I tried the previous v10.18.10.4061 driver on an HD 4000, and it was worse than the recommended (by us - for Einstein) v10.18.10.3621: slower,and more invalid results.
Perhaps you could test this one, and report the results in the Binary Radio Pulsar Search (Parkes PMPS XT) "BRP6" thread, since that's the primary focus for getting these apps/drivers/hardware incompatibilities sorted out.
I tried the previous v10.18.10.4061 driver on an HD 4000, and it was worse than the recommended (by us - for Einstein) v10.18.10.3621: slower,and more invalid results.
First 2 tasks completed and pending. BRP4 work units for the moment. All on this host with a sent time of 11 Apr 2015 @ 14:23:09 UTC.
Will let it run through them and see how they go before trying the 1.52 beta app.
Initial thoughts are its slower than the 3621 driver by about 190 sec/wu (3621 driver around 689 sec, 4176 driver around 879 sec). Need the rest of the wu to complete and validate.
As a final post on this subject, I did manage to run some of the (beta) BRP6 work units. Only about 3 before they took it out of beta so hardly a decent sample but I would say this one might actually work. I posted about it in the Tech News forum BRP6 beta thread.
I've gone back to the 3621 driver on the test machine as its faster.
Intel 10.18.10.4176 driver for HD 4000 graphics?
)
I tried the previous v10.18.10.4061 driver on an HD 4000, and it was worse than the recommended (by us - for Einstein) v10.18.10.3621: slower,and more invalid results.
Perhaps you could test this one, and report the results in the Binary Radio Pulsar Search (Parkes PMPS XT) "BRP6" thread, since that's the primary focus for getting these apps/drivers/hardware incompatibilities sorted out.
RE: I tried the previous
)
Running down the cache on one machine now
BOINC blog
First 2 tasks completed and
)
First 2 tasks completed and pending. BRP4 work units for the moment. All on this host with a sent time of 11 Apr 2015 @ 14:23:09 UTC.
Will let it run through them and see how they go before trying the 1.52 beta app.
Initial thoughts are its slower than the 3621 driver by about 190 sec/wu (3621 driver around 689 sec, 4176 driver around 879 sec). Need the rest of the wu to complete and validate.
BOINC blog
All except 1 pending have
)
All except 1 pending have validated. Will try the beta app now.
BOINC blog
As a final post on this
)
As a final post on this subject, I did manage to run some of the (beta) BRP6 work units. Only about 3 before they took it out of beta so hardly a decent sample but I would say this one might actually work. I posted about it in the Tech News forum BRP6 beta thread.
I've gone back to the 3621 driver on the test machine as its faster.
BOINC blog