3 ms·
My understanding is that when you use SSE/AVX instructions to perform floating point operations (and nowadays most compilers do when compiling for x86_64, even
by simonster 12y ago
My understanding is that when you use SSE/AVX instructions to perform floating point operations (and nowadays most compilers do when compiling for x86_64, even for code that isn't vectorizable), then Intel CPUs are what this article "single/double systems" and not "extended-based systems." But I'm not an FP expert, so please correct me if I'm wrong.
- AaronFriel 12y agoI believe that is correct, as the SSE/AVX instructions specify double arguments, and it's the not-quite deprecated x87 extension that supports 80-bit extended-precision floats.
- Zardoz84 12y agoSSE2 or superior