My iGPU WU's have been running quite smoothly (15min/WU, not bad); just this little snippet seems a bit odd in the stdout, but it doesnt seem to affect the WU completing successfully.
[09:49:30][4088][ERROR] Error in OpenCL context: Out of device memory.
[09:49:30][4088][ERROR] Error in OpenCL context: Out of device memory.
This seems consistant with other host's outputs as well, so it's probablly nothing.
[09:49:30][4088][ERROR] Error in OpenCL context: Out of device memory.
[09:49:30][4088][ERROR] Error in OpenCL context: Out of device memory.
Video memory is allocated dynamically, but has an upper limit which can be set up in the bios. You should set the limit above 800MB, better 1 GB if you have enought RAM installed.
Video memory is allocated dynamically, but has an upper limit which can be set up in the bios. You should set the limit above 800MB, better 1 GB if you have enought RAM installed.
Well since this is running on a macbook, I doubt I can do that. Not only is there no BIOS, but you can bet Apple locked the EFI.
It doesnt seem to affect the app though; it allocates ~400mb in memory according to task manager as well as dynamic memory in GPU-Z. My guess is it's a warning more so than an actual error.
I am getting Computation Error -1073741819 (0xffffffffc0000005) with the Binary Radio Pulsar Search (Arecibo) v1.34 (opencl-intel_gpu) application.
(example task# 421200278)
According to the stderr output the failure is caused by an access violation.
I have just enabled my IntelHD 4000 on-board GPU and upgraded to an OpenCL 1.2 driver so it is the first time I have received opencl workunits.
I also have an Nvidia GTX660 and the cuda application works well.
(Computer ID# 5512961, windows 7 x64)
I have extended the desktop to the Intel display (I did need to plug in a monitor to do this) and have had success running the seti@home V7 (opencl_intel_sah) application.
Note: Once the desktop was extended to the Intel I was able to remove the secondary monitor. The Intel GPU remained active even after restarting the boinc science applications.
I will disable intel GPU in the project settings for now.
There seems to be some problem with the 10-series Intel GPU drivers and running Einstein, try to downgrade the driver to the latest in the 9.xx series.
can someone tell me, is it ok for intel GPU WUs to take a slot from CPU slots?
When I start the intel GPU WU, one of the running CPU WU will be suspended.
Is there a way for me to use an extra slot for intel GPU WUs hence the CPU slots are not affected?
Thanks in advance.
====
Suddenly I find out that the two running GPU WUs apply more than one CPU thread.
The one from this project (Intel GPU) takes 0.5, and the other one (DistrRTgen, Nvidia GPU) takes 0.679.
If I suspend any of these two WUs, one CPU slot will be freed.
However, If I choose to run two the "System Idle Process" shows there's about 10% not applied. I think it's ok for me to run one more CPU WU? How to configure it?
can someone tell me, is it ok for intel GPU WUs to take a slot from CPU slots?
When I start the intel GPU WU, one of the running CPU WU will be suspended.
Is there a way for me to use an extra slot for intel GPU WUs hence the CPU slots are not affected?
Thanks in advance.
This behavior is normal. The task needs 0.5 CPUs + 1 Intel GPU. This is why one of your cores cannot process CPU tasks anymore. There is no way to run the workunit only on the GPU, since some of the calculations need to be done on the CPU. If you dig a bit deeper in the forum you will find more detailed explanations for that.
So it is a driver issue,
)
So it is a driver issue, picky app eh? haha. Can't accept anything less than 9.18 but nothing higher than 9.18 either. Thanks Alex.
Picky app or sh*tty driver -
)
Picky app or sh*tty driver - who knows?
MrS
Scanning for our furry friends since Jan 2002
Drivers indeed... My iGPU
)
Drivers indeed...
My iGPU WU's have been running quite smoothly (15min/WU, not bad); just this little snippet seems a bit odd in the stdout, but it doesnt seem to affect the WU completing successfully.
This seems consistant with other host's outputs as well, so it's probablly nothing.
RE: [09:49:30][4088][ERROR]
)
Video memory is allocated dynamically, but has an upper limit which can be set up in the bios. You should set the limit above 800MB, better 1 GB if you have enought RAM installed.
RE: Video memory is
)
Well since this is running on a macbook, I doubt I can do that. Not only is there no BIOS, but you can bet Apple locked the EFI.
It doesnt seem to affect the app though; it allocates ~400mb in memory according to task manager as well as dynamic memory in GPU-Z. My guess is it's a warning more so than an actual error.
I couldn't set this much
)
I couldn't set this much memory in my Gigabyte UEFI either.
MrS
Scanning for our furry friends since Jan 2002
Hi everyone, I am getting
)
Hi everyone,
I am getting Computation Error -1073741819 (0xffffffffc0000005) with the Binary Radio Pulsar Search (Arecibo) v1.34 (opencl-intel_gpu) application.
(example task# 421200278)
According to the stderr output the failure is caused by an access violation.
I have just enabled my IntelHD 4000 on-board GPU and upgraded to an OpenCL 1.2 driver so it is the first time I have received opencl workunits.
I also have an Nvidia GTX660 and the cuda application works well.
(Computer ID# 5512961, windows 7 x64)
I have extended the desktop to the Intel display (I did need to plug in a monitor to do this) and have had success running the seti@home V7 (opencl_intel_sah) application.
Note: Once the desktop was extended to the Intel I was able to remove the secondary monitor. The Intel GPU remained active even after restarting the boinc science applications.
I will disable intel GPU in the project settings for now.
Can anyone suggest a remedy to this problem?
There seems to be some
)
There seems to be some problem with the 10-series Intel GPU drivers and running Einstein, try to downgrade the driver to the latest in the 9.xx series.
See this thread for more info
Hi, can someone tell me,
)
Hi,
can someone tell me, is it ok for intel GPU WUs to take a slot from CPU slots?
When I start the intel GPU WU, one of the running CPU WU will be suspended.
Is there a way for me to use an extra slot for intel GPU WUs hence the CPU slots are not affected?
Thanks in advance.
====
Suddenly I find out that the two running GPU WUs apply more than one CPU thread.
The one from this project (Intel GPU) takes 0.5, and the other one (DistrRTgen, Nvidia GPU) takes 0.679.
If I suspend any of these two WUs, one CPU slot will be freed.
However, If I choose to run two the "System Idle Process" shows there's about 10% not applied. I think it's ok for me to run one more CPU WU? How to configure it?
Thanks.
RE: Hi, can someone tell
)
This behavior is normal. The task needs 0.5 CPUs + 1 Intel GPU. This is why one of your cores cannot process CPU tasks anymore. There is no way to run the workunit only on the GPU, since some of the calculations need to be done on the CPU. If you dig a bit deeper in the forum you will find more detailed explanations for that.