I've recently had almost all of my SET@home work units for the Intel onboard GPU fail with computation errors.

BOINC, seti@home, graphics drivers, windows etc are all up to date. I don't see this problem on other projects that also use the Intel GPU such as Eintein@home.

I don;t have a discrete graphics card. Just the Haswell CPU and it's onboard Intel HD 4600

---

Better to post a thread over here in number crunching

http://setiathome.berkeley.edu/forum_forum.php?id=10

Lot more users there and where they discuss a lot of the problems with the new applications.

exit code -1073741819 (0xc0000005) is a classic. And one difficult to trace. In this case I'd gamble on a problem with the drivers, or incompatibility between application and the GPU.

That Einstein runs normal cannot be compared, their applications are differently built, against an older BOINC API, and do something different than Seti's applications. About the only similarity between them is that they're both written in C++.

You may want to try to run the anonymous applications from Lunatics instead, as these may be better suited for the Intel GPU, being specifically built for them. Or you may want to private message Richard Haselgrove, as he has exactly the same system and runs without trouble. But then he does run Lunatics apps. In any case, he may tell you that the drivers you have are wrong, it happened before.

---

You are not alone, it drives me crazy, only compute errors with my new Intel Core i5-4590S on a Z97 chipset motherboard.

I could't find any solution for this problem so I removed Seti@Home from Boinc and stopped crunching on that pc.

With two posts under your belt, one done in 2009 and this one, you didn't ask for much help either. The solution really is to use the 3rd party Lunatics applications as these as purpose built for newer hardware, whereas the application released through Seti are of a more generic kind, that needs to be able to run on whatever hardware out there. And then for some this gives problems.

Just running away isn't exactly helping.
Why not read through Lunatics Windows Installer v0.43 Release Notes for a bit?

---

With two posts under your belt, one done in 2009 and this one, you didn't ask for much help either. The solution really is to use the 3rd party Lunatics applications as these as purpose built for newer hardware, whereas the application released through Seti are of a more generic kind, that needs to be able to run on whatever hardware out there. And then for some this gives problems.

Just running away isn't exactly helping.
Why not read through Lunatics Windows Installer v0.43 Release Notes for a bit?

Haha...two posts under my belt, thanks for the reminder. Since when is that of any value ?

To answer why, I never had any problems with the standard application until my new Haswell build from last week. After a few searches I found this thread and see what it brought a third post in 14 years, isn't that great ? haha.

I will check out the Lunatics application, thank you.. Oh before I forget to mention, same errors occure with Rosetta@Home.

I have decided not to use the lunatics application, I don't want third party software on my system.
The developers of the Boinc project has to provide us with a client that works with the latest generation of Intel processors.
Until this is resolved no new workunits for my my new Intel Core i5-4590S system.
If the problem "compute error" comes from somewhere else then the Intel platform and this version of boinc, I would like to hear it, thanks.

PS. Jord be careful with your statements, you might scare off people. It isn't important how many posts users have in a forum.

exit code -1073741819 (0xc0000005) is a classic. And one difficult to trace. In this case I'd gamble on a problem with the drivers, or incompatibility between application and the GPU.

That Einstein runs normal cannot be compared, their applications are differently built, against an older BOINC API, and do something different than Seti's applications. About the only similarity between them is that they're both written in C++.

You may want to try to run the anonymous applications from Lunatics instead, as these may be better suited for the Intel GPU, being specifically built for them. Or you may want to private message Richard Haselgrove, as he has exactly the same system and runs without trouble. But then he does run Lunatics apps. In any case, he may tell you that the drivers you have are wrong, it happened before.

There are now reports that the Einstein@Home project application for Intel GPU has problems with an Intel driver released this month (October 2014): A lot of invalid results (opencl-intel_gpu).

The Intel GPU application supplied as a 'BOINC application' by this SETI@Home project was written and supplied by the same third-party programming team as the applications packaged in the Lunatics Installer. All that SETI has done is to test it for safety and endorsed it with their digital certificate: there should be no difference in compatibility.

I would recommend testing the system using Intel driver version 10.18.10.3621, released May 21 2014. Problems were reported with the initial August release, and I believe it was withdrawn for a while: the same thing may be happening now. I'm having very few problems with 3621 - on two iGPUs from different generations (HD 4000 and 4600) and two different BOINC projects.

exit code -1073741819 (0xc0000005) is a classic. And one difficult to trace. In this case I'd gamble on a problem with the drivers, or incompatibility between application and the GPU.

That Einstein runs normal cannot be compared, their applications are differently built, against an older BOINC API, and do something different than Seti's applications. About the only similarity between them is that they're both written in C++.

You may want to try to run the anonymous applications from Lunatics instead, as these may be better suited for the Intel GPU, being specifically built for them. Or you may want to private message Richard Haselgrove, as he has exactly the same system and runs without trouble. But then he does run Lunatics apps. In any case, he may tell you that the drivers you have are wrong, it happened before.

There are now reports that the Einstein@Home project application for Intel GPU has problems with an Intel driver released this month (October 2014): A lot of invalid results (opencl-intel_gpu).

The Intel GPU application supplied as a 'BOINC application' by this SETI@Home project was written and supplied by the same third-party programming team as the applications packaged in the Lunatics Installer. All that SETI has done is to test it for safety and endorsed it with their digital certificate: there should be no difference in compatibility.

I would recommend testing the system using Intel driver version 10.18.10.3621, released May 21 2014. Problems were reported with the initial August release, and I believe it was withdrawn for a while: the same thing may be happening now. I'm having very few problems with 3621 - on two iGPUs from different generations (HD 4000 and 4600) and two different BOINC projects.

To be on the safe side, I am talking about the CPU and not GPU, I don't use the internal GPU for Boinc.

To be on the safe side, I am talking about the CPU and not GPU, I don't use the internal GPU for Boinc.

I think perhaps the title of the thread may have confused people.

The original poster was talking about his Intel GPU, not CPU.

Might be better to ask you question in the "number crunching" forum

I am not sure you should be getting errors with that CPU, but the guys at NC will know.

---

exit code -1073741819 (0xc0000005) is a classic. And one difficult to trace. In this case I'd gamble on a problem with the drivers, or incompatibility between application and the GPU.

That Einstein runs normal cannot be compared, their applications are differently built, against an older BOINC API, and do something different than Seti's applications. About the only similarity between them is that they're both written in C++.

You may want to try to run the anonymous applications from Lunatics instead, as these may be better suited for the Intel GPU, being specifically built for them. Or you may want to private message Richard Haselgrove, as he has exactly the same system and runs without trouble. But then he does run Lunatics apps. In any case, he may tell you that the drivers you have are wrong, it happened before.

There are now reports that the Einstein@Home project application for Intel GPU has problems with an Intel driver released this month (October 2014): A lot of invalid results (opencl-intel_gpu).

The Intel GPU application supplied as a 'BOINC application' by this SETI@Home project was written and supplied by the same third-party programming team as the applications packaged in the Lunatics Installer. All that SETI has done is to test it for safety and endorsed it with their digital certificate: there should be no difference in compatibility.

I would recommend testing the system using Intel driver version 10.18.10.3621, released May 21 2014. Problems were reported with the initial August release, and I believe it was withdrawn for a while: the same thing may be happening now. I'm having very few problems with 3621 - on two iGPUs from different generations (HD 4000 and 4600) and two different BOINC projects.

To be on the safe side, I am talking about the CPU and not GPU, I don't use the internal GPU for Boinc.

Jord's post, and this whole discussion thread, were specifically concerned with the internal, integral, iGPU component of the CPU.

The project has tried-and-tested applications which run on the traditional CPU cores of your Intel Core i5-4590S. Nothing you have read in this thread need put you off using those traditional applications - you can exclude the iGPU if you wish, by using the controls on the SETI@home preferences page for your account.

I have decided not to use the lunatics application, I don't want third party software on my system.

Haha (as you say) - for your info: the GPU apps used now as stock are Lunatics apps (check in stderr.txt)

The developers of the Boinc project ...

There is no such thing as "the Boinc project"

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

PS. Jord be careful with your statements, you might scare off people. It isn't important how many posts users have in a forum.

Nice flame there, Richard & Merle.

Luckily, by the end of the day you're gone again, walking away from the forums, thinking other thoughts. Then tomorrow other people will come here actually looking for help, asking for help, instead of ranting on and on. And then I'll just help those people, in my own way, as I have always done, in most all of the --how many posts do I have these days?

Apparently you didn't think that other people will read your post(s) in this thread and think that their easiest way out is also not to ask, but to remove Seti and just turn tail, walk away. For every problem there is a solution, all you have to do is get over your fear and ask.

The developers of the Boinc project has to provide us with a client that works with the latest generation of Intel processors.

BOINC, the client, doesn't use the CPU. It may use the GPU, but only for rendering the BOINC screen saver. Any other calculations are done by the science applications, a territory outside of BOINC its realm. While BOINC may be the managing program, it isn't much else than that. BOINC doesn't do the science, that's done by the applications provided for by the projects, in this case Seti.

To be able to use the science applications correctly, the user has to make sure his system is otherwise stable, up-to-date. That all drivers are correctly installed and the operating system updated.

If the problem "compute error" comes from somewhere else then the Intel platform and this version of boinc, I would like to hear it, thanks.

Since you have your computers hidden, there's no way any of the helpers can check what actual errors your computer throws. Do know that none of us can see any details about the computer, other than what CPU is in there and the task list. Private things as IP address and computer name are only visible to you, the user.

You don't have to unhide the computer, but then at least give a link to one of the tasks that had a compute error. You can do so by going to your Task list, right clicking on the Task ID number, do Copy link location, and paste that in an answer window here. Then any helper around here who is still willing to help you will check what error your computer actually throws and post what possible fixes there are for that.

Then it's up to you if you want to follow that advice. Or tell that person off as well.
It's easier to just tell all the helpers to bugger off or that they help wrong, post wrong, say things wrong. Instead, you can always do that one thing better, and step in and go voluntarily help out other people yourself.

---

There are now reports that the Einstein@Home project application for Intel GPU has problems with an Intel driver released this month (October 2014)

I left Einstein, in anticipation of their Drupal disaster. I suspect I won't be the only one when that dragon exits beta and is put on main. Of course, they have a 'reason' now to put Drupal on, because the present forums are all of a sudden slow and unreliable. So bring on the new thing. Just hope they updated their back-end to include the fix against the automated attacks.

The Intel GPU application supplied as a 'BOINC application' by this SETI@Home project was written and supplied by the same third-party programming team as the applications packaged in the Lunatics Installer. All that SETI has done is to test it for safety and endorsed it with their digital certificate: there should be no difference in compatibility.

Until that changes... in installer 0.44

---

@Ageless

I didn't realized that my computers were hidden, I have changed that.
A link to a workunit that gave the error:

http://setiathome.berkeley.edu/result.php?resultid=3801802475

There are many more like that one, so have fun with it :)

Oh I guess you are a little suspicious, don't be I wasn't flaming you.

I see now that I was wrong with my first reply into this forum.

My compute errors are related to a CPU error as I do not use the Intel GPU for crunching.

Maybe it is best that I stop pollute this thread, it will cause only more confusion.

I am sorry.

Test the computer with Linpack (which is Intel made program)

LinX - The simpliest Linpack interface (GUI)
http://www.xtremesystems.org/forums/showthread.php?201670-LinX-A-simple-Linpack-interface

Be warned: Linpack makes a heavy load on CPU - be prepared to watch the CPU temperature
LinX / Linpack tests CPU, RAM, bus between CPU and RAM, PSU (heavy load on CPU makes load on PSU +12 V)

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

Test the computer with Linpack (which is Intel made program)

LinX - The simpliest Linpack interface (GUI)
http://www.xtremesystems.org/forums/showthread.php?201670-LinX-A-simple-Linpack-interface

Be warned: Linpack makes a heavy load on CPU - be prepared to watch the CPU temperature
LinX / Linpack tests CPU, RAM, bus between CPU and RAM, PSU (heavy load on CPU makes load on PSU +12 V)

Haha, funny...I ran a Asus system (stress test) benchmark yesterday evenening, no errors.

Never heard of the LinX program but I downloaded the latest version and tried it a few times, check the img.

I see now that I was wrong with my first reply into this forum.

My compute errors are related to a CPU error as I do not use the Intel GPU for crunching.

I see your errors are on your Intel Core i5 running Windows 2008 R2 "Home Premium". Is this some sort of Windows Home server or Small Business server? Or is this mis-identified by BOINC?


The errors you're getting are "Access Violation (0xc0000005) at address [address] read attempt to address [address]". In general this is caused by the processor itself.

First, you need to understand that addresses in a user-mode process are virtual addresses. They are not the actual addresses used to access hardware. Rather, there is a virtual-to-physical translation circuit in the CPU (part of the memory management unit) which finds a matching entry in the "Translation Look-aside Buffer". During each context switch, the OS fills the TLB with the memory mappings belonging to your process.

So there is no way to try to access memory belonging to other processes, nor can you try to access hardware. It's not that this access is detected and fails, it's that no mapping exists for memory that does not belong to your program.

If your program accesses an address that does not map to anywhere, a trap will occur as Hans said. It's the same trap for "page faults" and "access violations". First the OS will check if the address is valid but not in the TLB (for example, your PC ran out of memory and some was swapped out to disk). In that case the OS will move the data back into physical RAM, set the proper mapping in the TLB, and continue running your program. If the OS determines that the address is completely invalid (there's no swap location associated with it), it will generate an "access violation" (Windows naming) or "segmentation fault" (POSIX naming).

Usually the cause is a logic bug, but if you had e.g. a RAM failure that changed a bit in one of your pointers, the hardware failure could trigger an access violation also.

If this were truly caused by a logic bug in the application, the same bug would manifest itself on every other system running this application. As there isn't a deluge of reports of this issue, we can probably rule that out. So the next thing to try would be testing your RAM using something like MemTest86+ or even Windows' built-in memory tester (don't know if it exists in your edition of Windows).

@OzzFan.

Yes WHS 2011, I am using this OS for many years now, and since the hardware update (2 weeks ago) I have these compute errors.

I have test the system RAM with the Windows built-in memory test program, nothing is wrong.

---