FGRPB1G now shows on the server status page as 36 Millions units remaining:
FGRPB1G search progress
Total needed
Already done
Work still remaining
175,921,402 units
139,835,162 units
36,086,240 units
2617.4 days
2163.2 days
454.2 days (estimated)
Keith Myers wrote:
No not official. They just have advised that the Gamma Ray Pulsar search was winding down back in summer and only a few more months of work was available. Then because they didn't get the BRP7 campaign ramped up as fast as they expected, they prolonged the GR#1 task generation for bit longer.
The scarcity of tasks is just because they haven't recovered the computers that do the pre-processing of those tasks from the last system wide project failure.
The end of these tasks could come at a moment's notice however. I would start making plans for other work for your host.
Speaking of this, we are going to be mainly offline for the next two weeks due to the holidays. The AC is usually not on in the buildings during this time, so I will only leave the cooler running hosts on.
I am not saying we (Boca Raton HS) have much of an impact on the overall timeline of these tasks, but we know every little bit helps.
No obvious pattern on the invalids after a quick look through the results. Mostly Windows CUDA wingmen, but also a handful of invalids against Windows AMD and Linux hosts.
I've noticed recently on the status page in the BRP section that strange negative values started showing up for Arecibo again
Does this mean arecibo data needs re-crunching ? GBT times started to fall behind too - wondering why....
I've noticed recently on the status page in the BRP section that strange negative values started showing up for Arecibo again
Does this mean arecibo data needs re-crunching ? GBT times started to fall behind too - wondering why....
GBT
22,031,744
994,729
11,210
89
Arecibo
140,611,327
-749,721
201,419
-4
I'm guessing the GBT times are because of this and will start changing if they really have run out of tasks:
I've noticed recently on the status page in the BRP section that strange negative values started showing up for Arecibo again
Does this mean arecibo data needs re-crunching ? GBT times started to fall behind too - wondering why....
GBT
22,031,744
994,729
11,210
89
Arecibo
140,611,327
-749,721
201,419
-4
Until now all available ressources for BRP4 were used for GBT, so it is slowing down now because part of those ressources are used for Arecibo tasks now. The total of Arecibo tasks has not been increased when the new ones were released, that's why the script is showing weird values. At least that's my theory about what we see there.
I have to admit that the BRP progress counting doesn't work so well, in particular the communication with the pre-processing. The GBT beams that aren't finished yet were sorted out by the pre-processing, because it was detected that these wouldn't process well on E@H (too much RFI). OTOH we found some more Arecibo beams needed to be processed (again), these are not yet counted in the total Arecibo beams. I'll fix that as soon as I get the necessary information.
FGRPB1G now shows on the
)
FGRPB1G now shows on the server status page as 36 Millions units remaining:
FGRPB1G search progress
Filipe wrote: FGRPB1G now
)
Nice to see that tracker back. Let's get it done!
_________________________________________________________________________
Astonishing! With all our
)
Astonishing! With all our combined GPU power is going to take about 1 year to finish them off?
Speaking of this, we are
)
Speaking of this, we are going to be mainly offline for the next two weeks due to the holidays. The AC is usually not on in the buildings during this time, so I will only leave the cooler running hosts on.
I am not saying we (Boca Raton HS) have much of an impact on the overall timeline of these tasks, but we know every little bit helps.
I am back running MeerKAT on
)
I am back running MeerKAT on a pair of Radeon VIIs. The invalid rate is about 5-6%. Is this invalid rate within the expected range? https://einsteinathome.org/host/12883788/tasks/5/0
No obvious pattern on the invalids after a quick look through the results. Mostly Windows CUDA wingmen, but also a handful of invalids against Windows AMD and Linux hosts.
Filipe wrote: Astonishing!
)
Obviously it would be alot faster if they were the only tasks available
I've noticed recently on the
)
I've noticed recently on the status page in the BRP section that strange negative values started showing up for Arecibo again
Does this mean arecibo data needs re-crunching ? GBT times started to fall behind too - wondering why....
Marcin wrote: I've noticed
)
I'm guessing the GBT times are because of this and will start changing if they really have run out of tasks:
O3MD search progress
Marcin wrote:I've noticed
)
Until now all available ressources for BRP4 were used for GBT, so it is slowing down now because part of those ressources are used for Arecibo tasks now. The total of Arecibo tasks has not been increased when the new ones were released, that's why the script is showing weird values. At least that's my theory about what we see there.
.
I have to admit that the BRP
)
I have to admit that the BRP progress counting doesn't work so well, in particular the communication with the pre-processing. The GBT beams that aren't finished yet were sorted out by the pre-processing, because it was detected that these wouldn't process well on E@H (too much RFI). OTOH we found some more Arecibo beams needed to be processed (again), these are not yet counted in the total Arecibo beams. I'll fix that as soon as I get the necessary information.
BM