BOINC Manager tab "Properties" shows/says for "Estimated computation size" for a WU of type
GW --> 144,000 Flops Credit 1,000
GR --> 525,000 Flops Credit 3,465
What did Einstein say about "Time"?
Seems OK to me.
I guess the flop counts are so much less than for the GW tasks have to do with the specialized calculations that are being run on the much slower CPU. so that's why it takes so long for GW tass to run with such low flop counts.
from what i remember from last time i looked at some of the instructions being used in the GW app, it should be theoretically possible to shift those calculations to the GPU, and maybe even possible to shift them to the Tensor cores on newer Nvidia GPUs. that would GREATLY speedup calculations. but it will take effort from the devs to create the an application that will properly leverage these resources, and would probably have to be CUDA specific.
BOINC Manager tab "Properties" shows/says for "Estimated computation size" for a WU of type
GW --> 144,000 Flops Credit 1,000
GR --> 525,000 Flops Credit 3,465
What did Einstein say about "Time"?
Seems OK to me.
where do you see this? there is no tab in Boinc Manager labeled "Properties" i see [Notices/Projects/Tasks/Transfers/Statistics/Disk]
and the properties button on the Projects tab does not show this info.
--edit--
Found it. on the "Tasks" tab. select an individual task and then click the properties button on the button panel on the left side.
but I'd be interested to know where those values come from. if they are set by the project, or if they are estimated by BOINC (boinc isnt good at flop counting, especially GPUs).
Task FLOPS estimate is set by the project. Why people complain all the time when the estimated time to finish is so far off in some cases. If the project admins don't get it someplace close to "in the ballpark" to actuality, it can throw the smooth running of BOINC off course on a multi-project host.
One of Richard Haselgrove's pet peeves.
In the case of the GW tasks, it is very much so under the actual value of FLOPS needed to complete the calculation and is awarded much less credit per time. As you surmise, the fact that a lot of the calculation is done on the cpu but is not factored into the total FLOPS used is the problem.
San-Fernando-Valley
)
I guess the flop counts are so much less than for the GW tasks have to do with the specialized calculations that are being run on the much slower CPU. so that's why it takes so long for GW tass to run with such low flop counts.
from what i remember from last time i looked at some of the instructions being used in the GW app, it should be theoretically possible to shift those calculations to the GPU, and maybe even possible to shift them to the Tensor cores on newer Nvidia GPUs. that would GREATLY speedup calculations. but it will take effort from the devs to create the an application that will properly leverage these resources, and would probably have to be CUDA specific.
_________________________________________________________________________
Ian&Steve C.
)
Task FLOPS estimate is set by the project. Why people complain all the time when the estimated time to finish is so far off in some cases. If the project admins don't get it someplace close to "in the ballpark" to actuality, it can throw the smooth running of BOINC off course on a multi-project host.
One of Richard Haselgrove's pet peeves.
In the case of the GW tasks, it is very much so under the actual value of FLOPS needed to complete the calculation and is awarded much less credit per time. As you surmise, the fact that a lot of the calculation is done on the cpu but is not factored into the total FLOPS used is the problem.