Gamma-ray pulsar binary search #1 on GPUs

Ian&Steve C.
Ian&Steve C.
Joined: 19 Jan 20
Posts: 3962
Credit: 47103162642
RAC: 65310696

there seems to be a problem

there seems to be a problem with the new LATeah3001L00 tasks. none of them will process on my systems.

 

possibly specific to linux or nvidia cards.

 

see here: https://einsteinathome.org/content/gamma-ray-gpu-tasks-hanging

 

reposting here since it seems like the devs monitor this forum more often

_________________________________________________________________________

Ian&Steve C.
Ian&Steve C.
Joined: 19 Jan 20
Posts: 3962
Credit: 47103162642
RAC: 65310696

Ian&Steve C. wrote: there

Ian&Steve C. wrote:

there seems to be a problem with the new LATeah3001L00 tasks. none of them will process on my systems.

 

possibly specific to linux or nvidia cards.

 

see here: https://einsteinathome.org/content/gamma-ray-gpu-tasks-hanging

 

reposting here since it seems like the devs monitor this forum more often

looks like this is an issue on Nvidia Volta/Turing/Ampere cards. previous gen (Pascal and before) seem unaffected.

but the new tasks are broken for newer nvidia cards. please fix.

_________________________________________________________________________

San-Fernando-Valley
San-Fernando-Valley
Joined: 16 Mar 16
Posts: 410
Credit: 10227113455
RAC: 21021667

Gamma-Ray tasks: I'm

Gamma-Ray tasks:

I'm getting driver errors en masse.

Driver error 116.

GPUs (i.e. TITAN V + GTX 1650) try repeatedly to recover.

After a certain count of error 116 System goes down with BSOD 116 driver issue.

Have newest driver from NVIDIA installed.

 

I'm aborting all GR-tasks.

Hope the tech guys have taken notice.

morpheus
morpheus
Joined: 20 Mar 05
Posts: 4
Credit: 1236880692
RAC: 11602

My Windows 10 machine with

My Windows 10 machine with 'GTX 970' is doing well.
My Windows 10 machine with 'RTX 2070 super' sucks!

Stay healthy! :)

.:morpheus:.

Betreger
Betreger
Joined: 25 Feb 05
Posts: 992
Credit: 1592825676
RAC: 776057

My I5 host with a pair of 3GB

My I5 host with a pair of 3GB GTX1060s is doing very well, the other I5 host with a GTX1660syper is having big problems. 

Tom M
Tom M
Joined: 2 Feb 06
Posts: 6464
Credit: 9588945356
RAC: 6795152

Is anyone with a problem in

Is anyone with a problem in processing have your "test" applications toggled on?

Might be getting test apps instead of production apps.

So far none of my Radeon cards across Windows and Linux seem to be having trouble (I think).

I just restarted a GR on a Windows 10/Gtx 1660 Super combo.  It hasn't immediately errored out.

I have been experiencing some trouble if I have the "wrong" Windows driver installed (for my rx cards).  The trouble is it seems to be a moving target about which works and which doesn't. ;)

I am not sure I have been having any trouble with my Nvidia drivers.

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!

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 4968
Credit: 18759095571
RAC: 7159356

The tasks don't immediately

The tasks don't immediately error out.  They just never finish and only use half the normal power for crunching.

If you let them run, the will eventually error out with the "exceeded task time limit" error.

So they just waste time and space on your computer.  Best to abort them all and switch to GW.

 

Ian&Steve C.
Ian&Steve C.
Joined: 19 Jan 20
Posts: 3962
Credit: 47103162642
RAC: 65310696

It’s not test/beta

It’s not test/beta applications. It’s the stock app, but it doesn’t appear to be the stock app that’s the problem, it’s the new tasks. They have some incorrect parameter or setting in them that’s causing issues with newer nvidia architectures. Basically anything Volta and newer. If you process some of the older task types, there is no issue on the same app. 
 

Volta = Titan V 

Turing = GTX16xx, RTX20xx

Ampere = RTX30x

 

The above cards will not be able to process these new files with symptoms ranging from immediate errors, to tasks that “run” indefinitely until they eventually hit the timeout or are aborted by the user. This is an issue on any OS (reported issues from Win7 through Win10, and several flavors of Linux), and any driver version. this has been clearly laid out in the referenced thread. 
 

I hope the devs can isolate the issues with these tasks with the newer nvidia cards. 

_________________________________________________________________________

Betreger
Betreger
Joined: 25 Feb 05
Posts: 992
Credit: 1592825676
RAC: 776057

My GTX1660 super is over

My GTX1660 super is over 17hrs so far. I'll let it run until it dies or reports and abort the remaining pulsars and only do GWs on that host. 

Betreger
Betreger
Joined: 25 Feb 05
Posts: 992
Credit: 1592825676
RAC: 776057

At the 20hr, 25min, 56sec

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

Comment viewing options

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