We are still working on it. There is some progress, but we are not done yet. The work on an ABP1 CUDA App is currently a bit ahead of a S5R5 one, so that might come out first, but there is still a lot of work ahead. Please bear with us.
We are still working on it. There is some progress, but we are not done yet. The work on an ABP1 CUDA App is currently a bit ahead of a S5R5 one, so that might come out first, but there is still a lot of work ahead. Please bear with us.
BM
This question may not make sense since EAH is hosting the pulsar search, but your comment above brings it to mind. Any thought on enabling the preferences to allow for the ABP search only?
I realize you might risk losing some EAH participants (perhaps), but it seems to me the upside is you would get additional volunteers who would want to crunch the ABP data only.
It would be nice to see ATI Radeon HD 38xx and 48xx GPU clients as well. Milkyway went the ATI route as they needed double precision in their science and the nVidia GTX 2xx series did not meet this requirement. Now there is efforts to make their client CUDA GPU responsive.
Shih-Tzu are clever, cuddly, playful and rule!! Jack Russell are feisty!
We are still working on it. There is some progress, but we are not done yet. The work on an ABP1 CUDA App is currently a bit ahead of a S5R5 one, so that might come out first, but there is still a lot of work ahead. Please bear with us.
BM
This question may not make sense since EAH is hosting the pulsar search, but your comment above brings it to mind. Any thought on enabling the preferences to allow for the ABP search only?
No, the fact that you cannot opt-out of the LIGO data processing experiment (currently S5R5) was a deliberate choice. Project management recently decided to keep it this way (discussed in another thread).
As BOINC does not support \ have the skil to write a code that runs on AMD
cards we can only hope for a new Cluster Physik, I wish that he could be cloned :)
At the moment my single HD4870 card spiting out 10X the credit as my Q6600 and i7 put togeather. My creds have skyrocketing from 5000 to 50000 a day.:)
As BOINC does not support \ have the skil to write a code that runs on AMD
cards we can only hope for a new Cluster Physik, I wish that he could be cloned :)
At the moment my single HD4870 card spiting out 10X the credit as my Q6600 and i7 put togeather. My creds have skyrocketing from 5000 to 50000 a day.:)
Keap on crawling Einstein ;)
You have to remember that Nvidia supplied a couple of cards and the programming expertise to get cuda supported. ATI did not. That is why it isn't currently supported. ATI support is, according to the BOINC roadmap, slated for the 6.10 versions.
While we could debate the merits of either brand the reality is Nvidia put its money where its mouth is and ATI choose not to. If you want your ATI card supported why not ask the executives at ATI to provide resources so that ATI support can be added sooner?
It is already supporte by Folding@home and I crunch for them also
as they are kapable to write codes for both AMD and NVIDIA.
If BOINC chose to not include the chrunching power of AMD
just becaus they did not provide the card we can
put it to good use outside BOINC thank's to the Stanford group. ;)
I guess more important than getting hardware is direct support from NVIDIA experts and driver developers. Both ATI and NVIDIA will support the OpenCL standard in the future and we'll see more cross-GPU-platform projects then, but in the "first round" of GPGPU computing, NVIDIA clearly has offered better support for developers.
Any update on the Einstein
)
Any update on the Einstein cuda app?
Are you waiting on some other changes (such as boinc upgrades, CUDA 2.2 to settle down, etc) before releasing it for testing?
Anything we can do to help the cause?
Cheers
BOINC blog
Is it polite to make a bump
)
Is it polite to make a bump and a very polite request for some update?
We are still working on it.
)
We are still working on it. There is some progress, but we are not done yet. The work on an ABP1 CUDA App is currently a bit ahead of a S5R5 one, so that might come out first, but there is still a lot of work ahead. Please bear with us.
BM
BM
RE: We are still working on
)
This question may not make sense since EAH is hosting the pulsar search, but your comment above brings it to mind. Any thought on enabling the preferences to allow for the ABP search only?
I realize you might risk losing some EAH participants (perhaps), but it seems to me the upside is you would get additional volunteers who would want to crunch the ABP data only.
Just a thought.
Thanks!
It would be nice to see ATI
)
It would be nice to see ATI Radeon HD 38xx and 48xx GPU clients as well. Milkyway went the ATI route as they needed double precision in their science and the nVidia GTX 2xx series did not meet this requirement. Now there is efforts to make their client CUDA GPU responsive.
Shih-Tzu are clever, cuddly, playful and rule!! Jack Russell are feisty!
RE: RE: We are still
)
No, the fact that you cannot opt-out of the LIGO data processing experiment (currently S5R5) was a deliberate choice. Project management recently decided to keep it this way (discussed in another thread).
CU
Bikeman
As BOINC does not support \
)
As BOINC does not support \ have the skil to write a code that runs on AMD
cards we can only hope for a new Cluster Physik, I wish that he could be cloned :)
At the moment my single HD4870 card spiting out 10X the credit as my Q6600 and i7 put togeather. My creds have skyrocketing from 5000 to 50000 a day.:)
Keap on crawling Einstein ;)
RE: As BOINC does not
)
You have to remember that Nvidia supplied a couple of cards and the programming expertise to get cuda supported. ATI did not. That is why it isn't currently supported. ATI support is, according to the BOINC roadmap, slated for the 6.10 versions.
While we could debate the merits of either brand the reality is Nvidia put its money where its mouth is and ATI choose not to. If you want your ATI card supported why not ask the executives at ATI to provide resources so that ATI support can be added sooner?
BOINC blog
It is already supporte by
)
It is already supporte by Folding@home and I crunch for them also
as they are kapable to write codes for both AMD and NVIDIA.
If BOINC chose to not include the chrunching power of AMD
just becaus they did not provide the card we can
put it to good use outside BOINC thank's to the Stanford group. ;)
I guess more important than
)
I guess more important than getting hardware is direct support from NVIDIA experts and driver developers. Both ATI and NVIDIA will support the OpenCL standard in the future and we'll see more cross-GPU-platform projects then, but in the "first round" of GPGPU computing, NVIDIA clearly has offered better support for developers.
CU
Bikeman