Hi there. This page seems to be stuck from last year. Is there any work being done to get this back on track? I'd like to see how much progress there is on the project.
Thanks!
http://einstein6.aei.uni-hannover.de/EinsteinAtHome/download/progress/BRP4-progress/
Copyright © 2024 Einstein@Home. All rights reserved.
Now validator is stuck too
)
Now validator is stuck too
According to server status
)
According to server status page it's disabled, so intentionally stopped and not just stuck I guess. They have moved some of the GPU applications from BRP4G back to BRP4, maybe that's the reason?
.
Is it possible to finish BRP4
)
Is it possible to finish BRP4 under 10 years with current speed boost?
I do BRP4 in ten and a half
)
I do BRP4 in ten and a half minutes on my Nano. Under 3 hours on my Pi4.
So much, much less than ten years.
and even faster if we can get
)
and even faster if we can get a CUDA or OpenCL app for Nvidia on x86. currently only Intel GPU supported for BRP4. or the custom CUDA/arm app floating around.
might have to snag one of the new Intel ARC GPUs to try it out if they don't come out with more apps for Nvidia/AMD.
I've already tested BRP4 on my Intel Xe graphics (iGPU of i7-1165G7) and performance is fairly impressive vs previous gen Intel iGPUs, about 2x as fast. even faster than the Nvidia Jetson Nano that Keith has.
_________________________________________________________________________
I haven't receive a BRP4G
)
I haven't receive a BRP4G since I inquired the status of BRP4 from Bernd and he informed me the focus was on the new Arecibo data they are pushing through BRP4 applications.
They actually moved the data over from BRP4G to BRP4.
Keith Myers wrote: I haven't
)
yeah it doesn't make too much difference since the BRP4 and BRP4G WUs are essentially identical in processing. the GPU tasks were just several bundled together and ran longer. the same exact application works for both as we discovered. the project probably just decided that they didnt need to maintain separate pools for these tasks and just process them all as BRP4 only.
my laptop Intel Xe iGPU (30w total power consumption for whole laptop) did the BRP4G in just under an hour, and the BRP4 in about 200 seconds. I have to imagine that a substantially more powerful discrete GPU would blast through the normal small BRP4 "small" tasks in 10's of seconds lol.
_________________________________________________________________________
I meant all BRP data, not one
)
I meant all BRP data, not one task
Keith Myers wrote: I do BRP4
)
Close to 8 minutes on Jetson TX1 and 5 minutes on Jetson TX2.
Jetson Tk1 is not so impressive - a bit more that 13 minutes.
Yes, the later versions are
)
Yes, the later versions are much faster. Still I think a little over ten minutes for paltry 128 cores is pretty impressive.
I really should have purchased the Xavier NX module for $399 when it first came out to upgrade my Nano.