There is new driver.

I'll test them. Because 185.66 or 185.66 make always errors.

I hope it's good driver with this project.

it seem to work fine on my sys. with gtx 260.

I test, and I have again the same errors, like 185.66, 185.68.

08/05/2009 19:27:31 GPUGRID Starting task 21-KASHIF_HIVPR_dim_ba2-4-100-RND7086_0 using acemd version 664
08/05/2009 19:28:04 GPUGRID Computation for task 21-KASHIF_HIVPR_dim_ba2-4-100-RND7086_0 finished
08/05/2009 19:28:04 GPUGRID Output file 21-KASHIF_HIVPR_dim_ba2-4-100-RND7086_0_1 for task 21-KASHIF_HIVPR_dim_ba2-4-100-RND7086_0 absent
08/05/2009 19:28:04 GPUGRID Output file 21-KASHIF_HIVPR_dim_ba2-4-100-RND7086_0_2 for task 21-KASHIF_HIVPR_dim_ba2-4-100-RND7086_0 absent
08/05/2009 19:28:04 GPUGRID Output file 21-KASHIF_HIVPR_dim_ba2-4-100-RND7086_0_3 for task 21-KASHIF_HIVPR_dim_ba2-4-100-RND7086_0 absent
08/05/2009 19:28:08 GPUGRID Finished upload of 21-KASHIF_HIVPR_dim_ba2-4-100-RND7086_0_4

<core_client_version>6.6.20</core_client_version>
<![CDATA[
<message>
- exit code 98 (0x62)
</message>
<stderr_txt>
# Using CUDA device 0
# Device 0: "GeForce GTX 260"
# Clock rate: 1404000 kilohertz
# Total amount of global memory: 939524096 bytes
# Number of multiprocessors: 27
# Number of cores: 216
# Device 1: "GeForce 8800 GT"
# Clock rate: 1500000 kilohertz
# Total amount of global memory: 536870912 bytes
# Number of multiprocessors: 14
# Number of cores: 112
# Amber: readparm : Reading parm file parameters
# PARM file in AMBER 7 format
# Encounter 10-12 H-bond term
WARNING: parameters.cu, line 568: Found zero 10-12 H-bond term.
WARNING: parameters.cu, line 568: Found zero 10-12 H-bond term.
MDIO ERROR: cannot open file "restart.coor"
ERROR: c:\cygwin\home\speechserver\gpumd2\src\pme\CPME_cufft.cu, line 104: cufftExecC2R (gridcalc3)
called boinc_finish

</stderr_txt>

I use boinc 6.6.20 on vista 64, and I have 1 GTX260 and one 8800GT.

It works good with ancient driver official 182.50.

Rivatuner cannot detect the 185.85 driver for some people--others it does..go figure. I've lost RT in the 1 rig running 185.85. I've DLed Nvidia System Tools....looks simple enough....can't get the damn thing to work though....:(

EDIT: Seem to have it going atm....pretty whacked, but what else can be expected from Nvidia these days...sheesh.

EDIT 2: Switched to Precision tool...a bit more intuitive...

I have been using Precision for a while, pretty easy although if you set the fan on auto *sometimes* it forgets and temps start to rise. I have not narrowed it down to exactly when, and there certainly is the possibility that it was me that screwed up and not the tool :am:

---

Thanks - Steve

Works good on Win. XP64 with Boinc 6.6.20 with gtx260 and gt9800.

I have been using Precision for a while, pretty easy although if you set the fan on auto *sometimes* it forgets and temps start to rise. I have not narrowed it down to exactly when, and there certainly is the possibility that it was me that screwed up and not the tool :am:

I'm WCed, but it's always good to know when something doesn't work 100% the way it should.....;) One thing that I like about Precision tool is that it will OC all Gpus on startup--that doesn't work with RT....

Works good on Win. XP64 with Boinc 6.6.20 with gtx260 and gt9800.

Vista 64 here...6.6.28, 2x 8800GS

I've just finished my first WU on Windows XP Home with 185.85 WHQL and Boinc 6.6.28 with a GTS250 here (and the most recent beta 185.81 here) and my timestep has dramatically reduced to 47.36ms (42.65ms for the latest beta 185.81), which is the lowest I have ever had.

I'll have to see some other results in the next coming days, but I had been averaging about 65-70ms before on most WU's.

Anyone else seen such improvements?

Rob

185.85, Vista 64, BOINC 6.6.28, GTX295
18 hours GPUGrid, GPUZ, EVGA Precision, no errors.

---

Thanks - Steve

The problems with the units only involve the cards not being a 2XX.
So if your card is based on 9xxx or similar it will probably error out on these famous units.
Seems all 2X5 cards run these units without issues on many different drivers, and in fact the newest drivers will give them some performance gains.
The reason why this units fails seems to be a mistery since some do finish but most will crash on our cards.
I also use this 185.85 driver for some time now and seems darn stable, so for me looks like a winner untill now.

I've just finished my first WU on Windows XP Home with 185.85 WHQL and Boinc 6.6.28 with a GTS250 here (and the most recent beta 185.81 here) and my timestep has dramatically reduced to 47.36ms (42.65ms for the latest beta 185.81), which is the lowest I have ever had.

I'll have to see some other results in the next coming days, but I had been averaging about 65-70ms before on most WU's.

Anyone else seen such improvements?

Rob

The 185.xx drivers offer enhanced crunching/folding performance, but I wasn't aware that there maybe a bit of an increase with these 185.85 drivers. Lately I've been busy with the crashing/hanging WUs, so I haven't had the opportunity to check much else out.

On my i7 rig the 185.xx have been a nightmare. This 185.85 driver runs and the only issue that I am having is it always resets the display to 12??x??? instead of the 1024x768 that I use (still on a 19" LCD here). Still an improvement in performance over the pre-185.xx drivers. I'm just hoping something shows up soon that will run 100% on this rig. My other rigs (775) don't have any issues with 185.xx drivers....

The problems with the units only involve the cards not being a 2XX.
So if your card is based on 9xxx or similar it will probably error out on these famous units.
Seems all 2X5 cards run these units without issues on many different drivers, and in fact the newest drivers will give them some performance gains.
The reason why this units fails seems to be a mistery since some do finish but most will crash on our cards.
I also use this 185.85 driver for some time now and seems darn stable, so for me looks like a winner untill now.

It's not these drivers that are causing the crashing that you are refering to....

It may be the WUs or the insufficient memory of the lower end cards....they don't know, but those WUs are crashing regardless of which driver is used....

.. and my timestep has dramatically reduced to 47.36ms (42.65ms for the latest beta 185.81), which is the lowest I have ever had.

I'll have to see some other results in the next coming days, but I had been averaging about 65-70ms before on most WU's.

WUs with different names and different credits likely feature different complexity (i.e. number of atoms) and differ in the time per step. A bit less than 50 ms used to be pretty normal for my 9800GTX+ last autumn. Now most WUs take longer, but not all of them.

MrS

---

Scanning for our furry friends since Jan 2002

Always errors for me...
http://www.gpugrid.net/results.php?userid=8645

I only have a few lags on my Hauppauge TV card, picture and sound stopped for 2sec after that running in slow motion, not seen this on 182.50.

Anyone else seeing a 2-3% slowdown on v185.85 compared with v185.66?
This is happening on both my ASUS and PNY 9600 GSO cards.

There seems to be a driver issue with old cards and new cuda 2.2 drivers. Maybe an incompatibility with the 2.1 toolkit.

If you have an old card, stay away from new drivers for a little while.

This is unrelated to the problem with KASHIF workunits. The driver problem happens with all workunits.

gdf

Thanks for the information, this may very well explain some of the recently unusually high failure rates.

So which is the last known good one, 182.50?

MrS

---

Scanning for our furry friends since Jan 2002

Thanks for the information, this may very well explain some of the recently unusually high failure rates.

So which is the last known good one, 182.50?

MrS

See this recent post:

http://www.gpugrid.net/forum_thread.php?id=1034&nowrap=true#9655

Failed WUs and he is indeed using driver v182.50...

You're right, that seems like the error. So we'd have to search for an older driver (too bad NV does not just say which CUDA version they support.. or, not that I could find it easily) or just wait it out. Since the project team removed the questionable WUs I didn't get any further errors. But then I only run 2 a day, hardly representative.

MrS

---

Scanning for our furry friends since Jan 2002