github-jwinterm/xmr-stak
1
0
Fork 0
mirror of https://github.com/jwinterm/xmr-stak.git synced 2026-11-15 09:23:56 +01:00
Code Issues Packages Projects Releases Wiki Activity

Add warning and fallback when auto intensity is 0

Browse Source 
Occassionally the auto adjust doesn't find enough memory and the
intensity is detected too low and aligned to 0 with the compute units.
This patch fixes this situation by issuing a warning with a suggestion
to set environment vars and then ignoring the alignment to 0

Per several issues:

Principally:
https://github.com/fireice-uk/xmr-stak/issues/81

Related:
https://github.com/fireice-uk/xmr-stak/issues/490
https://github.com/fireice-uk/xmr-stak/issues/472
This commit is contained in:
Doug Johnson
2017-12-31 00:14:02 -07:00
committed by Doug Johnson
parent 7584ea8e7a
commit 0c845b3569
1 changed files with 13 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
xmrstak/backend/amd/autoAdjust.hpp
+13
View File
@@ -118,6 +118,19 @@ private:
size_t possibleIntensity = std::min( maxThreads , maxIntensity );
// map intensity to a multiple of the compute unit count, 8 is the number of threads per work group
size_t intensity = (possibleIntensity / (8 * ctx.computeUnits)) * ctx.computeUnits * 8;
//If the intensity is 0, then it's because the multiple of the unit count is greater than intensity
if (intensity == 0) {
/* See Issues:
* https://github.com/fireice-uk/xmr-stak/issues/81
* https://github.com/fireice-uk/xmr-stak/issues/472
* https://github.com/fireice-uk/xmr-stak/issues/490
* Note that it appears that Northern Islands GPUs (HD 6XXX) are unaffected by
* these environment variables, according to my testing (dougvj)
*/
printer::inst()->print_msg(L0, "WARNING: Autodetected intensity unexpectedly low. Try setting GPU_SINGLE_ALLOC_PERCENT and etc.");
intensity = possibleIntensity;
}
conf += std::string(" // gpu: ") + ctx.name + " memory:" + std::to_string(availableMem / byteToMiB) + "\n";
conf += std::string(" // compute units: ") + std::to_string(ctx.computeUnits) + "\n";
// set 8 threads per block (this is a good value for the most gpus)