correct typo (backport #18750) (#18751)

* correct typo

(cherry picked from commit 1507477306)

* fix another typo

(cherry picked from commit 269028360c)

Co-authored-by: Yang Li <yangli1990@live.com.au>
This commit is contained in:
mergify[bot]
2021-07-18 13:52:57 +00:00
committed by GitHub
parent 19049ca91b
commit 927057df26

View File

@@ -94,7 +94,7 @@ attempts to use a float operation that is not supported, the runtime will report
an unresolved symbol error.
Float operations are performed via software libraries, specifically LLVM's float
builtins. Due to be software emulated they consume more compute units than
builtins. Due to the software emulated they consume more compute units than
integer operations. In general, fixed point operations are recommended where
possible.
@@ -108,7 +108,7 @@ To run the test, sync the repo, and run:
Recent results show the float operations take more instructions compared to
integers equivalents. Fixed point implementations may vary but will also be
less then the float equivalents:
less than the float equivalents:
```
u64 f32