GPUOpen On Reducing Vulkan Overhead With Volk, Possible 1~5% Savings

[phoronix]

Phoronix: GPUOpen On Reducing Vulkan Overhead With Volk, Possible 1~5% Savings

A guest post on AMD's GPUOpen blog outlines the overhead issues with using the Vulkan loader library and possible performance advantages to using vkGetDeviceProcAddr or more easily via a little heard of project called Volk...

http://www.phoronix.com/scan.php?page=news_item&px=Volk-Vulkan-Direct-Access

[ernstp]

I wonder what Feral is doing...

[tildearrow]

Typo:

MIT-licensed Vulkan meta-loder. By including

[dungeon]

Volkenstein

BTW, not sure what 1-5% savings mean? Is that mostly 1% or mostly 5% or should we think of average 3%

[dungeon]

I wonder what Feral is doing...

Feral need many more wolfs around, in range of +20% for sure... at least there is a rabbit to run after

I must laugh on this, as this is like a ZOO. Since VLK in some slavic languages means exactly wolf... so some people there read/understand it as AMD-WOLF likely

[GunpowaderGuy]

From the GitHub page : "volk is written in C89 and supports Windows, Linux and Android."
Please tell me that vulkano already does something like this and has no need to use volk

[schmidtbag]

BTW, not sure what 1-5% savings mean? Is that mostly 1% or mostly 5% or should we think of average 3%

To my understanding, it means the savings can be as little as 1% or as high as 5%. An average of 3% isn't that impressive, but saying "a 5% savings" could be misleading.
But... I do think saying "up to a 5% savings" would've sufficed.

[MartinPL]

WiLK = wolf in Polish 🐺

[aejsmith]

I wonder what Feral is doing...

We already do what this library does.

[stalkerg]

> We already do what this library does.

And what main problem now? I mean you use maximum optimizations but looks like this not enough.