Message boards :
Number crunching :
Intel GPU app hosts with failures - please post here
Message board moderation
Previous · 1 · 2
Author | Message |
---|---|
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
I now have 8 valid, 13 inconclusive and 4 invalid from the INTEL GPU. Is it worth crunching the remaining WUs (over 50) allocated to this GPU or should I just abort them? He mentioned Dell. Lots of people have said that Dell supplied them with an OpenCL 1.1 OEM driver, haven't provided an OEM 1.2 driver, and have locked it in some way so that the Intel drivers won't install. The strange thing is that the Dell i5 HD 4600 I bought to test a BOINC installation problem came with precisely the right drivers to run the Einstein app (which is fussier than we are) right out of the box. So I can't reproduce whatever the problem is and take it back to Dell. |
MikeN Send message Joined: 24 Jan 11 Posts: 319 Credit: 64,719,409 RAC: 85 |
I now have 8 valid, 13 inconclusive and 4 invalid from the INTEL GPU. Is it worth crunching the remaining WUs (over 50) allocated to this GPU or should I just abort them? Richard is correct, dell appear to have an OEM1.2 driver for their 4600 series processors but not for the 2500 series. I can only assume that this is a ploy to force people to buy their more expensive systems. If anyone knows how to unlock a dell system so that it will download normal intel drivers, please post it as I and many others would like to know and it may solve many of the problems in this thread. Possibly changing bios settings? |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
I now have 8 valid, 13 inconclusive and 4 invalid from the INTEL GPU. Is it worth crunching the remaining WUs (over 50) allocated to this GPU or should I just abort them? Oh sorry I didn't quite understand the locked issue. We recently got a few dozen Dell Optiplex 7010 systems at work. Which looks to be a similar configuration to your systems. Same CPU w/ built in graphics, RAM, & OS at least. However, No add in GPU for them. They are setup on our corporate domain so I have less control than my lab test machines but I did update the them with the latest driver packages from Intel. The Dell packages I download are just a extraction package. They extract to a path such as "C:\Dell\Drivers\1PRMJ" and then run the Intel setup.exe. Which is digitally signed by Intel. So it doesn't look like Dell did anything jiggery pokery to it. Dell may have enabled some of the built in OS features that can prevent some types of drivers to be installed. These should be able to be disabled if you wish. Probably either in the Local Security Policy or Group Policy settings. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
MikeN Send message Joined: 24 Jan 11 Posts: 319 Credit: 64,719,409 RAC: 85 |
Just got my first Intel GPU invalid Not sure if this is relevant, but the majority of the WUs I am crunching are getting validated. The number of invalids is staying low, currently 1 and most invalids are ending up validated. It looks like just a certain type of WU causes problems and most are handled OK. I will let the PC crunch the remaining WUs and see what happens. |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
Just got my first Intel GPU invalid you could try this build instead: https://dl.dropboxusercontent.com/u/60381958/MB7_win_x86_SSE_OpenCL_Intel_r2135.7z SETI apps news We're not gonna fight them. We're gonna transcend them. |
SongBird Send message Joined: 23 Oct 01 Posts: 104 Credit: 164,826,157 RAC: 297 |
You didn't say if you mean MB or AP. Anyway here is a machine with a bunch of intel GPU invalids. I haven't done any investigation as to why they happened. I just removed the offending app. I would still like to crunch on the gpu though... Computer 7019029 straight to the invalids |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
INTEL Intel(R) HD Graphics 4000 (1496MB) OpenCL: 1.01 As discussed OpenCL 1.2 drivers are required. Not because app is OpenCL 1.2 but because Intel was unable to make compliant drivers before last ones, marked as OpenCl 1.2 SETI apps news We're not gonna fight them. We're gonna transcend them. |
MikeN Send message Joined: 24 Jan 11 Posts: 319 Credit: 64,719,409 RAC: 85 |
Just got my first Intel GPU invalid Thanks, will do. Interestingly, the number of invalids spiked yesterday at 9. I wonder if that is because I was using the PC a lot yesterday, Maybe it is fine when just crunching seti on the intel GPU, but not when trying to do other things as well. Anyway, nearly out of my INTEL GPU stash, then will put that processor back to collatz until I can get an openCl 1.2 driver. |
MikeN Send message Joined: 24 Jan 11 Posts: 319 Credit: 64,719,409 RAC: 85 |
Thanks, will do. Interestingly, the number of invalids spiked yesterday at 9. I wonder if that is because I was using the PC a lot yesterday, Maybe it is fine when just crunching seti on the intel GPU, but not when trying to do other things as well. Anyway, nearly out of my INTEL GPU stash, then will put that processor back to collatz until I can get an openCl 1.2 driver. Now that APs are back I got fed up of waiting for these INTEL MB WUs to finish, so I reinstalled lunatics to remove the INTEL apps. As soon as BOINC restarted, they were all aborted and then sent straight back to me as resends but to the Nvidea GPU. If I had known that would happen, I would have done it days ago and got them all crunched by now instead of letting my Nvidea GPU cruch Einstein whilst it waited for some more APs. |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
I've reported the servers misreporting of the OpenCL version to the boinc_dev list. Claggy |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
I've reported the servers misreporting of the OpenCL version to the boinc_dev list. I think the keyword there is 'reporting'. So far as I can tell (and please check my codewalk), actual decision-making on whether the OpenCL version is correct is done with opencl_device_version_int, defined as "100*maj + min" in lib\opencl_boinc.cpp In most cases, minor version numbers are displayed to two significant digits: in that multiplication, "02" will give the same answer as "2". The display format used may confuse us humans, but I don't think it should result in the wrong version being allowed. |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
In most cases, minor version numbers are displayed to two significant digits: in that multiplication, "02" will give the same answer as "2". The display format used may confuse us humans, but I don't think it should result in the wrong version being allowed. Agree, it's just the Human readable display format that is wrong. (as far as I can tell) Claggy |
doug Send message Joined: 10 Jul 09 Posts: 202 Credit: 10,828,067 RAC: 0 |
I see this: ERROR: OpenCL kernel/call 'clGetDeviceIDs' call failed (-1) in file ..\..\..\src\GPU_lock.cpp near line 672. This is a driver call. From the Intel driver or the Nvidia driver? Doug Oh sorry. I was reading from the top of the thread instead of the latest. I guess you already have it figured out. Doug |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.