That... is quite an improvement margin.
Damn.
That... is quite an improvement margin.
Damn.
"It is more efficient to compute the child index of the current node inside the parent node and write the bounds when available. The previous code could load up to 16 AABBs to compute the new ones. The new code also only needs 1/7 of the previously used scratch memory. The new code seems to be around 30% faster (0.5ms) in GOTG on a 6700XT."
Shitty missleading clickbait headline though. 0.5ms improvement probably wont translate to 30% more fps with raytracing on RDNA2 cards as the "30% faster on rdna2" in the headline suggest.
Sweet. I will ride out my RDNA2 card until it dies. While i'm not a fan of raytracing this will come in handy for the few RT only titles i may be interested in.
And this is while AMD is trying to abandon support for it and RDNA1 in Windows.
These improvements are for MESA, by an open source developer for Valve. So not in the AMD driver itself. Older AMD cards do well on linux, because AMD open sourced the driver and incorporated it in the kernel. My AMD experience on linux thus far has been that it just works. Valve is spending a lot of resources on improving the graphics stack on linux.
all 9 comments