Aras Pranckevičius is a user on You can follow them or interact with them if you have an account anywhere in the fediverse. If you don't, you can sign up here.

Aras Pranckevičius

@shivoa ohh, 2500K, yeah indeed time for an upgrade. My GPU impl might be "suboptimal" for nvidia, I guess (structured buffer usage perhaps?). On the Mac I'm also getting better results from Intel than from NV.

Daily Pathtracer part 10: small update to have all implementations on the same page again; and a note on Mono C# performance tweaks.

Next up: let's do a stream/buffers oriented path tracer setup! (why? I don't know)

Daily Pathtracer part 9: a wild ryg appears and optimizes my code!

Next up: put these optimizations into C#, C#/Burst, GPU implementations.

@KeefJudge I saw someone mention as an alternative web client, haven't tried it

@KeefJudge I mean, mastodon client (and server) source is on github. If you manage to land changes into official release, they will appear here eventually too :)

I blogged about compilation time – compiling with -std=c++98 is (sometimes) much faster than -std=c++11

@KeefJudge it's open source! which means... I dunno, maybe, if it breaks you can keep both pieces :)

Daily Pathtracer part 8, where I'm learning more SIMD, and add more spheres to the test scene.

@zenmumbler "the thing about internet is, no one knows you're a dog" -- I could have said I'm an expert on SIMD and there's no way to verify that :)

I do know some people who, for exact same amount of experience, would say that they are experts...

Daily Pathtracer is back! Part 7, where I'm learning about SIMD.

Next up: more SIMD.

New Unity Linux builds for 2017.4.0f1 and 2018.1.0b13 are available!

Release notes and downloads at

Daily Pathtracer, Part 6. DX11/HLSL implementation (still not optimized whatsoever).

Next up: will do a break for a couple days; sustaining daily posts is hard :)

@ScreennameTaken in my experience it just is (on some other workloads too). According to the internets, it can do more "math" indeed (Intel Iris Pro: 832 Gflops, GeForce GT 750M: 722 Gflops). It doesn't have that much memory bandwidth, so YMMV.

@jakob meh, that sounds like work. I don't like work :)

@jakob that's my takeaway from other benchmarks too; the Intel one is actually *better* at computing power, it only is slower if bottlenecked by memory bandwidth