Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
Freemo
@freemo:qoto.org
[m]
@CoreRasurae: By the way if your ever looking for remote work, likely using aparapi, let me know.
CoreRasurae
@CoreRasurae
@freemo:qoto.org Sure, I will
CoreRasurae
@CoreRasurae
Hi @freemo:qoto.org @grfrost I found an issue with AMD OpenCL ICD, that can deadlock an application while executing threaded Aparapi applications that concurrently call OpenCL APIs. Namely when compiling a kernel while another is being executed. This does not happen with NVIDIA which has thread safe OpenCL APIs. I have a simple workaround for this that can be inserted into Aparapi. Do you want me include it in the current MR?
@freemo:qoto.org BTW Do you have an expected date for proceeding with the MR analysis?
Probably this should be reported to AMD too.
grfrost
@grfrost

Hey @CoreRasurae regarding possible AMD runtime/driver bug. Can you reproduce using a small C++ example? My guess is that AMD will not 'go the lengths' of installing Aparapi. but if you have a C++ test case they will take a look.

I would be happy to validate too if you can produce a C++ testcase.

WRT MR I will defer to @freemo:qoto.org

CoreRasurae
@CoreRasurae
@grfrost I can try to recreate something like that with a C++ testcase. Let me try if can build it up in the next few days.
CoreRasurae
@CoreRasurae
@grfrost Sorry, I couldn't find the time to create the testcase, not yet...
grfrost
@grfrost
No worries. Busy times. I am expecting a call to be the new UK Prime Minister any time now :)
CoreRasurae
@CoreRasurae
:)
CoreRasurae
@CoreRasurae
@freemo:qoto.org Hi there, I hope all is well. I am writing to get a status update over the Aparapi MR. What is your plan for these MRs?
Freemo
@freemo:qoto.org
[m]
Hi
Forgot about thisa, been busy.. i think IIRC the CI had broken.. i can take a look this week
AS long as the CI fix isnt too crazy I should have time for it
CoreRasurae
@CoreRasurae
@freemo:qoto.org That would be great, as I am finalizing my application and nearing the first release. This would allow me release it against an official Aparapi release. BTW can I publish my code for the application, which will also be open source, on qoto.org?
Freemo
@freemo:qoto.org
[m]
@CoreRasurae: wonderful, i will be sure to work on it for you then... and yes you are most welcome. Anything open source for any purpose is always welcome onn qoto
CoreRasurae
@CoreRasurae
Ok, great!
CoreRasurae
@CoreRasurae
@freemo:qoto.org There is one thing that is not addressed with this MR though, which I found at a later stage, which is that AMD OpenCL ICD drivers do not deal well with concurrent kernel compile and execute calls at least. I haven't had the time to map to actual OpenCL calls, however the OpenCL ICD driver can deadlock, causing the application to stall. This does not happen with NVIDIA OpenCL ICD. This happens on Windows AMD drivers. I wasn't able to test on Linux, since none of my machines with AMD GPUs have working AMD drivers on Linux. I have a simple fix for this issue, which is to synchronize Aparapi-JNI calls for kernel compile and kernel execute. I am not sure if the OpenCL spec says something about the need for the API calls to be thread safe, or not, so this may, or may not be an AMD driver bug. Still, since this issue is already happening with current AMD released Windows drivers, it may be best to protect Aparapi that way. Should I add an MR for this?
Another thing that seem to be happening is that I may have introduced a regression with compiled kernel caching in 3.0.0 that seems that Aparapi is always compiling the same kernel on the same cards, again and again, when called concurrently for the same device from different threads. I have to fix this too, its just that I am lacking the time due to still being quite involved in the PhD... final stages... big boss... :D
Freemo
@freemo:qoto.org
[m]
i have AMD linux and can test here.. do we have any unit tests that that trigger this issue that I can test ?
but yea i think an MR for the fix, if for no other reasont han for me to review it, might be worth while
CoreRasurae
@CoreRasurae
ok, I don't have a unit test, since it would deadlock
unless in case of deadlock if include a test timeout time
*if=we
Freemo
@freemo:qoto.org
[m]
well the point of unit tests is they fail when there is a problem... so we would probably want a timeout and then have the test fail
CoreRasurae
@CoreRasurae
I will try to create a unit test for Aparapi for this issue in the next few days
Freemo
@freemo:qoto.org
[m]
wonderful, thanks
Freemo
@freemo:qoto.org
[m]
I think i finally have a breather, going to work on CI this weekend I think. Sorry for the delays guys
the link https://git.qoto.org/aparapi seems to have some problem, sometime is not responding, is this the main link? thx
Freemo
@freemo:qoto.org
[m]
@marco.stefanetti_gitlab: works for me jsut fine.. but one of our load balancers had bad hardware for a few days.. was fixed ysterday
1 reply
there is a mirror on github also
Bikramjit
@Bikramjit84_twitter
Hi we have micro service running in kubernetes which is listening to kafka topic. Expected to receive million of messages in a day and need to calculate expensive calculation and the results need to be updated in MSSQL. We are planning GPU node in kubernetes and planning to use aparapi. Let me know your thought on this use case and how best we can utilize the GPU so that the processing of Kafka message is near real time
Freemo
@freemo:qoto.org
[m]
Only thought that comes to mind, which you are already aware of id imagine, is you need to batch jobs, cant do them real time (one at a time) on GPU but in groups.. so you need some sort of latency calculator that decides how long its willing to wait and if the size is large enough its worth doing on a GPU but if you need ito get it out to reduce latency in the system but not ehough of a batch has accumulated you'd want to fall back to CPU
Fili-Catalan
@Fili-Catalan
Hello everyone! I'm very new to Aparapi and I'm trying to get it to work on my Mac. I'm using NetBeans to test the GPU capabilities of Aparapi but whenever I try running my test, I keep getting an error saying that the opencl.dll/opencl.so files can't be located. It looks like this error is only relevant to Linux and Windows machines, so I'm trying to figure out how to get Aparapi to recognize/locate OpenCL, since the OpenCL framework does exist within my machine. Is there anything I'm missing? Does anyone know if there's a fix for this for Mac machines? Any help is appreciated.