Well i have received a test unit so we gonna see if the driver is the issue.
I had rebooted and was to be honest not expecting to see gpugrid downloaded new units, but since it did i thought ill let it finish and see what happens.
Those 2 i got both with ended without issue
Lets see if this test errors out.
Still want to remind you guys that, according to boinc my card seems only to be able todo 1.0 Cuda, what strikes me is why nvidia does not inform which cards benefit from this new cuda 2.X versions or is it again boinc which does not see the right cuda version.

Well i have received a test unit so we gonna see if the driver is the issue.
I had rebooted and was to be honest not expecting to see gpugrid downloaded new units, but since it did i thought ill let it finish and see what happens.
Those 2 i got both with ended without issue
Lets see if this test errors out.
Still want to remind you guys that, according to boinc my card seems only to be able todo 1.0 Cuda, what strikes me is why nvidia does not inform which cards benefit from this new cuda 2.X versions or is it again boinc which does not see the right cuda version.

The version reported by BOINC is the compute capability of the card, not the cuda version. They have reworded it from 6.6.24 onwards. What sort of card is it?

---

BOINC blog

9600 GT haven't found info anywhere what version of cuda is supported by it.

Hardware capabilities should be, right out of my head:

G80 - 1.0
>G80 & <G200 - 1.1
G200 - 1.3

MrS

---

Scanning for our furry friends since Jan 2002

Yea i forgot indeed is 1.1

Sadly many hours lost again with broken units, rebooted back into win XP

I reinstalled old driver set 182.50 and reversed back to cuda 2.1 with 6.6.28
if i probably change back to 6.5.0 but ill try this a few days more because its time estimates really are ok.

Now back to sleep ill see tommorow if the units still fail

You might want to go into the Nvida control panel -> Help -> System information and just check what version it thinks it is. Apparently when I upgraded, while it went through the motions and rebooted it still thought it had the old drivers.

To do the upgrade properly I had to uninstall the old drivers and then install new ones before it came up as 185.85 drivers.

---

BOINC blog

Whell, I was using 182.08 and it works grat.

I have 6.6.20 and I upgrade to 6.6.28.

You say that 185.85 works on GTX26x and up, so I try and:


<core_client_version>6.6.28</core_client_version>
<![CDATA[
<message>
- exit code 98 (0x62)
</message>
<stderr_txt>
# Using CUDA device 0
# Device 0: "GeForce GTX 260"
# Clock rate: 1242000 kilohertz
# Total amount of global memory: 939196416 bytes
# Number of multiprocessors: 24
# Number of cores: 192
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>
]]>

next:

<core_client_version>6.6.28</core_client_version>
<![CDATA[
<message>
- exit code 98 (0x62)
</message>
<stderr_txt>
# Using CUDA device 0
# Device 0: "GeForce GTX 260"
# Clock rate: 1242000 kilohertz
# Total amount of global memory: 939196416 bytes
# Number of multiprocessors: 24
# Number of cores: 192
# 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 50: cufftExecC2C (gridcalc2.1)
called boinc_finish

</stderr_txt>
]]>

next:

<core_client_version>6.6.28</core_client_version>
<![CDATA[
<message>
Niepoprawna funkcja. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
# Using CUDA device 0
# Device 0: "GeForce GTX 260"
# Clock rate: 1242000 kilohertz
# Total amount of global memory: 939196416 bytes
# Number of multiprocessors: 24
# Number of cores: 192
MDIO ERROR: cannot open file "restart.coor"
# Using CUDA device 0
# Device 0: "GeForce GTX 260"
# Clock rate: 1242000 kilohertz
# Total amount of global memory: 939196416 bytes
# Number of multiprocessors: 24
# Number of cores: 192
MDIO ERROR: cannot open file "restart.coor"
Cuda error: Kernel [langevin_thermostat] failed in file 'thermostat.cu' in line 34 : unspecified launch failure.

</stderr_txt>
]]>

<core_client_version>6.6.28</core_client_version>
<![CDATA[
<message>
- exit code 98 (0x62)
</message>
<stderr_txt>
# Using CUDA device 0
# Device 0: "GeForce GTX 260"
# Clock rate: 1242000 kilohertz
# Total amount of global memory: 939196416 bytes
# Number of multiprocessors: 24
# Number of cores: 192
MDIO ERROR: cannot open file "restart.coor"
# Using CUDA device 0
# Device 0: "GeForce GTX 260"
# Clock rate: 1242000 kilohertz
# Total amount of global memory: 939196416 bytes
# Number of multiprocessors: 24
# Number of cores: 192
# Using CUDA device 0
# Device 0: "GeForce GTX 260"
# Clock rate: 1404000 kilohertz
# Total amount of global memory: 939196416 bytes
# Number of multiprocessors: 24
# Number of cores: 192
# Using CUDA device 0
# Device 0: "GeForce GTX 260"
# Clock rate: 1404000 kilohertz
# Total amount of global memory: 939196416 bytes
# Number of multiprocessors: 24
# Number of cores: 192
# Using CUDA device 0
# Device 0: "GeForce GTX 260"
# Clock rate: 1242000 kilohertz
# Total amount of global memory: 939196416 bytes
# Number of multiprocessors: 24
# Number of cores: 192
# Using CUDA device 0
# Device 0: "GeForce GTX 260"
# Clock rate: 1242000 kilohertz
# Total amount of global memory: 939196416 bytes
# Number of multiprocessors: 24
# Number of cores: 192
ERROR: c:\cygwin\home\speechserver\gpumd2\src\pme\CPME_cufft.cu, line 104: cufftExecC2R (gridcalc3)
called boinc_finish

</stderr_txt>
]]>

So I will try now 182.50 - like testing :) If it also fails I will go back to 182.08...

---

POLISH NATIONAL TEAM - Join! Crunch! Win!

I come back with new problems :
Remember my host : 8400 + 1 GTX 260 + 1 8800 GT on vista 64.
It work good with boinc 6.6.2 and Nvidia 182.5.

I upgrade boinc to 6.6.28
First i have :

22/05/2009 01:17:16 Starting BOINC client version 6.6.28 for windows_x86_64
22/05/2009 01:17:16 Processor: 2 GenuineIntel Intel(R) Core(TM)2 Duo CPU E8400 @ 3.00GHz [Intel64 Family 6 Model 23 Stepping 6]
22/05/2009 01:17:16 OS: Microsoft Windows Vista: Home Premium x64 Edition, Service Pack 1, (06.00.6001.00)
22/05/2009 01:17:16 Memory: 4.00 GB physical, 8.17 GB virtual
22/05/2009 01:17:16 CUDA device: GeForce 8800 GT (driver version 18585, compute capability 1.1, 512MB, est. 60GFLOPS)
22/05/2009 01:17:16 CUDA device (not used): GeForce 8800 GT (driver version 18585, compute capability 1.1, 512MB, est. 60GFLOPS)
22/05/2009 01:17:16 Not using a proxy

The GTX260 don't appear (and only one thread run)
and all my wus go in errors like

22/05/2009 09:42:02 GPUGRID Computation for task p1270000-IBUCH_pYIpYVk1_2105-0-10-RND0895_1 finished
22/05/2009 09:42:02 GPUGRID Output file p1270000-IBUCH_pYIpYVk1_2105-0-10-RND0895_1_1 for task p1270000-IBUCH_pYIpYVk1_2105-0-10-RND0895_1 absent
22/05/2009 09:42:02 GPUGRID Output file p1270000-IBUCH_pYIpYVk1_2105-0-10-RND0895_1_2 for task p1270000-IBUCH_pYIpYVk1_2105-0-10-RND0895_1 absent
22/05/2009 09:42:02 GPUGRID Output file p1270000-IBUCH_pYIpYVk1_2105-0-10-RND0895_1_3 for task p1270000-IBUCH_pYIpYVk1_2105-0-10-RND0895_1 absent

Then I upgrade driver Nvidia to 185.85, and I have still the pb.

- you used 6.6.20, not 6.6.2 ;)

- the messages you quote say nothing, you have to look into the individual task outputs (e.g. here) .. but currently this probably won't tell you much more

- regarding the cards being used: in the newer clients you have to add an option in the cc_config if you want all GPUs being used.. sorry, don't know the exact details

- one of your WUs finished with 6.6.28, but I don't know why you suddenly get so many errors. The WUs run on GTX 260 and 8800GT, despite of the 260 missing in the messages.

MrS

---

Scanning for our furry friends since Jan 2002

- regarding the cards being used: in the newer clients you have to add an option in the cc_config if you want all GPUs being used.. sorry, don't know the exact details

MrS

That would be

<cc_config>
<options>
<use_all_gpus>1</use_all_gpus>
</options>
</cc_config>

---

pixelicious.at - my little photoblog

Thanks, but i don't find a cc_config. :/

I notice an another bug :
I work for an other project (µfluid) and Gpugrid stay in waiting.
So with boinc 6.6.20, the 2 projects work in same time.
2 threads for µfluid, and 2 threads for GPUgridµ.

You have to create it yourself...just text, but save it as...

cc_config.xml

It goes in the Boinc folder....

---

Consciousness: That annoying time between naps......

Experience is a wonderful thing: it enables you to recognize a mistake every time you repeat it.

You have to create it yourself...just text, but save it as...

cc_config.xml

It goes in the Boinc folder....

Yes it does ...

and Gpugrid stay in waiting.

What's the exact message why it is waiting?

MrS

---

Scanning for our furry friends since Jan 2002

Look :


with boinc 6.6.20, the four wus run in same time.

I find the solution :
in advanced option : Use GPU while computer in use.


But i run with 182.5. With 185.85 I still have errors.

In the sticky thread regarding this "problem" paul says that from 6.6.20 up you have to manually switch this settign (if you want you GPU to run when you work).. so I assumed you already knew about this, since you've been using 6.6.20. Was the change introduced later on?

MrS

---

Scanning for our furry friends since Jan 2002

In the sticky thread regarding this "problem" paul says that from 6.6.20 up you have to manually switch this settign (if you want you GPU to run when you work).. so I assumed you already knew about this, since you've been using 6.6.20. Was the change introduced later on?

MrS

In 6.6.20 I don't see this message.

Yes, but did you have to set the preference manually? Is it there at all? (sorry, I'd take a look myself if I'd use 6.6.20)

MrS

---

Scanning for our furry friends since Jan 2002

on this version their are 2 really nasty bugs
once affect display settiing and another wont let you access the control for nivida