Well that's just absurd. Who's responsible for a silly idea like that? The progress information is just totally bogus and therefore worthless.
That would be the Boinc developers and the "feature" has been around for a few years now. I seem to remember the reasoning was that participants of various projects got upset because Boinc didn't report any progress for certain tasks and it was decided that a simulated progress report was better than no report at all.
Once a task actually reports some progress to Boinc the simulated progress is stopped and the actual progress value is displayed. Normally this only shows up during the first few minutes of a task running but if the tasks is running very slow or isn't making any progress then it can be quite misleading.
Once you get a feel for how tasks progress on your machine it's quite easy to distinguish between the simulated and the real progress.
Sorry for going on about this but I'm still trying to get my head around this. In my case at least, the "Gravitational Wave search O2 Multi-Directional GPU" application takes about 50 times longer than the "Gamma-ray pulsar binary search #1 on GPUs" application. That's not five times longer, which would be surprising in itself, but ten times that.
Despite this, the GW jobs award (presumably) only slightly more than one quarter as much credit. I don't care too much about credit but I assume that the awarded credit is based on the amount of work done so it appears to be out by a factor of about 200. And if the newer faster graphics cards can chew through a GW job in something like an hour then they'd be capable of completing the Gamma-ray pulsar searches in about a minute.
Or is this factor of 50 times not actually universally the case across all hardware? Is there some sort of massive bottleneck or deadlock that only occurs on slower or certain types of GPUs? If the slower GPUs are not up to the task then the scheduler should be excluding them when assigning these jobs.
By the way, earlier in this year there was the GPU version of the O1 GW search which ran quite well on my GPU. One could argue that it's not the same thing but, really, is such a massive difference to be expected? As far as credit awarded, they are the same thing supposedly.
Or is this factor of 50 times not actually universally the case across all hardware?
Your Radeon HD 7700 has GCN 1.0 architecture. Based on user experiences so far, all AMD cards with GCN 1.0 architecture have been observed to be incompatible with this current 'Gravitational Wave search O2 Multi-Directional GPU' app. That's why that card isn't making any real progress and a task will end up crashing in a way or another.
There's been talk about this under Cruncher's Corner and Problems and Bug Reports.
Your Radeon HD 7700 has GCN 1.0 architecture. Based on user experiences so far, all AMD cards with GCN 1.0 architecture have been observed to be incompatible with this current 'Gravitational Wave search O2 Multi-Directional GPU' app. That's why that card isn't making any real progress and a task will end up crashing in a way or another.
There's been talk about this under Cruncher's Corner and Problems and Bug Reports.
Okay thanks. This is an answer that I can deal with.
It seems that only AMD cards on Windows machines are validating against each other.
Not sure I understand exactly what you meant, but my last two validations on my Linux host running ATI cards were with Windows hosts, one with a NVIDIA card, the other with an ATI card. From about 5000 validations, I have no invalids. Woo hoo! Knock on wood.
Ideas are not fixed, nor should they be; we live in model-dependent reality.
Mr Anderson wrote:Well that's
)
That would be the Boinc developers and the "feature" has been around for a few years now. I seem to remember the reasoning was that participants of various projects got upset because Boinc didn't report any progress for certain tasks and it was decided that a simulated progress report was better than no report at all.
Once a task actually reports some progress to Boinc the simulated progress is stopped and the actual progress value is displayed. Normally this only shows up during the first few minutes of a task running but if the tasks is running very slow or isn't making any progress then it can be quite misleading.
Once you get a feel for how tasks progress on your machine it's quite easy to distinguish between the simulated and the real progress.
Sorry for going on about this
)
Sorry for going on about this but I'm still trying to get my head around this. In my case at least, the "Gravitational Wave search O2 Multi-Directional GPU" application takes about 50 times longer than the "Gamma-ray pulsar binary search #1 on GPUs" application. That's not five times longer, which would be surprising in itself, but ten times that.
Despite this, the GW jobs award (presumably) only slightly more than one quarter as much credit. I don't care too much about credit but I assume that the awarded credit is based on the amount of work done so it appears to be out by a factor of about 200. And if the newer faster graphics cards can chew through a GW job in something like an hour then they'd be capable of completing the Gamma-ray pulsar searches in about a minute.
Or is this factor of 50 times not actually universally the case across all hardware? Is there some sort of massive bottleneck or deadlock that only occurs on slower or certain types of GPUs? If the slower GPUs are not up to the task then the scheduler should be excluding them when assigning these jobs.
By the way, earlier in this year there was the GPU version of the O1 GW search which ran quite well on my GPU. One could argue that it's not the same thing but, really, is such a massive difference to be expected? As far as credit awarded, they are the same thing supposedly.
Mr Anderson wrote:Or is this
)
Your Radeon HD 7700 has GCN 1.0 architecture. Based on user experiences so far, all AMD cards with GCN 1.0 architecture have been observed to be incompatible with this current 'Gravitational Wave search O2 Multi-Directional GPU' app. That's why that card isn't making any real progress and a task will end up crashing in a way or another.
There's been talk about this under Cruncher's Corner and Problems and Bug Reports.
Richie wrote:Your Radeon HD
)
Okay thanks. This is an answer that I can deal with.
Validate errors are back
)
Validate errors are back
Since the 5th I've got 13 of
)
Since the 5th I've got 13 of these validate errors
https://einsteinathome.org/workunit/433164893Betreger wrote:Since the 5th
)
I see only one validate error for you. Do you have other accounts?
https://einsteinathome.org/host/6654821/tasks/5/0
It seems that only AMD cards on Windows machines are validating against each other.
They still have work to do. But my RX 570 died, so I am out until the next generation.
(Or unless they get a CUDA version for Nvidia cards.)
Jim1348 wrote:It seems that
)
Not sure I understand exactly what you meant, but my last two validations on my Linux host running ATI cards were with Windows hosts, one with a NVIDIA card, the other with an ATI card. From about 5000 validations, I have no invalids. Woo hoo! Knock on wood.
Ideas are not fixed, nor should they be; we live in model-dependent reality.
Getting a few validate errors
)
Getting a few validate errors during the weekend:
https://einsteinathome.org/workunit/436287854
https://einsteinathome.org/workunit/436353150
https://einsteinathome.org/workunit/436356291
https://einsteinathome.org/workunit/436389219
https://einsteinathome.org/workunit/436426137
https://einsteinathome.org/workunit/436453510
https://einsteinathome.org/workunit/436287131
And more reports on other parts of the forum.
Yep, I've gotten 6 since the
)
Yep, I've gotten 6 since the 24th.
Like yours they all show this status "Tasks are pending for this workunit." It seems odd they aren't be sent out again.