[brentpresley]

PPD is about 1/2 what it was for WU 5900 here on 5 9800GX2s and 2 9800GTXs.


EDIT - can someone explain to me why one of these gives me 2500PPD on one GPU of a GX2, and another 5902 WU gives me 4200 PPD?

[P5-133XL]

PPD is about 1/2 what it was for WU 5900 here on 5 9800GX2s and 2 9800GTXs.


EDIT - can someone explain to me why one of these gives me 2500PPD on one GPU of a GX2, and another 5902 WU gives me 4200 PPD?

The likely problem is that you are using FahMon using the three-frame measurement setup. With the 5900/5901/5902's the variablity between individual frame times is much higher than with other projects so that if you are measuring PPD using just a few frames you can expect much more variability in PPD's. Rather than using 3-frame try using the whole WU measuring setting.

[brentpresley]

You are incorrect. I am using FAHMon to calculate PPD from the entire simulation, not just the last 3 frames.

[werty316]

Regardless of whichever option a user chooses to monitor their PPD, each value of each setting can be viewed in the benchmarks screen of Fahmon.

Avg. Time = All frames
Cur. Time = Last frame only
R3F. Time = Last 3 frames
Eff. Time = Effective rate

If you are running a CPU client as well as the GPU client on the same machine be sure to recheck the priority of each core.