[SERVER-74732] mcpu=native flag for arm Created: 09/Mar/23 Updated: 25/Mar/23 Resolved: 14/Mar/23 |
|
| Status: | Closed |
| Project: | Core Server |
| Component/s: | None |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Improvement | Priority: | Major - P3 |
| Reporter: | Alex Neben | Assignee: | Alex Neben |
| Resolution: | Done | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Attachments: |
|
||||||||
| Issue Links: |
|
||||||||
| Assigned Teams: |
Server Development Platform
|
||||||||
| Backwards Compatibility: | Fully Compatible | ||||||||
| Participants: | |||||||||
| Description |
|
During the talk on 3/9 James Greenhalgh recommend we compile with "-O3 -mcpu=native". We should see what kinds of improvements we can unlock by building with those flags. |
| Comments |
| Comment by Alex Neben [ 14/Mar/23 ] |
|
All graphs show no difference for each combination. So I am going to close this as no change. |
| Comment by Alex Neben [ 13/Mar/23 ] |
|
Tried -mcpu=native, -mcpu=native + O3, -mcpu=native + Ofast. Unclear if we can use Ofast because it does some math things. Just getting some data before figuring out if i need to talk to other teams about possibly enabling it for real. |
| Comment by Geert Bosch [ 10/Mar/23 ] |
|
-O3 enables more aggressive inlining and loop unrolling among other things, and tends to make the code larger, often resulting in overall slowdown when indiscriminately applied to a large project, while often being beneficial on small synthetic tests such as our tests based on the Google Benchmark framework. If anything, there might be specific objects/libraries that could benefit. |