iMac Pro The real FLOPS & Vector use on the iMac Pro

[bxs]

I'm interested in knowing how the Processor hardware embedded Vector instruction set improves the FLOPS number from the Apple's claimed number on their iMP web page ?

I suspect Geekbench 4 does not, at this time, have its code recompiled to employ the vector instructions and therefore will only show/display Apple's FLOPS claim.

I guess I can write a simple program to exploit the new vector instructions and have it report the FLOPS to compare against what Geekbench 4 reports.

Thoughts.... ?

 

[jerwin]

https://www.geekbench.com/doc/geekbench4-cpu-workloads.pdf

Floating Point Workloads GEMM

GEMM (General Matrix Multiplication) computes the result C = AB + C, where A, B, and C are matrices.

The GEMM workload uses 512x512 single-precision matrices. The GEMM implementations are written using hand-tuned vector instructions for the following instruction sets:

• AVX + FMA
• AVX
• SSE3
• SSE2

• ARMv8 NEON • ARMv7 NEON

so-- I think that this is already incorporated. see https://www.intel.com/content/www/us/en/architecture-and-technology/avx-512-overview.html

With ultra-wide 512-bit vector operations capabilities, Intel® AVX-512 can handle your most demanding computational tasks.

Applications can pack 32 double precision and 64 single precision floating point operations per second per clock cycle within the 512-bit vectors, as well as eight 64-bit and sixteen 32-bit integers, with up to two 512-bit fused-multiply add (FMA) units, thus doubling the width of data registers, doubling the number of registers, and, doubling the width of FMA units, compared to Intel® Advanced Vector Extensions 2.0 (Intel® AVX2).3

Is there an anomalously high GEMM sub-score among cpus that support avx-512?