intermittent black screen

[bmxjumperc]

If you're able to provide more information about which PROJ, RUN, CLONE, GENs this occurs with and which GPU(s), as well as whether the core was running in OpenCL or CUDA mode, that would be super helpful in tracking down the offending kernel(s)!

~ John Chodera // MSKCC

I'm not sure if this is exactly what you're looking for? I don't know about RUN, CLONE, or GEN?
The primary monitor is on display-port.
Plenty of monitor blank-outs and here was the latest from the log:

Code: Select all

11:54:08:WU02:FS02:0x22:Completed 800000 out of 1250000 steps (64%)
11:55:29:WU02:FS02:0x22:Completed 812500 out of 1250000 steps (65%)
11:55:31:WU02:FS02:0x22:Checkpoint completed at step 812500
11:55:31:WU00:FS00:0xa7:Completed 140000 out of 500000 steps (28%)
11:56:52:WU02:FS02:0x22:Completed 825000 out of 1250000 steps (66%)
11:58:15:WU02:FS02:0x22:Completed 837500 out of 1250000 steps (67%)
11:59:01:WU00:FS00:0xa7:Completed 145000 out of 500000 steps (29%)
11:59:34:WU01:FS01:0x22:Completed 1760000 out of 2000000 steps (88%)
11:59:37:WU02:FS02:0x22:Completed 850000 out of 1250000 steps (68%)

I am not actually seeing these blank-outs in connection with TDR events.

[ajm]

Each "WU" (Work Unit) is part of
a project (PROJ), that is, a protein under study,
a run, that is, a particular simulation of that protein started from a specific situation (conformation),
a clone, that is, such a simulation at a specific initial velocity,
and a generation (GEN), that is a piece of that clone.

A WU is thus designated by four numbers called the PRCG numbers : Project (Run, Clone, Generation), eg. 16918 (139, 146, 0).

All the information John Chodera was talking about are in your log: viewtopic.php?p=327412&f=24#p327412

[bmxjumperc]

Thanks,

A reaction to overheating is not ruled out. I had been out of the f@h game for a while and was rusty on the tools such as MSI Afterburner to program the GPU fan curve. I am using higher GPU fan speeds and it is keeping the GPU about 15 C cooler.

Probably helpful

[bruce]

If you've been out for a while, there's a reasonable chance that your hardware has accumulated more dust than it needs in the cooling passages. (Zero is good ) Consider a careful internal cleaning of passages where air is supposed to cool things down.

[bmxjumperc]

If you've been out for a while, there's a reasonable chance that your hardware has accumulated more dust than it needs in the cooling passages. (Zero is good ) Consider a careful internal cleaning of passages where air is supposed to cool things down.

Yah, it was just the software tools to monitor the the GPU etc. that I wasn't using for a while. I'm not even sure if I ever F@H on this 2070 so I'm not all that familiar with its behavior.
I still have no idea what for certain is causing the monitor blank-outs. The innards of my tower are sparkly clean. The NZXT H700i seems to have quite good positive/negative pressure plus quite fine dust screens. I also use a DataVac on it once in a while.

We're looking into this! Last time we encountered this, there was a specific kernel call that was taking a bit too long to make the windows timeout. We might be able to easily fix this without requiring settings to be changed.

~ John Chodera // MSKCC

Looking forward to future details even if I only provoked research and refactoring than that's great too.

[Neil-B]

Windows has a intermittent black screen issue so it may not actually be a FaH issue as such just that FaH triggers/exacerbates a windows issue

[bmxjumperc]

Windows has a intermittent black screen issue so it may not actually be a FaH issue as such just that FaH triggers/exacerbates a windows issue

Thank you.