Zink OpenGL-Over-Vulkan Driver - Performance Is Turning Out Better Than Expected

When looking at the performance of Zink's OpenGL over Vulkan implementation just about one year ago the performance had a lot to be desired. But since then they have patches bringing it all the way to OpenGL 4.6 compared to the OpenGL 2.1 days and there has also been a lot of work on the performance. The performance at least for select operations is now turning out better than even the developers were expecting.

Mike Blumenkrantz formerly of Samsung OSG and Enlightenment fame who has been devoting much time to Zink in recent months came out with some new performance figures. He was curious about how well Zink atop Intel's "ANV" Vulkan drivere compares to Intel's modern Iris Gallium3D OpenGL driver.

Blumenkrantz noted, "Both runs used the same base checkout from mesa, so all the core/gallium/nir parts were identical. The results weren’t what I expected...I’m not claiming zink is faster than a native GL driver, only that for some cases, our performance is oddly better than I expected."

When looking at the performance on a micro-level, there are cases where Zink is either faster or at least very competitive to Intel's dedicated OpenGL driver. Blumenkrantz went on to note:

is massively better than zink in successful test completion, with a near-perfect 99.4% pass rate compared to zink’s measly 91%, and that’s across 2500 more tests too. This is important also since timediff only compares between passing tests.

With that said, somehow zink’s codepath is significantly faster when it comes to dealing with high numbers of varying outputs, and also, weirdly, a bunch of dmat4 tests, even though they’re both using the same softfp64 path since my icelake hardware doesn’t support native 64bit operations.

But there were a lot of cases still where Zink is much slower than the real OpenGL driver. See more details on Mike's blog.

In any case these micro-benchmarks are interesting. Coming up shortly I'll run some Zink vs. OpenGL driver benchmarks on Intel and Radeon hardware with different GL games for seeing how the performance compares for the more complex workloads.