Discussion Thread for the Continuous GW Search known as O2MD1 (now O2MDF - GPUs only)

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 4700
Credit: 17544937123
RAC: 6412878

They stopped generating any

They stopped generating any new O2MD1 cpu work since they published their research paper. All you can get now are timed out or errored resends.  They still are distributing O2MDF gpu work though.  The project SSP page tells the tale. There is very little resend work to snag unless you are very lucky.

Project server status page

 

Tom M
Tom M
Joined: 2 Feb 06
Posts: 5586
Credit: 7674012813
RAC: 1787200

It finally occurred to me

It finally occurred to me that if I use the drop-down menu on the right of the tasks listing I can rapidly find the top producing GW gpu system.

Possible top GW gpu producing system

A Proud member of the O.F.A.  (Old Farts Association).  Be well, do good work, and keep in touch.® (Garrison Keillor)

tullio
tullio
Joined: 22 Jan 05
Posts: 2118
Credit: 61407735
RAC: 0

I have installed a 4 GB board

I have installed a 4 GB board Video RAM as recommended. But many tasks fail with an internal error, XLAL function call failed. Why?

Tullio

Mr P Hucker
Mr P Hucker
Joined: 12 Aug 06
Posts: 819
Credit: 481368012
RAC: 4215

Tom M wrote: It finally

Tom M wrote:

It finally occurred to me that if I use the drop-down menu on the right of the tasks listing I can rapidly find the top producing GW gpu system.

Possible top GW gpu producing system

That's only a third better than you.

If this page takes an hour to load, reduce posts per page to 20 in your settings, then the tinpot 486 Einstein uses can handle it.

Ian&Steve C.
Ian&Steve C.
Joined: 19 Jan 20
Posts: 3681
Credit: 33829720132
RAC: 37741370

tullio wrote: I have

tullio wrote:

I have installed a 4 GB board Video RAM as recommended. But many tasks fail with an internal error, XLAL function call failed. Why?

Tullio

 

OpenCL Create Context failed: CL_OUT_OF_HOST_MEMORY

 

it's pretty self explanatory. you're out of memory. though I'm not sure if this error refers to system (CPU) memory, or the GPU memory.

 

are you trying to run more than 1 WU at a time?

do you have other applications running on the GPU at the same time? (such as other boinc projects, 3D applications or games, etc)

are you running CPU projects that are using a large portion of your installed 12GB memory?

 

 

_________________________________________________________________________

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6534
Credit: 284710168
RAC: 110462

My reading at khronos.org

My reading at khronos.org leads me to believe that host refers to the CPU, and not whatever device was chosen to instantiate an OpenCL state machine ( this would be the GPU here ). Typically the error arises when 'clEnqueueNDRangeKernel' is called by the host, this function sets the overall size of the work to be done. 

{ CL_​OUT_​OF_​RESOURCES is used if it is an issue with the device memory }

So yes, it seems a bit odd that 12GB of system memory is being challenged.

Cheers, Mike.

I have made this letter longer than usual because I lack the time to make it shorter ...

... and my other CPU is a Ryzen 5950X :-) Blaise Pascal

Ian&Steve C.
Ian&Steve C.
Joined: 19 Jan 20
Posts: 3681
Credit: 33829720132
RAC: 37741370

well a few pages back, he did

well a few pages back, he did claim to have 11GB of system memory used. so he must be filling it up with CPU tasks that use a lot of RAM. I think he runs Rosetta also.

tullio wrote:

 A GW GPU task has started on my GTX 1060 with its 3 GB Video RAM and seems OK also on GPU-Z. Memory used is 1962 MB. Task has completed and is pending. 11 GB of system memory used.

Tullio

_________________________________________________________________________

tullio
tullio
Joined: 22 Jan 05
Posts: 2118
Credit: 61407735
RAC: 0

I never run more than one

I never run more than one GPU  task at the time. GPU-Z shows that GW tasks use about 3333 MB of Video RAM and I have 4 GB on a GTX 1050 Ti board. The PC is brand new, has an i5 Intel CPU and 12 GB of system RAM. GPU-Z shows that about 6 GB are used. So it is not a matter of memory.

Tullio

Mr P Hucker
Mr P Hucker
Joined: 12 Aug 06
Posts: 819
Credit: 481368012
RAC: 4215

Mike Hewson wrote: My

Mike Hewson wrote:

My reading at khronos.org leads me to believe that host refers to the CPU, and not whatever device was chosen to instantiate an OpenCL state machine ( this would be the GPU here ). Typically the error arises when 'clEnqueueNDRangeKernel' is called by the host, this function sets the overall size of the work to be done. 

{ CL_​OUT_​OF_​RESOURCES is used if it is an issue with the device memory }

So yes, it seems a bit odd that 12GB of system memory is being challenged.

Cheers, Mike.

 

Isn't it impossible for a PC to run out of memory?  It would just use the page file.  If you're low on memory, all that will happen is Boinc sees this and doesn't start tasks (so as to avoid the page file), I wouldn't expect a task to start and fail.

If this page takes an hour to load, reduce posts per page to 20 in your settings, then the tinpot 486 Einstein uses can handle it.

Ian&Steve C.
Ian&Steve C.
Joined: 19 Jan 20
Posts: 3681
Credit: 33829720132
RAC: 37741370

tullio wrote: I never run

tullio wrote:

I never run more than one GPU  task at the time. GPU-Z shows that GW tasks use about 3333 MB of Video RAM and I have 4 GB on a GTX 1050 Ti board. The PC is brand new, has an i5 Intel CPU and 12 GB of system RAM. GPU-Z shows that about 6 GB are used. So it is not a matter of memory.

Tullio

how much memory are you allowing BOINC to use in the compute preferences? by default if you havent changed it, BOINC is set to only use 50% of your memory, which is ~6GB in your case. try increasing that to 75% and see what happens.

_________________________________________________________________________

Comment viewing options

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