Dan Gohman

Thanks @chadaustin, I haven't had time to look at everything in detail, but it looks really helpful!

@juj: The purpose of my request for benchmarks and testcases was to allow me to evaluate how good our current design is. If we get data and it exposes minor...

@juj: I should also mention that the SSE1 tests you have here look like really great tools, and I'm definitely looking forward to using them. Being synthetic benchmarks, they'll give...

My proposal above conveniently omitted mention of how the iterations of a SIMD.Long operation might be ordered :). I agree that it's desirable to coordinate with the C++ committee here...

Oops. Anyone have a suggestion for the best way to fix this?

Yes, our (Firefox) current implementation canonicalizes NaNs at the point when a scalar is extracted from a SIMD value or when a scalar is loaded from a typed array. I...

This issue is actually causing two tests in the "float32x4 int32x4 bit conversion" test to fail on SpiderMonkey. Currently, bitcasting from int32x4 to float32x4 and back to int32x4 is not...

This turns out to be a problem for emulating intrinsics like _mm_cmpeq_ps in too. Since it's defined to return __m128, which is float32x4, we bitcast the int32x4 result to float32x4,...

Yes, I can arrange this. My understanding is that we have pretty good coverage in the 'Float32x4 operators' and similar tests already, so the only thing we're missing are the...

My big-picture view is that the current iteration of SIMD.js is primarily about delivering basic functionality and establishing how SIMD integrates with JS in general, since that's a fairly complex...