Comparing cuda23 to cuda50 to openCL_ati...

Message boards : Number crunching : Comparing cuda23 to cuda50 to openCL_ati...
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 5124
Credit: 276,046,078
RAC: 462
Message 1552033 - Posted: 4 Aug 2014, 1:38:11 UTC
Last modified: 4 Aug 2014, 1:38:56 UTC

I think I have something of a handle on the way the Cuda23-50 map. The higher the number, the newer the GPU chip set will run more efficiently on it. At least that is what I am getting from the documentation and the sample .CFG files and yes my GTX 750Ti appears to be running "like greased lightning" on Cuda50 files.

But what about the Radeon/ATI equivalents?

I see at the moment:

opencl_ati_cat123
opencl_ati5_?
etc.

Is there some kind of mapping to either the equivalent CudaXX file or something that gives you a feel for what each type of file usually works best with? For instance my Radeon GPU for Seti parameter files have both a mb_.....ati___.txt and a mb_....ati5_....txt parameter file. So clearly there is a distinction between ati and ati5.

I am leaving Astropulse completely out of the question because I don't want to muddy the waters.

I am still working out how my A-10 5700 oughta be running so I am interested in the Radeon side of the house.

Thanks,
Tom
A proud member of the OFA (Old Farts Association).
ID: 1552033 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 1552090 - Posted: 4 Aug 2014, 4:32:30 UTC

There are 2 ATI apps.
OpenCL ATI
OpenCL ATI HD5
As I recall HD5 was intended for Radeon HD5000 series & newer. Where the other is for HD4000 & newer.

The other plan classes are primarily because ATI drivers don't have have an easy way to tell BOINC their versions information. So a new plan class has to be created when a version issue comes up. The most recent opencl_ati5_nocal is because newer cards/drivers don't provide a CAL version to BOINC.
CAL was ATI's version for GPGPU computing, but when AMD bought ATI they dropped support for CAL in favor of OpenCL.
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 1552090 · Report as offensive
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 5124
Credit: 276,046,078
RAC: 462
Message 1552162 - Posted: 4 Aug 2014, 12:01:51 UTC - in response to Message 1552090.  

Thank you for the explanation.

Do any of the groupings have a tendency to process faster on same Radeon gpu than others? Or is this one of those throw enough mud at the wall for your system and see which sticks issue?

Thanks,

Tom
A proud member of the OFA (Old Farts Association).
ID: 1552162 · Report as offensive
Profile Mike Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 01
Posts: 34258
Credit: 79,922,639
RAC: 80
Germany
Message 1552189 - Posted: 4 Aug 2014, 13:26:09 UTC - in response to Message 1552162.  

Thank you for the explanation.

Do any of the groupings have a tendency to process faster on same Radeon gpu than others? Or is this one of those throw enough mud at the wall for your system and see which sticks issue?

Thanks,

Tom


In principle you can say HD 5 version is faster on x770 and above cards.
Same for 270/280 and 290.


With each crime and every kindness we birth our future.
ID: 1552189 · Report as offensive
Josef W. Segur
Volunteer developer
Volunteer tester

Send message
Joined: 30 Oct 99
Posts: 4504
Credit: 1,414,761
RAC: 0
United States
Message 1552267 - Posted: 4 Aug 2014, 17:19:15 UTC

As a further point of clarification, the project plan classes with ati5 in the name are the non-HD5 app versions. IOW (opencl_ati_cat132) is HD5 but (opencl_ati5_cat132) is not.
                                                                   Joe
ID: 1552267 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14650
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1552269 - Posted: 4 Aug 2014, 17:25:07 UTC - in response to Message 1552267.  

As a further point of clarification, the project plan classes with ati5 in the name are the non-HD5 app versions. IOW (opencl_ati_cat132) is HD5 but (opencl_ati5_cat132) is not.
                                                                   Joe

Aaaaaaaarrgghh.
ID: 1552269 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 1552282 - Posted: 4 Aug 2014, 18:24:34 UTC - in response to Message 1552267.  

As a further point of clarification, the project plan classes with ati5 in the name are the non-HD5 app versions. IOW (opencl_ati_cat132) is HD5 but (opencl_ati5_cat132) is not.
                                                                   Joe

0.o
I am sure it made sense at the time... well maybe it did.
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 1552282 · Report as offensive
Profile Tom M
Volunteer tester

Send message
Joined: 28 Nov 02
Posts: 5124
Credit: 276,046,078
RAC: 462
Message 1552353 - Posted: 4 Aug 2014, 21:45:10 UTC - in response to Message 1552269.  

As a further point of clarification, the project plan classes with ati5 in the name are the non-HD5 app versions. IOW (opencl_ati_cat132) is HD5 but (opencl_ati5_cat132) is not.
                                                                   Joe

Aaaaaaaarrgghh.


I think I quite agree. There wasn't a logical progression on the naming convention(s)?

Someone want to post a more or less exhaustive list and divvy them up into the assorted classes they belong to for me?

Thanks,

Tom
A proud member of the OFA (Old Farts Association).
ID: 1552353 · Report as offensive
Wedge009
Volunteer tester
Avatar

Send message
Joined: 3 Apr 99
Posts: 451
Credit: 431,396,357
RAC: 553
Australia
Message 1553032 - Posted: 6 Aug 2014, 21:03:26 UTC

I have a vague recollection of a post explaining that the '5' in ati5 referred to the version number reported or received by BOINC when determining the class of ATI/AMD GPU, where '5' referred to HD4000+. It's an unfortunate coincidence, hence the confusion.

Fortunately, plan class names don't matter much on anonymous platforms.
Soli Deo Gloria
ID: 1553032 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 1553040 - Posted: 6 Aug 2014, 21:21:18 UTC - in response to Message 1553032.  

I have a vague recollection of a post explaining that the '5' in ati5 referred to the version number reported or received by BOINC when determining the class of ATI/AMD GPU, where '5' referred to HD4000+. It's an unfortunate coincidence, hence the confusion.

Fortunately, plan class names don't matter much on anonymous platforms.

I have decided I am going to star using the plan class waffles.
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 1553040 · Report as offensive

Message boards : Number crunching : Comparing cuda23 to cuda50 to openCL_ati...


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.