Open
Bug 1985439
Opened 1 year ago
Updated 1 year ago
the WLLAMA benchamrk in about:inference spends 550ms (SimpleAlloc) to 1200ms (BTAlloc) on TC threads
Categories
(Core :: JavaScript Engine, task, P3)
Core
JavaScript Engine
Tracking
()
NEW
People
(Reporter: mayankleoboy1, Unassigned)
References
(Blocks 1 open bug)
Details
Go to about:inference
Select WLLAMA benchamrk
Click on run
SA
- Total: https://share.firefox.dev/41V4LRc
- 450ms block: https://share.firefox.dev/3JBtbJe
- 90ms block: https://share.firefox.dev/47gird3
BT
- Total: https://share.firefox.dev/41rjjrE
- 900ms block: https://share.firefox.dev/45TKBIs
- 300ms block: https://share.firefox.dev/4n1R2jx
Not the biggest of perf issues though.
|
||
Comment 1•1 year ago
|
||
We're mostly switching away from wllama to use llama.cpp (native), so I don't that is of the highest priority.
|
Reporter | |
Updated•1 year ago
|
Blocks: sm-regalloc
|
||
Updated•1 year ago
|
Severity: -- → N/A
Priority: -- → P3
You need to log in
before you can comment on or make changes to this bug.
Description
•