Reputation: 1
The Intel® Xeon® Silver 4216 processor installed in the node supports the AVX-512 instruction set. When using AVX-512, how many FP32 operations can one core execute per clock cycle?
Hint: Consider the number of AVX-512 instructions executed per clock cycle and the number of FP32 operations per AVX-512 instruction.
When I looked up the specifications, I found the following:
Instruction set extensions: Intel® SSE4.2, Intel® AVX, Intel® AVX2, Intel® AVX-512
Number of AVX-512 FMA units: 1
Intel® Advanced Vector Extensions 512 (AVX-512), new instruction set extensions, delivering ultra-wide (512-bit) vector operations capabilities, with up to 2 FMAs (Fused Multiply Add instructions), to accelerate performance for your most demanding computational tasks.
I understand that AVX-512 can process 512/32 = 16 FP32 elements, but I'm not sure about how to interpret the phrase "two FMAs" - should I think of this as 2 additions and 2 multiplications, or should I just think of it as 2 operations(either addition or multiplication)?
Upvotes: 0
Views: 98