App for Intel HD graphics?
![Mumps [MM]](https://albertathome.org/sites/default/files/imagefield_default_images/default-avatar.png?1347855344 "Mumps [MM]"){kind=avatar}
15 Nov 2012 23:14:59 UTC
Topic 84901
There are an awful lot of Intel CPUs out there with Intel HD graphics. And with the 4000 version used in ivy bridge, it's actually a fairly powerful GPU. Any chance for an opencl app to use this wasted resource?
App for Intel HD graphics?
Sorry for the delay, this one had escaped my attention.
For this to work "out of the box", BOINC has to incorporate support for Intel GPUs as OpenCL capable co-processor.
Also, there have been numerous reports that Intel's OpsnCL drivers prevent other OpenCL drivers (e.g. AMDs) to work with BOINC, so in order to use e,g, Radeon gfx cards for crunching , volunteers had to uninstall the Intel OpenCL drivers. This would have to be solved as well.
Cheers
HB
RE: RE: There are an
Well the 1st bit is in the alpha-test clients, as in BOINC being able to detect it and schedule tasks to it.
The 2nd part with the Intel OpenCL drivers conflicting, that's still an issue, but only for those systems that have Intel HD graphics and ATI or Nvidia.
I was one of the users who
I was one of the users who had to uninstall the OpenCL part of the Intel graphics driver when I used Boinc 7.0.28 or all CUDA-task would fail.
I'm now running Boinc 7.0.52 where some bugs concerning Intel GPUs have been fixed and can now successfully run work on both my Nvidia card over at Einstein at the same time as I've been testing a beta-app on Seti@home that runs on the Intel HD Graphics 4000. A few problems exist with either the app or Intels drivers as can be seen in this thread over at Seti.
So there should be enough support in the later Boinc-builds to try this.
Hi! I think I can prepare
Hi!
I think I can prepare something to test this next week.
Cheers
HBE
Nice, looking forward to it!
Nice, looking forward to it! =)
Any progress on this? I have
Any progress on this? I have successfully crunched GPU tasks for both collatz and SETI (astropulse), on my intel HD 4000.
Dublin, California
Team: SETI.USA
RE: Any progress on this?
Hi!
No, unfortunately no progress yet. Our BOINC server side code is not yet able to handle Intel GPUs consistently (certainly not on Einstein and not even on Albert IMHO). We (the E@H admins and devs) have quite a lot of other stuff on our desks at the moment (to name just a few: launch of S6directed on Einstein, transition of BRP4 to a CPU only run, evaluating the early BRP5 results, a complete overhaul of the Web Homepage and forum platform, postprocessing of finished GW runs, ....).
Yeah, it's a shame really, especially with the next Intel CPU generation to be launched soon (Haswell) which is said to feature some relatively powerful GPUs. Sigh.
What would be possible at the moment: we provide (in the timeframe of a few days) an app to download that someone could test using an app_info.xml file that would pretend the app is a CPU app. So the Boinc client would ask for work for a CPU anonymous platform task, but the task would then in fact be calculated on the Intel GPU, so one would at least get an idea about the runtime.
Is anyone who is experienced in app_info.xml writing interested in such an experiment?
Cheers
HB
I wouldn't call myself
I wouldn't call myself experienced with app_info, but i'm certainly willing to give it a go if you can post an app for OSX.8. I tried compiling myself a version a few weeks back, generated a binary, but then either it or my attempted app_info caused BOINC 7.0.65 to completely hang when opening.
Machine: https://albertathome.org/host/7157
I have one system with an i3
I have one system with an i3 3220 (Intel Graphic 2500). Will that help? If yes I'll try it.
I'd be willing to give it a
I'd be willing to give it a try on my i7 3770K with HD 4000 graphics.
I'm interested to see if this app might work better than the beta app over at Seti as that one stops processing when I turn of my screen. According to the developer of that app it's a driver related problem but Intel hasn't responded with a fix yet to the bug report on their developers forum.