Suspended Project_JBird 737647

JBird
Joined: 22 Dec 14
Posts: 1963
Credit: 4046216051
RAC: 0
Topic 197894

Sorry, I apparently need to update my Graphics card. Several OpenGL errors causing several failed jobs from my GeForce 8400 GS with only 16 CUDA cores. I'll be back in 1 week with some muscle.
Happy New Year Einstein!

JBird
Joined: 22 Dec 14
Posts: 1963
Credit: 4046216051
RAC: 0

Suspended Project_JBird 737647

Follow up - Newegg shipping today: GeForce GT 640 2GB 128 bit DDR3 and 384 CUDA Cores with 900Mhz core clock OpenGL 4.2 = May get here in time to crunch suspended jobs

Claggy
Claggy
Joined: 29 Dec 06
Posts: 560
Credit: 2699403
RAC: 0

RE: Follow up - Newegg

Quote:
Follow up - Newegg shipping today: GeForce GT 640 2GB 128 bit DDR3 and 384 CUDA Cores with 900Mhz core clock OpenGL 4.2 = May get here in time to crunch suspended jobs


That may be hard, all your GPU tasks have errored already, and you're aborted your last CPU tasks, they weren't affected, you could have left them crunching with just GPU activity suspended,

So you have no suspended tasks to crunch:

All tasks for computer 11709374

In progress tasks for computer 11709374

Claggy

Claggy
Claggy
Joined: 29 Dec 06
Posts: 560
Credit: 2699403
RAC: 0

RE: Several OpenGL errors

Quote:
Several OpenGL errors causing several failed jobs from my GeForce 8400 GS with only 16 CUDA cores.


They weren't OpenGL (Open Graphics Library) errors, they weren't even OpenCL (Open Computing Language) errors, But were data transfer errors for the CUDA app, the errors were eithier:

Quote:

[01:46:17][6028][ERROR] Error during CUDA device->host HS data transfer (error: 999)
[01:46:17][6028][ERROR] Demodulation failed (error: 1008)!

[01:40:55][5104][ERROR] Error during CUDA device->host HS power spectrum transfer (error: 702)
[01:40:55][5104][ERROR] Demodulation failed (error: 1008)!

[01:00:15][5440][ERROR] Error during CUDA device->host time series length transfer (error: 999)
[01:00:15][5440][ERROR] Demodulation failed (error: 1008)!

[15:24:09][5348][ERROR] Error during CUDA device->host time series mean transfer (error: 999)
[15:24:09][5348][ERROR] Demodulation failed (error: 1008)!

Claggy

MAGIC Quantum Mechanic
MAGIC Quantum M...
Joined: 18 Jan 05
Posts: 1886
Credit: 1410167868
RAC: 1173906

Could have got about twice

Could have got about twice the Cuda's for a little more like the 750Ti that can be OC'd

Those are pretty cheap at that same website

JBird
Joined: 22 Dec 14
Posts: 1963
Credit: 4046216051
RAC: 0

@Claggy -hey thanks for the

@Claggy -hey thanks for the look-see at my problem. I recently built this PC and just discovered my NVIDIA card was EOL so I grabbed the GT 640 best upgrade for my budget at this time. It won't get here til Monday Jan 5 so I'll shut down Sunday and install new Corsair PS to meet 12v rail spec requirements in front of the NVIDIA install. I assume a Project Reset is my best bet. Do I need to look at Lunatics apps? Or do I need any config advice or will Boinc take care of that new hardware recognition and capability assessment and Assign tasks accordingly? Thanks in advance.
JBird

JBird
Joined: 22 Dec 14
Posts: 1963
Credit: 4046216051
RAC: 0

Thanks Magic. I didn't even

Thanks Magic. I didn't even look at that one but I will. Don't know my way around SLI or whatever it takes to enable multiple GPUs on same PC/project (if that even is possible here)

JBird
Joined: 22 Dec 14
Posts: 1963
Credit: 4046216051
RAC: 0

PS - do you recommend I just

PS - do you recommend I just Resume E@H then and let whatever crunching *can happen? I tried to resume Rosetta and it *didn't so I Suspended it again and set Resources to 1. Still haven't seen any bigger jobs for SETI and bunch of waiting to validate tasks there anyhow (dunno what to do about *that - did my thing but they're hungup for some reason)

Claggy
Claggy
Joined: 29 Dec 06
Posts: 560
Credit: 2699403
RAC: 0

RE: PS - do you recommend I

Quote:
PS - do you recommend I just Resume E@H then and let whatever crunching *can happen? I tried to resume Rosetta and it *didn't so I Suspended it again and set Resources to 1. Still haven't seen any bigger jobs for SETI and bunch of waiting to validate tasks there anyhow (dunno what to do about *that - did my thing but they're hungup for some reason)


Just resume Einstein, you're already got apps and data downloaded, Einstein is a bit special (on some of their apps) in that you download datasets,
then the project assigns tasks according to what datasets you have,
If you reset the project you loose those datasets, and you'll have to download new sets (as well as apps and dlls) increasing the amount of downloads you and the project have to do.

You won't get any GPU work at Seti for your 8400GS as it's on 340.52 drivers, the 340.xx drivers and legacy NV GPUs have been hit by an OpenCL driver Bug,
Seti has excluded legacy GPUs from getting work if they are running 340.xx drivers, (including for the Seti Cuda apps, inspite of them not being affected by this Bug)

Tasks that are pending validation, or inconclusive are waiting for your wingmen to complete their task, only then can they validate.

Claggy

JBird
Joined: 22 Dec 14
Posts: 1963
Credit: 4046216051
RAC: 0

RE: RE: PS - do you

Quote:
Quote:
PS - do you recommend I just Resume E@H then and let whatever crunching *can happen? I tried to resume Rosetta and it *didn't so I Suspended it again and set Resources to 1. Still haven't seen any bigger jobs for SETI and bunch of waiting to validate tasks there anyhow (dunno what to do about *that - did my thing but they're hungup for some reason)

Just resume Einstein, you're already got apps and data downloaded, Einstein is a bit special (on some of their apps) in that you download datasets,
then the project assigns tasks according to what datasets you have,
If you reset the project you loose those datasets, and you'll have to download new sets (as well as apps and dlls) increasing the amount of downloads you and the project have to do.

You won't get any GPU work at Seti for your 8400GS as it's on 340.52 drivers, the 340.xx drivers and legacy NV GPUs have been hit by an OpenCL driver Bug,
Seti has excluded legacy GPUs from getting work if they are running 340.xx drivers, (including for the Seti Cuda apps, inspite of them not being affected by this Bug)

Tasks that are pending validation, or inconclusive are waiting for your wingmen to complete their task, only then can they validate.

Claggy


Roger that Claggy and thanks for reinforcing my suspicion that my card was now EOL/Legacy class so upgraded to the GT 640 (budget consideration) in hopes of Driver progression from NV and stronger OpenCL and SSE and CUDA engines at the same time. Good to know that not much else I *can do other than maybe move my swap file over to the SSD *if that would help (it's currently on a separate HDD)that I don't think *anybody's even utilizing and I have a 4gig RAMDisk in place. I guess more physical RAM is next upgrade. I checked on a few of my wingmen and found more than I'd like using pretty underpowered systems or uncontacted server in months (go figure) - hafta fault BOINC for not reassigning stuff like that right?

JBird
Joined: 22 Dec 14
Posts: 1963
Credit: 4046216051
RAC: 0

OK well, NVIDIA OpenGL

OK well, NVIDIA OpenGL detected problem with Driver and cannot continue message twice after 2 hrs each instance; and odd behavior of task config/runs re: kinda over-amping my Resources prefs. So I just Suspended it and zero'd the resources so SETI could get back to it unhampered. Hardware updates Monday afternoon after card gets here, then I'll try again after I get it settled and confirm freshest Drivers etc. Any suggestions how to ensure or enable or tweak my Intel i5 and HD 4000 drivers are *happening here? All I've done so far is enable it in BIOS and Device Mgr but don't know what else to do about it - it doesn't seem to show in any computer config in BOINC. Would just plugging a DVI cable in it but not my monitor avail anything?(referencing a comment I saw about cheating it with a Dongle) My little 8400 GS beats the pants off it on my Desktop and what I do. Just want the HD 4000 available if anything BOINC wants it.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.