# Jeff Bezanson remarks on LLVM "getting slower and slower"

**URL:** https://discourse.julialang.org/t/jeff-bezanson-remarks-on-llvm-getting-slower-and-slower/39637
**Category:** Internals & Design
**Created:** [May 17, 2020, 2:37pm UTC](https://discourse.julialang.org/t/jeff-bezanson-remarks-on-llvm-getting-slower-and-slower/39637 "2020-05-17T14:37:52Z")
**Posts on this page:** 1
**Showing post:** 15

<div class="post-metadata">

### Author: ![Marc.Cox](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/marc.cox/32/7514_2.png) [@Marc.Cox](https://discourse.julialang.org/u/Marc.Cox)
#### Post date: [April 12, 2021, 12:12am UTC](https://discourse.julialang.org/t/jeff-bezanson-remarks-on-llvm-getting-slower-and-slower/39637/15 "2021-04-12T00:12:43Z")

</div>

Thx @ImreSamu,

Support for A JIT compiler should operate on the Pareto rule principle  
"… for many outcomes, roughly 80% of consequences come from 20% of the causes (the “vital few”).[1] "  
@@ [Pareto principle - Wikipedia](https://en.wikipedia.org/wiki/Pareto_principle)

Google Engineers contributed to LLVM 12 RC1 making  
“Machine Function Splitter” code generation optimization  
pass for splitting code functions into hot and cold parts.

They are doing this stemming from their research that in  
roughly half of code functions that  
**MORE than 50% of the code bytes are NEVER executed**  
**but generally loaded** into the CPU’s data cache.

[Google Engineers Propose “Machine Function Splitter” For Faster Performance](https://www.phoronix.com/scan.php?page=news_item&px=Machine-Function-Splitter)

---

_[View the full topic](https://discourse.julialang.org/t/jeff-bezanson-remarks-on-llvm-getting-slower-and-slower/39637)._