Provided the user is running a late-enough BOINC client to handle OpenCL 
'natively'.

Some projects have installed OpenCL applications under a plan_class that will 
issue them in response to  a native-mode (CUDA or ATI) GPU request, as a 
courtesy to users who don't want to upgrade beyond BOINC v6.



>________________________________
> From: Raistmer the Sorcerer <[email protected]>
>To: Michael Tughan <[email protected]> 
>Cc: [email protected] 
>Sent: Wednesday, 30 January 2013, 14:02
>Subject: Re: [boinc_dev] Feature request for server config
> 
>In this case BOINC will not report availability of OpenCL device and OpenCL 
>app will not be distributed to host. No special treatment required in this 
>case. But yes, it's another case where CAL available and OpenCL not.
>Hope David's approach will work.
>
>
>Среда, 30 января 2013, 7:46 -05:00 от Michael Tughan 
><[email protected]>:
>>One other problem I see with that is if the user is using old drivers that 
>>support CAL but not OpenCL. A somewhat more unlikely scenario, especially 
>>with the kind of people that would be inclined to run BOINC, but still a 
>>distinct possibility.
>>
>>Michael Tughan
>>
>>On 2013-01-30, at 4:14 AM, Raistmer the Sorcerer < [email protected] > wrote:
>>
>>> Charlie, you forgot about AMD GPUs that support CAL but can't support 
>>> OpenCL. Namely, HD2xxx, HD3xxx and their onboard analogues.
>>> Situation with ATi differs from sitaution with NV where all CUDA enabled 
>>> GPUs are also OpenCL enabled ones. Old AMD/ATi  GPUs support CAL but don't 
>>> support OpenCL.
>>> 
>>> 
>>> 
>>> Вторник, 29 января 2013, 22:28 -08:00 от Charlie Fenton < 
>>> [email protected] >:
>>>> I may be mistaken, but my understanding is that the ATI/AMD drivers which 
>>>> enable CAL on Windows and Linux also enable OpenCL for the ATI/AMD GPU.  
>>>> If that is the case, and if the OpenCL application is faster, then it 
>>>> seems that there should be no reason to have the CAL application at all.
>>>> 
>>>> Of course, CAL does not support Macs, but Apple's OpenCL implementation 
>>>> supports most modern ATI/AMD GPUs and is standard on current versions of 
>>>> OS X.  So it also makes sense to have OpenCL but not CAL applications for 
>>>> Macs.
>>>> 
>>>> Cheers,
>>>> --Charlie
>>>> 
>>>> --
>>>> Charlie Fenton  [email protected]
>>>> BOINC / SETI@home Macintosh & Windows Programmer
>>>> Space Sciences Laboratory
>>>> UC Berkeley
>>>> 
>>>> 
>>>> 
>>>> On Jan 29, 2013, at 4:27 PM, David Anderson wrote:
>>>> 
>>>>> The project should adjust its app_plan() function so that
>>>>> the estimated FLOPS for the OpenCL version
>>>>> is higher than that for the CAL version.
>>>>> -- David
>>>>> 
>>>>> On 29-Jan-2013 4:01 PM, Raistmer the Sorcerer wrote:
>>>>>> SETI@home has both OpenCL ATi and Brook+ (CAL) apps for one of its
>>>>>> subprojects. Currently  CAL app distributed to all CAL-capable ATi 
>>>>>> devices,
>>>>>> including OpenCL capable ones. As was shown with CUDA apps on SETI beta 
>>>>>> BOINC
>>>>>> can't select fastest app if few available. CAL app much slower than 
>>>>>> OpenCL
>>>>>> app. So, feature needed to allow server to send CAL app only to CAL-only
>>>>>> capable hosts (not send if host supports OpenCL too) and OpenCL app for
>>>>>> OpenCL+CAL/OpenCL hosts. AFAIK currently it''s not possible.
>>>> 
>>>> _______________________________________________
>>>> boinc_dev mailing list
>>>>  [email protected]
>>>>  http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
>>>> To unsubscribe, visit the above URL and
>>>> (near bottom of page) enter your email address.
>>> 
>>> _______________________________________________
>>> boinc_dev mailing list
>>>  [email protected]
>>>  http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
>>> To unsubscribe, visit the above URL and
>>> (near bottom of page) enter your email address.
>>
>
>_______________________________________________
>boinc_dev mailing list
>[email protected]
>http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
>To unsubscribe, visit the above URL and
>(near bottom of page) enter your email address.
>
>
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.

Reply via email to