I am using Vtune 2022.
I see that it reports ONLY about the utilization of FP vectorization (and not integer vectorization).
How can I get the utilization of vectorization in my integer-based computations ?
Could you please let us know the analysis that you have ran to get the floating point vectorization.
Also, could you please explain the need for integer based computations, so that we can inform the concerned team regarding this request.
I am running a basic matrix multiplication where all matrices are int322 based.
I don't use FP in my app.
I know the internal loops are vectorized.
Vtune always reports ONLY about FP vectorization (which in my case it is obviously almost 0% since no FP values) and NOTHING about the utilization of INTEGER vectorization. Why is that ?
Thanks for reaching out to us.
Are you using HPC Performance Characterization Analysis feature of Intel VTune? Currently, this feature only supports vectorization for floating-point loops . For vectorization optimization opportunities, please use the Intel Advisor to run a vectorization analysis. More information on vectorization analysis can be found here.