@doktor83 I'm pretty sure that you are aware of all these occasions, I mean with the hashdrop.
Cards with 4gb memory dont have this issue, at least to me. They reach maximum hashspeed immediately.
The problem is with the 8gb cards.
Whenever I start your miner one of them, sometimes more, can go up to maximum, others no and they stay low all the time
even after 2,3,4 hours of mining.
Next time I'm starting your miner some other cards or card do the opposite thing. It's a random situation.
To help things out, you of course the developer of a great miner to me, I found out that running GPU-Z at the time your miner runs
and passing throu all cards one by one and then exiting, voula all cards running full speed. Problem solved.
But of course this is not a solution!!!
Maybe something with the interrupts?
I'm trying to help!
I wanted to try this, and I got 1000 h/s across the board
just by firing up GPU-Z.. Awesome. Wish I had known this earlier! The GPU-Z launch was unusually slow... Very strange! Whatever it is doing to the gpu's, SRB can do too, this gives me hope.
GPU-Z is 2.8.0, and I'm ignoring the update to 2.9.0 in case it works differently, lol.
Hi, could you please explain a bit detail about steps to achieve it? Did you run GPU-Z right before ran SRB? Or while you clicked start SRB Miner, you then ran GPU-Z? Thanks.
Gentlemen,
things are very simple.
You have started SRBminer waiting a while, 2-3 minutes, to report the first hashrate total and give it some time to reach the top hashspeed as doktor83 says.
Then you realize the drop on some or all of your cards.
Now it's the time to fire up GPU-Z as miner runs and uses all the resources of GPUs.
I'm passing throu all GPUs one by one and then exiting from GPU-Z. If you stay running it you'll realize that there's a small drop in hashrate on all cards.
This is normal because of the memory resource usage of the 2 programs.
The great thing is that GPU-Z motivates SRBminer to reach the highest it can get.
As a programmer, I'm suspecting something with the interrupts or the memory management of GPUs.
For some cards it may need to run it a second time or and a third one.
Keep in mind that all this has to do with the RX 480-580
8gb cards which of course I run them with
2 threads and intensity 54, worksize 8.
I reach on them from 1070 to 1148 h/s.
There is
no problem at all with RX 570-580
4gb cards. These cards, of course with 1 thread - worksize 8 - intensity 60-61, reach maximum hashing speed at once.
I'm getting on these cards from 755 to 780 h/s.
Sorry, I've forgotten to mention that all these are on
heavy algorithm
I'm glad to help the community!
and waiting for a better miner from Doktor83.
GPU-Z didn't use any GPU's resources. it just connects to GPUs and monitors them.
All this connected to low-level accessing to GPUs via AMD ADL or other low-level driver such as phymem or like this.
I noticed that effect about year ago. It happens on GGS miner too. And not only GPU-Z has the same effect ))). Some hardware monitoring tools can do that too. Like HWInfo, but not all versions. It's exists method even without any tools to do that, but only for GPU wich display connected to.
Must to notice that Claymore's miners never had that issue. His miners allways runs at max speeds. This another one proof that Claymore knows more than others on AMD GPU programming. Looks like he realy worked or works now for AMD.