[llvm] [GlobalISel] Change MatchTable entries to 1 byte each (PR #74429)

[Amara Emerson via llvm-commits]

aemerson wrote:

> > I then looked at your PR here: #74823 and measuring this resulted in a no statistically significant change in CTMark overall compilation times vs baseline (so no regression anymore), however I did see a 0.8% reduction in average cycles for the llc test, compared to the baseline. So it seems that PR brings a needed tiny perf improvement.
> 
> 
> 
> I think the measurement result of #74823 may be noise too, because that PR reduces less code size than this PR. 
> 
> In a word, the measurement result shows no regression here. So, I think we are OK to go further?

I think that's possible perhaps but unlikely. These measurements were from 40 runs each, directly reading PMU counters. The root cause for the improvement I don't know yet, but yes agreed that either way it's worth the binary size reduction.

https://github.com/llvm/llvm-project/pull/74429