This list may be incomplete (I don't keep up with AMD/ATI cards), but might be a good starting place. Columms highlighted in green show the double precision capable cards.

http://en.wikipedia.org/wiki/Comparison_of_AMD_graphics_processing_units

Is OpenCL 1.2 a requirement?

My HD4850 is DP capable and supports OpenCL 1.0 but cannot run this application, it just errors out after around 10 seconds

Will I be able to somehow see it when I use a non-DP GPU?

Or is it then just producing "garbage" which does not validate against other results?

I am not sure if my APU and my GPU is DP capable as wikipedia is not very clear on that.

Greetings, =Lupus=
____________

The first generation of APU (Llano), whose CPU number format is A?-3??? doesn't support DP (if someone wants to try just to be sure, go ahead).

The second generation (Trinity), whose CPU number format is A?-5??? (in desktop) and A?-4??? (in notebooks) supports DP, but I'm not sure on this one. Again, some confirmation is welcome.

The third generation (Richland), whose CPU number format is A?-6??? (in desktop) and A?-4??? (in notebooks) supports DP. The GPU of Trinity and Richland are in the same level, so if one supports it, the other will too, but, again some confirmation is welcome.

To summarize (subject to confirmation, please report), one can run it in APU if it's A?-(>3)???.

On the low power range, the first generation (Brazos), doesn't support DP.

The second generation (Kabini), however, should support DP, but this one I really don't know. If anyone has this APU, please let us know if you can run GeneferOCL.


On the mobile GPU front, 7730M or higher should run it, and then all of the next generation (8???M) too.

8730 or higher also runs GeneferOCL.

But if anyone is actually sucessful (or unsucessful) in running GeneferOCL on their particular card/APU, please say here. An exaustive list would be very helpful.


____________

Oh, and while you are at it, the runtimes for the short tasks (even if they are growing, a indication of the size is already good) would be very useful information. I think the majority (to not say all) of mobile APU's, for instance, won't be able to complete the short tasks in time even running 24/7.
____________

I have an AMD Radeon HD 7670M, and I have computation errors running genefer (atiGFN). Is there some trouble or this card don't support any DP?

No, 7670M does not support DP.
____________

I have an AMD Radeon HD 7670M, and I have computation errors running genefer (atiGFN). Is there some trouble or this card don't support any DP?

Your GPU does NOT have DP. Only HD 77xxM or higher does.

I have an idea - would " I can run Milkyway-OCL which also need DP on this GPU/APU" be an indicator that I can also run genefer-OCL?
Their runtimes are much shorter and could be used as an indicator, and in their forums is a long list of supported GPUs...

Maybe you can also ask them for information ^.^
____________

On that note I should have mentioned in the first place that my 4850 happily runs Collatz and Milkyway OpenCL apps. I can also run PPS Sieve here, the only OCL app I can't get to run is Genefer

http://www.primegrid.com/result.php?resultid=482102691

Saaad news, saad news for my GPUs.
I tried what I said: Using Milkyway as a test object for double precision.

Here's what I got.
For my sweet ATI 7670:

Found 1 platform
Platform 0 information:
Name: AMD Accelerated Parallel Processing
Version: OpenCL 1.2 AMD-APP (1124.2)
Vendor: Advanced Micro Devices, Inc.
Extensions: cl_khr_icd cl_amd_event_callback cl_amd_offline_devices cl_khr_d3d10_sharing cl_khr_d3d11_sharing
Profile: FULL_PROFILE
Using device 0 on platform 0
Found 2 CL devices
Device 'Turks' (Advanced Micro Devices, Inc.:0x1002) (CL_DEVICE_TYPE_GPU)
Driver version: 1124.2 (VM)
Version: OpenCL 1.2 AMD-APP (1124.2)
Compute capability: 0.0
Max compute units: 6
Clock frequency: 800 Mhz
Global mem size: 1073741824
Local mem size: 32768
Max const buf size: 65536
Double extension: (none)
Device doesn't support double precision

Found 1 platform
Platform 0 information:
Name: AMD Accelerated Parallel Processing
Version: OpenCL 1.2 AMD-APP (1124.2)
Vendor: Advanced Micro Devices, Inc.
Extensions: cl_khr_icd cl_amd_event_callback cl_amd_offline_devices cl_khr_d3d10_sharing cl_khr_d3d11_sharing
Profile: FULL_PROFILE
Using device 1 on platform 0
Found 2 CL devices
Device 'Devastator' (Advanced Micro Devices, Inc.:0x1002) (CL_DEVICE_TYPE_GPU)
Board: AMD Radeon HD 7560D
Driver version: 1124.2 (VM)
Version: OpenCL 1.2 AMD-APP (1124.2)
Compute capability: 0.0
Max compute units: 4
Clock frequency: 800 Mhz
Global mem size: 536870912
Local mem size: 32768
Max const buf size: 65536
Double extension: cl_amd_fp64

double-precision constant is represented as single-precision
constant because double is not enabled

can confirm HD 7750 and HD 7950 work

so no hd 6870?

Trinity (A6-4455M) works:
http://www.primegrid.com/results.php?hostid=419390

And can be completed within the given timeframe.
____________

What is the likelyhood of the AMD R9 285 being supported in the near future?

Per the specs page here http://xfxforce.com/en-us/products/amd-radeon-r9-series/amd-radeon-r9-285-double-dissipation-edition-r9-285a-cdfc

It says "Double Precision Floating Point"-
Which is disappointing there's no output figure-
-and a relief because things moved very fast and before I knew it, I had a new system, built around this GPU & I'd be upset if there was no computing potential

As I understand, its the first of AMD's newest architecture revision. Here's to hoping its a bright one.

I've been struck by a Nvidia and an Intel recall the last two time I went on the edge

It says "Double Precision Floating Point"-
Which is disappointing there's no output figure

See http://en.wikipedia.org/wiki/List_of_AMD_graphics_processing_units for figures.

"Tonga is configured for 1/16 rate double precision performance, 1/4 the rate of the more powerful Tahiti".

AMD Radeon R9 285 Double Dissipation Edition: 3.34 TFLOPS Single Precision => 208 GFLOPS Double Precision.
AMD Radeon R9 280 Double Dissipation Edition: 3.34 TFLOPS Single Precision => 835 GFLOPS Double Precision.

Hello,

iam getting no GFN-Short/WR WUs for my Radeon 290X Card

Requesting new tasks for AMD/ATI GPU and Intel GPU
Scheduler request completed: got 0 new tasks

Catalyst 14.12
BOINC 7.4.36 x64
Win7-x64

PrimeGrid preferences :

Generalized Fermat Prime Search n=20 (GFN-Short) AMD/ATI GPU (OpenCL)
Generalized Fermat Prime Search World Record n=22 (GFN-WR) AMD/ATI GPU (OpenCL)

Do i need an app_info.xml ??? If Yes can you give me a link?

Best Regards

Thanks a lot,

now the Radeon R 290X starts Crunshing GNF-WUs.


I have Downloadet this Files

http://www.primegrid.com/download/primegrid_genefer_3_2_6_0_3.07_windows_intelx86__atiGFN.exe
http://www.primegrid.com/download/primegrid_genefer_3_2_6_0_3.07_windows_intelx86__atiGFNWR.exe

and Paste ist here

C:\ProgramData\BOINC\projects\www.primegrid.com

together with this app_info.xml

<app_info>
<app>
<name>genefer</name>
<user_friendly_name>Genefer</user_friendly_name>
</app>
<file_info>
<name>primegrid_genefer_3_2_6_0_3.07_windows_intelx86__atiGFN.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>genefer</app_name>
<version_num>210</version_num>
<api_version>7.0.64</api_version>
<avg_ncpus>1.000000</avg_ncpus>
<max_ncpus>1.000000</max_ncpus>
<file_ref>
<file_name>primegrid_genefer_3_2_6_0_3.07_windows_intelx86__atiGFN.exe</file_name>
<main_program/>
</file_ref>
<platform>windows_intelx86</platform>
<coproc>
<type>ATI</type>
<count>1.000000</count>
</coproc>
</app_version>
<app>
<name>genefer_wr</name>
<user_friendly_name>Genefer (World Record)</user_friendly_name>
</app>
<file_info>
<name>primegrid_genefer_3_2_6_0_3.07_windows_intelx86__atiGFN.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>genefer_wr</app_name>
<version_num>210</version_num>
<api_version>7.0.64</api_version>
<avg_ncpus>1.000000</avg_ncpus>
<max_ncpus>1.000000</max_ncpus>
<file_ref>
<file_name>primegrid_genefer_3_2_6_0_3.07_windows_intelx86__atiGFN.exe</file_name>
<main_program/>
</file_ref>
<platform>windows_intelx86</platform>
<coproc>
<type>ATI</type>
<count>1.000000</count>
</coproc>
</app_version>
<app>
<name>pps_sr2sieve</name>
<user_friendly_name>Proth Prime Search (Sieve)</user_friendly_name>
</app>
<file_info>
<name>primegrid_tpsieve_1.39_windows_intelx86__atiPPSsieve.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>pps_sr2sieve</app_name>
<version_num>124</version_num>
<platform>windows_x86_64</platform>
<avg_ncpus>1</avg_ncpus>
<max_ncpus>1</max_ncpus>
<flops>1.0e11</flops>
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>
<cmdline>-m 64</cmdline>
<file_ref>
<file_name>primegrid_tpsieve_1.39_windows_intelx86__atiPPSsieve.exe</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>

Today I tried to participate using my R9 290X (MSI Lightning) but no WUs were delivered to my system (Win7/x64, i5-2500K, 8 GB RAM, BOINC 7.4.36 x64, latest ATI driver), so I exactly cloned the following information:

Thanks a lot,

now the Radeon R 290X starts Crunshing GNF-WUs.


I have Downloadet this Files

http://www.primegrid.com/download/primegrid_genefer_3_2_6_0_3.07_windows_intelx86__atiGFN.exe
http://www.primegrid.com/download/primegrid_genefer_3_2_6_0_3.07_windows_intelx86__atiGFNWR.exe

and Paste ist here

C:\ProgramData\BOINC\projects\www.primegrid.com

together with this app_info.xml

<app_info>
<app>
<name>genefer</name>
<user_friendly_name>Genefer</user_friendly_name>
</app>
<file_info>
<name>primegrid_genefer_3_2_6_0_3.07_windows_intelx86__atiGFN.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>genefer</app_name>
<version_num>210</version_num>
<api_version>7.0.64</api_version>
<avg_ncpus>1.000000</avg_ncpus>
<max_ncpus>1.000000</max_ncpus>
<file_ref>
<file_name>primegrid_genefer_3_2_6_0_3.07_windows_intelx86__atiGFN.exe</file_name>
<main_program/>
</file_ref>
<platform>windows_intelx86</platform>
<coproc>
<type>ATI</type>
<count>1.000000</count>
</coproc>
</app_version>
<app>
<name>genefer_wr</name>
<user_friendly_name>Genefer (World Record)</user_friendly_name>
</app>
<file_info>
<name>primegrid_genefer_3_2_6_0_3.07_windows_intelx86__atiGFN.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>genefer_wr</app_name>
<version_num>210</version_num>
<api_version>7.0.64</api_version>
<avg_ncpus>1.000000</avg_ncpus>
<max_ncpus>1.000000</max_ncpus>
<file_ref>
<file_name>primegrid_genefer_3_2_6_0_3.07_windows_intelx86__atiGFN.exe</file_name>
<main_program/>
</file_ref>
<platform>windows_intelx86</platform>
<coproc>
<type>ATI</type>
<count>1.000000</count>
</coproc>
</app_version>
<app>
<name>pps_sr2sieve</name>
<user_friendly_name>Proth Prime Search (Sieve)</user_friendly_name>
</app>
<file_info>
<name>primegrid_tpsieve_1.39_windows_intelx86__atiPPSsieve.exe</name>
<executable/>
</file_info>
<app_version>
<app_name>pps_sr2sieve</app_name>
<version_num>124</version_num>
<platform>windows_x86_64</platform>
<avg_ncpus>1</avg_ncpus>
<max_ncpus>1</max_ncpus>
<flops>1.0e11</flops>
<coproc>
<type>ATI</type>
<count>1</count>
</coproc>
<cmdline>-m 64</cmdline>
<file_ref>
<file_name>primegrid_tpsieve_1.39_windows_intelx86__atiPPSsieve.exe</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>

This morning, I finally got a WU.

The runtime estimate is almost 153 hrs - is this normal for a "Genefer (World Record) 2.10" WU or isn't it rather possible that this is quite overestimated?
It sems quite long to me, because this type of WU is part of your upcoming challenge and would take 6 days to complete while the challenge itself lasts only for 8 days meaning that I can process exactly a single WU of this type during that whole challenge. ;-)

And one more question:
The website preference settings of this sub project state that "some ATI/AMD GPUs may require installation of the ATI Stream SDK". Is that required for a 290X card, does it speed up the computation?

Michael.
____________

This sounds pretty accurate.

With drawing up the challenge schedule I was aware that many people will only be able to complete 1 WR unit. That however is to me part of the challenge. Are you confident in your machine that you go for the big score with 1 WR unit and filling the left over time up with "short" units, with the risk that your big scoring unit gets invalidated for some reason? Or are you going to bet on completing as many short tasks as you can, missing out on the significant bonus credit WR's give, but minimizing the damage if one WU gets invalidated... :)
____________
PrimeGrid Challenge Overall standings --- Last update: From Pi to Paddy (2016)

Well, I will try the long one - so far I had no problems with my 290X.

Still, what about the AMD SDK requirement? What is it good for?
I mean, the WUs download and start without complaints - that's clear - but would there be any benefit having the SDK installed?

Michael.
____________

Well, it seems that I was right. Based on the current progress bar indication and the re-calculation of the remaining WU duration, that "long" WU should finish in significantly less time than half of the original estimate on my card (approx. 60 hrs per WU). So, it should be possible to work up 2 WUs during the challenge duration...

Of course, maybe even the progress bar indication is not reliable. ;-)
And also, it might be that this WU won't be validated at all as my card is factory-overclocked (it is the MSI R9 290X LIGHTNING, which I fancy a lot because of its excellent and comparably silent cooling system).
However, so far I have not experienced any faulty WUs with other DC projects and to my experience, PrimeGrid quite certainly is not the most GPU hardware challenging DC project out there in the wild as judged by GPU temperature.

Michael.
____________

It would actually have been nice, if - at least during an active challenge - the returned WUs could be validated. For several days, this one is sitting there with no other copy handed out for validation.
Thanks.

Michael.
____________