Gamma-ray pulsar binary search #1 on GPUs

archae86
archae86
Joined: 6 Dec 05
Posts: 3157
Credit: 7219624931
RAC: 975722

Betreger wrote: At the 20hr,

Betreger wrote:

At the 20hr, 25min, 56sec the GTX1660super errored out with a computation error. I'm now on to GWs. 

stderr contains:

EXIT_TIME_LIMIT_EXCEEDED

Traveller
Joined: 20 Feb 05
Posts: 3
Credit: 544295275
RAC: 886125

there seems to be a problem

From a previous post:

there seems to be a problem with the new LATeah3001L00 tasks

I'm seeing this for these same tasks on my RTX 2060 on Windows 10.

A different Task, LATeah2049Lae_172.0_0_0.0_12999070_3, went through earlier today without issue.

GWGeorge007
GWGeorge007
Joined: 8 Jan 18
Posts: 3060
Credit: 4963207686
RAC: 1415118

Traveller wrote: From a

Traveller wrote:

From a previous post:

there seems to be a problem with the new LATeah3001L00 tasks

I'm seeing this for these same tasks on my RTX 2060 on Windows 10.

A different Task, LATeah2049Lae_172.0_0_0.0_12999070_3, went through earlier today without issue.

This different task < LATeah2049Lae > ends with a "3" which indicates that it is one which has been through 3 other hosts and has been processed by you now.  The 'age' of the task does have a bearing whether or not it will be processed by your Touring card.

George

Proud member of the Old Farts Association

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 4963
Credit: 18705482198
RAC: 6281348

The LATeah3001L00 Gamma-Ray-P

The LATeah3001L00 Gamma-Ray-Pulsar tasks with the bolded part of their name are the problem tasks with any Volta/Turing/Ampere generation cards.  Any other kind of task seen so far runs OK.

Bernd (moderator/scientist/administrator) posted earlier today they are preventing any LATeah3001L00 tasks being sent to the aforementioned cards for now until the problem is understood and a fix being implemented.

 

Ian&Steve C.
Ian&Steve C.
Joined: 19 Jan 20
Posts: 3945
Credit: 46663852642
RAC: 64143916

Keith Myers wrote: Bernd

Keith Myers wrote:

Bernd (moderator/scientist/administrator) posted earlier today they are preventing any LATeah3001L00 tasks being sent to the aforementioned cards for now until the problem is understood and a fix being implemented.

it appears that they have actually blocked ALL Gamma ray tasks for the affected cards (anything CC >7.0), not just these new 3001L00 tasks. so you won't even get resends of the older tasks that do work.

_________________________________________________________________________

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 4963
Credit: 18705482198
RAC: 6281348

Yes, I saw that when I looked

Yes, I saw that when I looked for any resends of the older tasks and saw none and the server logfile shows not even requesting FGRPB1G tasks.

 

CaptainGonzo
CaptainGonzo
Joined: 4 Mar 21
Posts: 1
Credit: 8729083
RAC: 0

I got a 6800XT it should eat

I got a 6800XT it should eat that up!

Tom M
Tom M
Joined: 2 Feb 06
Posts: 6434
Credit: 9563084476
RAC: 9429103

CaptainGonzo wrote:I got a

CaptainGonzo wrote:

I got a 6800XT it should eat that up!

Congratulations.  You may be the 3rd E@H user with a Big Navi GPU.  If you want us to admire the system in "real-time" you will need to unhide it.

Tom M

A Proud member of the O.F.A.  (Old Farts Association).  Be well, do good work, and keep in touch.® (Garrison Keillor)  I want some more patience. RIGHT NOW!

cpalmer
cpalmer
Joined: 13 Feb 21
Posts: 1
Credit: 7609840
RAC: 0

On 15 March I had two

On 15 March I had two gamma-ray pulsar binary search GPU tasks marked invalid. Nothing at all obvious in the outputs. When I look at one WU in particular (532165066) it's strange because one of the cohorts marked valid actually had an opencl compilation error! But maybe that was benign? Grateful if anyone could share any insight into these. The other WU was 531989259.

Thanks, Chris

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4312
Credit: 250390161
RAC: 34702

cpalmer wrote: On 15 March I

cpalmer wrote:

On 15 March I had two gamma-ray pulsar binary search GPU tasks marked invalid. Nothing at all obvious in the outputs. When I look at one WU in particular (532165066) it's strange because one of the cohorts marked valid actually had an opencl compilation error! But maybe that was benign?

The OpenCL error was actually the result of testing whether the device supports FP64, and normal for devices which don't. Computation takes a different (slower) code for these devices.

Other than that, your result just doesn't match those of the other two closely enough for the tolerances defined in our validator - I'm not sure how that came about, but this occasionally happens to all of us.

BM

Comment viewing options

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