Skip to content

Commit

Permalink
Update README.md
Browse files Browse the repository at this point in the history
  • Loading branch information
vaishakp authored Jun 26, 2024
1 parent dcf76b7 commit 17e85e1
Showing 1 changed file with 1 addition and 1 deletion.
2 changes: 1 addition & 1 deletion README.md
Original file line number Diff line number Diff line change
Expand Up @@ -163,7 +163,7 @@ The following can be observed when the above code is run (with e.g. `gcc-13.3`):
![image](https://github.com/vaishakp/vaishakp.github.io/assets/36019754/dbb09c06-502c-4a47-b1be-52f3737c496e)

Notes:
1. It can be seen that when fma is turned on (LHS), native SIMD instructions (`vfmsub132ss`, `vfmadd132ss`, etc.) are used to carry out FMA.
1. It can be seen that when `fma` is turned on (LHS), native SIMD instructions (`vfmsub132ss`, `vfmadd132ss`, etc.) are used to carry out FMA.
2. On the RHS, `fma` was turned off with `-ffp-contract=off`. Here, only scalar operations are used (`vmulss`, `vsubss`, etc.).
3. The avx2 xmm registers are being used in both cases.
4. The code with FMA on (i.e. LHS) completes the execution with fewer instructions because of FMA.
Expand Down

0 comments on commit 17e85e1

Please sign in to comment.