Closed
Bug 838886
Opened 12 years ago
Closed 2 years ago
Limit gradient patterns cache memory usage
Categories
(Core :: Graphics, defect)
Tracking
()
RESOLVED
INVALID
People
(Reporter: avih, Unassigned)
References
Details
(Whiteboard: [MemShrink:P2])
Gradient pattern cache items could use a lot of memory per item (see bug 837542 for several MB of cached items), especially when using D2D. While less cache misses due to better cache keys would improve the overall cache memory usage (bug 838758), we might still want to impose some limit on this cache, as it can quickly grow quite large.
Reporter | ||
Comment 1•12 years ago
|
||
Correction: see bug 837542 for several hundreds MB of cached items.
Updated•12 years ago
|
Whiteboard: [MemShrink]
Comment 2•12 years ago
|
||
So I think using immense amounts of memory is usually due to drivers doing odd things and using much more memory than expected. In general a reasonable limit here is probably something like 1000 patterns which should amount to 16MB.
![]() |
||
Comment 3•12 years ago
|
||
Removed the MemShrink tag because we have a tag on 837542.
Whiteboard: [MemShrink]
Comment 4•12 years ago
|
||
(In reply to Nicholas Nethercote [:njn] from comment #3)
> Removed the MemShrink tag because we have a tag on 837542.
May be the MemShrink tag should be inserted again here, because 837542 is replaced by this bug.
![]() |
||
Updated•12 years ago
|
Whiteboard: [MemShrink]
![]() |
||
Updated•12 years ago
|
Whiteboard: [MemShrink] → [MemShrink:P2]
Updated•3 years ago
|
Severity: normal → S3
Comment 5•2 years ago
|
||
So, there's this bug and also bug 848742 (and maybe a few more pertaining to gradient memory usage). Is it still a concern? My understanding is that WebRender has cached render tasks for gradients, so I didn't want to close all of these without asking first.
Thanks!
Flags: needinfo?(nical.bugzilla)
Comment 6•2 years ago
|
||
Things are different enough that we should close this now and open new bugs if need be.
Status: NEW → RESOLVED
Closed: 2 years ago
Flags: needinfo?(nical.bugzilla)
Resolution: --- → INVALID
You need to log in
before you can comment on or make changes to this bug.
Description
•