[IMPALA-12314] Revisit pre-compiled IR at O1

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: Impala 4.3.0
• Component/s: Backend
• Epic Color: ghx-label-12
• Assignee: Michael Smith
• Reporter: Michael Smith
• Created: 25/Jul/23 16:45
• Updated: 09/Aug/23 23:43
• Resolved: 07/Aug/23 21:52

Description

From https://github.com/apache/impala/blob/4.2.0/be/CMakeLists.txt#L298-L300

We enable basic optimizations (-O1) to reduce the IR size and speed up runtime JIT. Empirically, the runtime JIT produces slightly better code when starting with IR that was optimized at -O1. Higher optimization levels tend to bloat the code.

This evaluation was done with LLVM 3.3. We're now on LLVM 5, and this evaluation may no longer hold. Revisit whether this makes sense or whether another option (O2, Os, O3) may make more sense.

Issue Links

• relates to: IMPALA-5081 Expose IR optimization level via query option