New app seems slightly slower than old on 330M, but reasonably faster on 680MX. no validation issues noticed on a reasonable amount of tasks. I say go for it (and on Albert too maybe).
For my iMac's 680MX, running 2 at once, i'm getting ~10,000s rather than ~13,000s.
I do not know if this is more to do with the App or with the tasks however.
The application binary is actually identical, this is mostly a matter of deployment.
Eventually the CUDA 5.5 64Bit build will be integrated into our semi-automatic deployment system, but this hasn't been done yet. It will be a while until we find the time to do that.
Deploying CUDA applications by hand is pretty cumbersome and error-prone, I don't think I will do this for this app version.
I chose BRP5 for this app version as an experiment, because the supply of BRP4G work depends on the data that we get from Arecibo, and is therefore less predictable.
RE: Sorry, should be fixed
)
Awesome!!!! I have a GTX 750 crunching 3 at a time.
I intend to promote this app
)
I intend to promote this app version out of "Beta" status today. Any objections?
BM
BM
New app seems slightly slower
)
New app seems slightly slower than old on 330M, but reasonably faster on 680MX. no validation issues noticed on a reasonable amount of tasks. I say go for it (and on Albert too maybe).
EDIT: somehow double posted
)
EDIT: somehow double posted that...
further info then: both 330M and 680MX are on Yosemite systems with CUDA 6.5 installed.
Thank you for making this
)
Thank you for making this app. Will the speed also increase compared to the old app? By which amount?
Proud member of SETI.Germany
For my iMac's 680MX, running
)
For my iMac's 680MX, running 2 at once, i'm getting ~10,000s rather than ~13,000s.
I do not know if this is more to do with the App or with the tasks however.
All BRP5 tasks are absolutely
)
All BRP5 tasks are absolutely identical (except for the actual observatory data).
BM
BM
Will the BRP4G app be ported
)
Will the BRP4G app be ported to 64-bit?
The application binary is
)
The application binary is actually identical, this is mostly a matter of deployment.
Eventually the CUDA 5.5 64Bit build will be integrated into our semi-automatic deployment system, but this hasn't been done yet. It will be a while until we find the time to do that.
Deploying CUDA applications by hand is pretty cumbersome and error-prone, I don't think I will do this for this app version.
I chose BRP5 for this app version as an experiment, because the supply of BRP4G work depends on the data that we get from Arecibo, and is therefore less predictable.
BM
BM