Not luck; I've tried many times, including when tasks were plentiful. This computer has received 0 tasks, the other computer has received 52 tasks. Both are set to request tasks at intervals of no more than 7 minutes.

[snip]
I've found that when an output file is absent, the cause is usually one not reported in the event log, but is reported in a different output file.

You may need to copy all files in the slot into a different directory so you can check the other files for error messages.

I think it would be fair for the points to be the same per gflop, since more powerful processors(cpu or gpu) should get more points in the same amount of time. I do not really care so much about the points though. Definitely do not hold up the release to figure out the points! It can be worked on after we start chewing through the rest of OPN(G).

As for time spent, elapsed time is definitely better than cpu time. Really though, it would be fairer to treat the gpu like a cpu. The way that it is now does not treat them equally. Using one of my computers as an example(just cpu or just gpu):

4c/8t i7 x 1hr = 8 hrs credit. This averages ~5.8 jobs total in that hour.
24c igpu x 1 hr = 1hr credit. This averages ~66.3 jobs total in that hour

This illustrates that it is not just more powerful total, but it is actually more powerful per core too, since it does a lot more jobs per core. Yet, it gets treated as if it were a 1 core cpu. It really should be given time credit per core. I do understand that it is probably non trivial to actually make that happen though. A compromise could be to have a multiplier on it though. Meaning elapsed time x Multiplier = time credit given.

Just want to point out that running just the GPU results in way more science happening than running just the CPU or running both the CPU and GPU. But if I just run the GPU, it will mean way less time credit. I do plan on just running the GPU since it is better for science of course. But I can't help regretting the loss of time credit, I was going to get my first 20Y diamond badge, but now will not with just running the GPU with the current loss of time credit.

I replaced GPU in my Win10Pro desktop from RTX 2060 Super to ATI Radeon HD 5750.

On RTX 2060 Super I had no errors since OPNG Beta started.

On ATI Radeon HD 5750 I cannot complete any task. Results have Error/Invalid status. According to GPU-Z this card supports OpenCL 1.2 and I have newest possible drivers (Aug 03, 2015) from AMD support site. When antivirus software is enabled tasks end with Error status (tasks reach 99% in few minutes and 100% in 2-3 hours, 0% GPU load when processing). When I disable AV tasks end with Invalid status (tasks reach 100% in 1 minute, 0% GPU load and higher GPU/memory clocks when processing). I see it's a common problem with older ATI/AMD cards. There's a 100% GPU load in other projects with AV enabled.

I have gotten 39 WU's on my windows 10 machine with AMD RX 480, part of the Ellesmere family ( Ellesmere [Radeon RX 470/480/570/570X/580/580X]).
And 10 WU's on ManjaroLinux machine with AMD RX 560, part of Baffin family ( Baffin [Radeon RX 460/560D / Pro 450/455/460/555/555X/560/560X])

All were found to be valid. Looks like AMD is working in this beta. There can still be a problem, but not for everybody so it seems.