Gamma-ray pulsar binary search #1 on GPUs

C0d3r
C0d3r
Joined: 31 Dec 09
Posts: 10
Credit: 143,548,322
RAC: 132,013

Thank you!

Thank you!

Jeroen
Jeroen
Joined: 25 Nov 05
Posts: 379
Credit: 738,546,503
RAC: 7,673

I installed my AMD R9-290x

I installed my AMD R9-290x alongside my 7970 card on my Linux host. An issue I had with previous BRP applications is that tasks would fail to validate when running more than one task at a time on the 290x. With the FGRP GPU application, I am running 2 tasks on the 290x and so far tasks are validating okay. Run time is about 370 seconds per task which is about 12-14% faster than my 7970.

TimeLord04
TimeLord04
Joined: 8 Sep 06
Posts: 1,442
Credit: 72,378,840
RAC: 0

Bernd Machenschalk wrote:I

Bernd Machenschalk wrote:
I just issued app vesrion 1.17 (currently OSX only, Beta test) that, amog other things, should have the progress counting fixed.

I'm picking up 1.17 Units on both of my computers, now, but won't be getting to them for a couple of days while the MAC finishes the 1.14 Units, and XP Pro x64 finishes the 1.16 Units.

The Invalids still hold at the 5 I reported earlier; NO NEW Invalids to report on 1.14 for MAC.  Smile

 

TL

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join SETI Refugees

Mad_Max
Mad_Max
Joined: 2 Jan 10
Posts: 147
Credit: 1,768,961,952
RAC: 734,630

Kai Leibrandt wrote:Hi

Kai Leibrandt wrote:

Hi all,

I need some help here - I'm getting wildly different runtimes on two computers with identical R9 280x's. One is getting runtimes of about 975s where the other gets about 420s. Luxmark is showing similar behaviour (4300 vs 12700). Both machines are running exclusively FGRP tasks (one at a time for now until I have sorted this out). Both machines are running idle apart from E@H and have ample cpu and ram to support the gpus (the slower one has 2x L5630, 16 threads, the faster one 2x X5670, so 24 threads, both have 24Gb ram). Both are running OSX El Capitan. I have tried swapping the gpus out to exclude the possibilty of a dodgy gpu, and got identical results, the slow machine remaining the slower one. BIOS settings are identical.

The only difference I can see is that one has it's gpu in a PCIE x8 slot (a server board without any x16 slots) where the other has the gpu a proper x16 slot. My understanding so far however was that the reduced bandwidth of the x8 slot shouldnt affect opencl performance, at least not to this extent.

Does anyone have any ideas what I could try to at least reduce the difference between computation performance?

Many thanks in advance,

 Kailee.

Defender_2 wrote:
Since E@H is verhy bandwith-hungry I guess it's caused by the different PCI-lanes. But I can't say more about it since I'm no expert.

Kai Leibrandt wrote:

Could someone please confirm this? I would have thought as much data as possible would be transferred to gpu memory initially where it would then be crunched... This would mean pie bandwidth wouldn't have such a large effect on performance.

Tia,Kailee.

I think PCI-E can not have significant impact on the sped of current FGRP GPU app. Because i have many identical GPUs (AMD HD 7870 2 Gb) running in wide mix of PCI-E buses: from x16 down to 8x, 4x and even 1x (this last connected via flex PCI-E riser 1x==>16x, because its 3rd GPU in computer and MB have only 2 full sized PCI-E slots)
And see only minor variation from PCI-E connection. In best(x16) vs worst(x1) cases it around 20-25% maximum.
So in x16 vs x8 situation should generally be no any significant difference.

Mad_Max
Mad_Max
Joined: 2 Jan 10
Posts: 147
Credit: 1,768,961,952
RAC: 734,630

C0d3r wrote:I switched my

C0d3r wrote:

I switched my browser Mozilla Firefox to Google Chrome and it works much better. I can play videos in HD it's running fine and scrolling is smooth. But don't know why!? I experience a little bit of lag in some apps but I can tolerate it, my first and biggest problem was the browser thing. It seems resolved.

In FireFox you can try alter "use hardware acceleration" setting in preferences menu (about:preferences#advanced )

Usual it solve problems with lags when GPU under heavy load.Seems Chrome just already set to not use  hardware acceleration.

Or as alternative you can set BOINC to NOT run GPU tasks while computer in use - it will allow GPU work only after few minutes of user inactivity and pause GPU work after firs mouse move or button pressed: Options ==> Computing preferences ==> Computing tab ==> Suspend GPU computing when computer is in use

 

Holmis
Joined: 4 Jan 05
Posts: 1,118
Credit: 811,676,463
RAC: 235,057

Bernd Machenschalk wrote:I

Bernd Machenschalk wrote:
I just issued app version 1.17 (currently OSX only, Beta test) that, among other things, should have the progress counting fixed.

I see that you also released a 1.17 version for windows and that the progress counting now works much better, thank you!

Kailee71
Kailee71
Joined: 22 Nov 16
Posts: 35
Credit: 42,623,563
RAC: 0

Mad_Max wrote: I think PCI-E

Mad_Max wrote:

I think PCI-E can not have significant impact on the sped of current FGRP GPU app. Because i have many identical GPUs (AMD HD 7870 2 Gb) running in wide mix of PCI-E buses: from x16 down to 8x, 4x and even 1x (this last connected via flex PCI-E riser 1x==>16x, because its 3rd GPU in computer and MB have only 2 full sized PCI-E slots)

And see only minor variation from PCI-E connection. In best(x16) vs worst(x1) cases it around 20-25% maximum.
So in x16 vs x8 situation should generally be no any significant difference.

Hi Mad_Max,

that's exactly what I thought. Have given up on this issue for now, just swapped the 280x for a gtx580 as that is for some weird reason much happier in that machine... Well at least it's not being hampered nearly so bad by the same issue. The 280x is now back in a windows machine doing milkyway.

If anyone can think of something to try to get the ati to get better results - I would be very grateful and stick the 280x back into the einstein machine...

Thanks,

 

Kailee.

Jim1348
Jim1348
Joined: 19 Jan 06
Posts: 389
Credit: 202,199,824
RAC: 6,097

Kai Leibrandt wrote:If anyone

Kai Leibrandt wrote:
If anyone can think of something to try to get the ati to get better results - I would be very grateful and stick the 280x back into the einstein machine... 

On one of the forums (I forget which), it was found that turning on the AERO intereface helped reduce lag.  I think you have one Win7 machine, so you could try it there.  That is somewhat counter-intuitive, but I think it adds a buffer.

WhiteWulfe
Joined: 3 Mar 15
Posts: 31
Credit: 61,177,387
RAC: 0

Holmis wrote:Bernd

Holmis wrote:
Bernd Machenschalk wrote:
I just issued app version 1.17 (currently OSX only, Beta test) that, among other things, should have the progress counting fixed.

I see that you also released a 1.17 version for windows and that the progress counting now works much better, thank you!

 

This is definitely good to hear ^_^

Kailee71
Kailee71
Joined: 22 Nov 16
Posts: 35
Credit: 42,623,563
RAC: 0

Jim1348 wrote:Kai Leibrandt

Jim1348 wrote:
Kai Leibrandt wrote:
If anyone can think of something to try to get the ati to get better results - I would be very grateful and stick the 280x back into the einstein machine... 

On one of the forums (I forget which), it was found that turning on the AERO intereface helped reduce lag.  I think you have one Win7 machine, so you could try it there.  That is somewhat counter-intuitive, but I think it adds a buffer.

Hey Jim,

thanks for the effort but my Einstein machines are now OSX. But you gave me an idea - might try Windows on that hardware to see if that makes a difference.

Ta,

Kailee.

Comment viewing options

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