Quote from: Ledipaa on November 13, 2017, 09:49:43 PM

Quote from: Morgon on November 13, 2017, 09:02:04 PM

Quote from: soothaa on November 13, 2017, 08:39:58 PM

{ "index" : 0, "intensity" : 2016, "worksize" : 8, "affine_to_cpu" : false },
{ "index" : 0, "intensity" : 1700, "worksize" : 8, "affine_to_cpu" : false },

Are these Vega 56 or 64? When I was using Stak on my 56s, I had my second thread worksize at 1600.

I might suggest testing out CastXMR, even if just long enough to rule out a configuration issue versus a system issue.

Quote from: dallase on November 13, 2017, 08:59:30 PM

Hmm, I have 5 vega 56s running no problem on a 64GB SSD. Virt size is 32GB, and I have 16GB of physical. Getting 1800+ on ea

Interesting. What's your physical RAM usage when you have a workload? You prompted me to look at my own system, and I am noticing that with 4 GPUs, my VirtMem commit is only 32GB... maybe we only need 8GB Virtual per card?

Edit: Actually I'm even more confused. Maybe I'm doing something wrong, but following this guide to check my pagefile usage reveals 0 actual use. I wonder how we got here to thinking that we needed a high Virtual Memory counter for things to work properly.

Its probl from the time where we thought you would need lots or RAM to properly use HBCC. And now that it doesnt matter how much ram you have, since HBCC has nothing to do with vegas hashrate.

I also had ["cl out of host memory"] error when my page file was managed by windows. I set pagefile at 32000 and no more errors.

I'm not saying 32GB is right or wrong, I just picked it because thats about all I had to work with on a 64GB SSD. And that cleared up that message.