| 78222e63630280b96488de0d51d1b2578e26f814 |
|
30-Jan-2013 |
Jerome Glisse <jglisse@redhat.com> |
r600g: add cs memory usage accounting and limit it v3 (backport for mesa 9.0)
We are now seing cs that can go over the vram+gtt size to avoid
failing flush early cs that goes over 70% (gtt+vram) usage. 70%
is use to allow some fragmentation.
The idea is to compute a gross estimate of memory requirement of
each draw call. After each draw call, memory will be precisely
accounted. So the uncertainty is only on the current draw call.
In practice this gave very good estimate (+/- 10% of the target
memory limit).
v2: Remove left over from testing version, remove useless NULL
checking. Improve commit message.
v3: Add comment to code on memory accounting precision
This version is a backport for mesa 9.0
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Reviewed-by: Marek Olšák <maraeo@gmail.com>
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| f6a66a33f7ba52e36242db9346573bf477efa04e |
|
19-Sep-2012 |
Vadim Girlin <vadimgirlin@gmail.com> |
winsys/radeon: fix relocs caching
Don't cache pointers to elements of reallocatable array.
In some circumstances it caused false cache hits resulting in incorrect
command stream and gpu lockup.
Note: This is a candidate for the stable branches.
Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>
Reviewed-by: Marek Olšák <maraeo@gmail.com>
(cherry picked from commit 9aa8bac98b823e8783bc3a06a6e5b23fbf8d87fb)
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| a7f4d3b740d4c85b0dc2b006c30c2bc4a3ed8597 |
|
04-Aug-2012 |
Marek Olšák <maraeo@gmail.com> |
winsys/radeon: print error if CS is overflowed
and don't submit the CS to the kernel.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 92af184690995d3b16731518f7becfaac3538edb |
|
21-May-2012 |
Christian König <deathsimple@vodafone.de> |
winsys/radeon: enable IB submission to compute rings v2
This allows to submit things to the compute only
rings on cayman+
v2: rebased on current master and actually make use
of the new flag in evergreen_compute.c
Signed-off-by: Christian König <deathsimple@vodafone.de>
Reviewed-by: Marek Olšák <maraeo@gmail.com>
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 6bb0151f1fd87b4e15f177c7122fc28fea29497e |
|
22-May-2012 |
Maarten Lankhorst <m.b.lankhorst@gmail.com> |
winsys/radeon: Remove unnecessary pipe_thread_destroy in radeon_drm_cs_destroy
Fixes crash bug introduced with 210ddf0819b5 fd.o #49198
pthread_detach after a pthread_join is unneeded.
Signed-off-by: Maarten Lankhorst <m.b.lankhorst@gmail.com>
Signed-off-by: Marek Olšák <maraeo@gmail.com>
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 29e55bc5f1b6d7375b6a86e24ca4ae58e399011e |
|
26-Feb-2012 |
Marek Olšák <maraeo@gmail.com> |
winsys/radeon: add usage parameter to cs_is_buffer_referenced
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| bb1f0cf3508630a9a93512c79badf8c493c46743 |
|
02-Dec-2011 |
Jerome Glisse <jglisse@redhat.com> |
r600g: add support for virtual address space on cayman v11
Virtual address space put the userspace in charge of their GPU
address space. It's up to userspace to bind bo into the virtual
address space. Command stream can them be executed using the
IB_VM chunck.
This patch add support for this configuration. It doesn't remove
the 64K ib size limit thought this limit can be extanded up to
1M for IB_VM chunk.
v2: fix rendering
v3: fix rendering when using index buffer
v4: make vm conditional on kernel support add basic va management
v5: catch the case when we already have va for a bo
v6: agd5f: update on top of ioctl changes
v7: agd5f: further ioctl updates
v8: indentation cleanup + fix non cayman
v9: rebase against lastest mesa + improvement from Marek & Michel
v10: fix cut/paste bug
v11: don't rely on updated radeon_drm.h
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 93f4e3cb6c1ca303ee1f5c2a2491a8eff33f2633 |
|
24-Dec-2011 |
Marek Olšák <maraeo@gmail.com> |
winsys/radeon: move managing GEM domains back to drivers
This partially reverts commit 363ff844753c46ac9c13866627e096b091ea81f8.
It caused severe performance drops in Nexuiz. Reported by Phoronix.
Tested by me on r300g and by IRC people on r600g.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 41b75dc92c921c078e64962f47b3d51b1e4f7b5a |
|
25-Oct-2011 |
Marek Olšák <maraeo@gmail.com> |
winsys/radeon: add flush option not to rewrite tiling flags in registers
Not used yet.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 363ff844753c46ac9c13866627e096b091ea81f8 |
|
27-Sep-2011 |
Marek Olšák <maraeo@gmail.com> |
winsys/radeon: move GEM domains out of the drivers into winsys
The drivers don't need to care about the domains. All they need to set
are the bind and usage flags. This simplifies the winsys too.
This also fixes on r600g:
- fbo-depth-GL_DEPTH_COMPONENT32F-copypixels
- fbo-depth-GL_DEPTH_COMPONENT16-copypixels
- fbo-depth-GL_DEPTH_COMPONENT24-copypixels
- fbo-depth-GL_DEPTH_COMPONENT32-copypixels
- fbo-depth-GL_DEPTH24_STENCIL8-copypixels
I can't explain it.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| a3cd2c6c9b3724dbc3aa565dab98968c46bde963 |
|
27-Sep-2011 |
Marek Olšák <maraeo@gmail.com> |
winsys/radeon: remove redundant member radeon_bo::size
It's part of pb_buffer already.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 798ebc91ebe383a702faf597ffbf2033dc467f50 |
|
27-Sep-2011 |
Marek Olšák <maraeo@gmail.com> |
winsys/radeon: simplify updating GEM domains for relocations
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 210ddf0819b5acf87a614214b6d4b02193aafa4a |
|
27-Aug-2011 |
Maarten Lankhorst <m.b.lankhorst@gmail.com> |
winsys/radeon: Create async thread only once
I noticed that a thread was created for every time async flush was called, so I moved it and used some semaphores to synch.
Signed-off-by: Maarten Lankhorst <m.b.lankhorst@gmail.com>
Reviewed-by: Marek Olšák <maraeo@gmail.com>
Signed-off-by: Christian König <deathsimple@vodafone.de>
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 296b8990956fcbd7ce47902d7c108a5973db9397 |
|
07-Aug-2011 |
Marek Olšák <maraeo@gmail.com> |
winsys/radeon: remove broken bo-is-busy-for-write guessing
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| e2e1dc9e66ff348caa97b7b35e558d75c6cc6899 |
|
04-Aug-2011 |
Marek Olšák <maraeo@gmail.com> |
r600g: set the flush callback in radeon_winsys
I have also renamed the winsys function.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 03b25ad8ffd72f1f88b6c80a2ebfe3cf6e8a6390 |
|
04-Aug-2011 |
Marek Olšák <maraeo@gmail.com> |
winsys/radeon: consolidate the add_reloc function
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| e6fb62594fca714883af9bba9795be8838c16900 |
|
04-Aug-2011 |
Marek Olšák <maraeo@gmail.com> |
r600g: emit CS using radeon_winsys
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 638d75185e66727faaba5dc2df1b6e14c7c0c075 |
|
03-Aug-2011 |
Marek Olšák <maraeo@gmail.com> |
r600g: let radeon_winsys maintain the list of relocations
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 5b005ecc2b624a0ffb577ab760abacf069694f8d |
|
03-Aug-2011 |
Marek Olšák <maraeo@gmail.com> |
winsys/radeon: do the CS cleanup in the CS ioctl thread
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 64ab39b035f755510a644643b96451431bbe5f27 |
|
03-Aug-2011 |
Marek Olšák <maraeo@gmail.com> |
winsys/radeon: fix space checking
We should remove the relocations which caused a validation failure
from the list, so that the kernel receives only the validated ones.
NOTE: This is a candidate for the 7.11 branch.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 28a336dc38c478b809544e7404c4d1fddd873333 |
|
22-Jul-2011 |
Marek Olšák <maraeo@gmail.com> |
winsys/radeon: simplify how value queries work
This drops the get_value query and adds a function query_info, which returns
all the values in one nice structure.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| d35aeff4bb0b03450b2c3c08bd7f84db5bf43283 |
|
18-Apr-2011 |
Marek Olšák <maraeo@gmail.com> |
r300g/winsys: rename r300->radeon and do a little cleanup
Renaming a few files, types, and functions.
Also make the winsys independent of r300g.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| aac4bae98651c928ce298aecf158b0aa1a7fabce |
|
08-Apr-2011 |
Marek Olšák <maraeo@gmail.com> |
r300g: fix crash when adding a new relocation
No idea why it's so hard to reproduce.
Broken with: c35572352e3e92683988ee8d151b47f4190d62f9
Thanks to Toni Spets for assistance.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 9d5de0fcb6ced7a4da85a09ad25dcbc2b21bfdf9 |
|
07-Apr-2011 |
Marek Olšák <maraeo@gmail.com> |
r300g: add 'radeon: ' prefix to error messages in winsys
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 6caac3ecb8bc32d92c35fdb1f0a67541ffa8af29 |
|
03-Apr-2011 |
Marek Olšák <maraeo@gmail.com> |
r300g: do not wait for a busy BO if neither GPU nor CPU is changing it
Improves frame rate in apps with at least one user vertex buffer and
a hw index buffer.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| c35572352e3e92683988ee8d151b47f4190d62f9 |
|
03-Apr-2011 |
Marek Olšák <maraeo@gmail.com> |
r300g: remove the redundant reference counter in radeon_bo
We already have pb_buffer::reference::count.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 5650a719f0c69c00954e47bd7a7b3e9433cb551d |
|
08-Mar-2011 |
Marek Olšák <maraeo@gmail.com> |
r300g: decide whether a flush should be asynchronous when calling it
Thread offloading is not sometimes desirable, e.g. when mapping a buffer.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| f6dbcb92bf4a3597c7b3da890ffafc84951f25df |
|
02-Mar-2011 |
Marek Olšák <maraeo@gmail.com> |
r300g: do not use ioctl thread offloading on single-core machines
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| e9e5380f22c230ea1276208baa40c3fe6482b296 |
|
19-Feb-2011 |
Marek Olšák <maraeo@gmail.com> |
r300g: fix invalid dereference in winsys
radeon_bo_unref may destroy the buffer, so call it after p_atomic_dec, not before.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| fa3f1348e49feeac511dbe5b22bbddc47f56ba81 |
|
16-Feb-2011 |
Marek Olšák <maraeo@gmail.com> |
r300g: fix a race between CS and SET_TILING ioctls
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| b9e2cde6006b557a3a23a82384899f4d5a5ac7b8 |
|
15-Feb-2011 |
Marek Olšák <maraeo@gmail.com> |
r300g: offload the CS ioctl to another thread
This is a multi-threading optimization which hides the kernel overhead
behind a thread. It improves performance in CPU-limited apps by 2-15%.
Of course you must have at least 2 cores for it to make any difference.
It can be disabled with:
export RADEON_THREAD=0
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| df54b53b7d12a3bca5867b6649cb308feb36f0da |
|
11-Feb-2011 |
Marek Olšák <maraeo@gmail.com> |
r300g: improve function radeon_bo_is_referenced_by_cs
This should prevent calling into radeon_get_reloc when there's
only one context.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 6ccab620a0e7364ab6c0d902b3ddf58ee988f7fa |
|
23-Dec-2010 |
Marek Olšák <maraeo@gmail.com> |
r300g: import the last bits of libdrm and cleanup the whole thing
Based on Dave's branch.
The majority of this commit is a cleanup, mainly renaming things.
There wasn't much code to import, just ioctl calls.
Also done:
- implemented unsynchronized bo_map (important optimization!)
- radeon_bo_is_referenced_by_cs is no longer a refcount hack
- dropped the libdrm_radeon dependency
I'm surprised that this has resulted in less code in the end.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|
| 1f0348c4a2ba4f8ca812a2daccf3b01c3f984ef1 |
|
04-Dec-2010 |
Marek Olšák <maraeo@gmail.com> |
r300g: rework command submission and resource space checking
The motivation behind this rework is to get some speed by reducing
CPU overhead. The performance increase depends on many factors,
but it's measurable (I think it's about 10% increase in Torcs).
This commit replaces libdrm's radeon_cs_gem with our own implemention.
It's optimized specifically for r300g, but r600g could use it as well.
Reloc writes and space checking are faster and simpler than their
counterparts in libdrm (the time complexity of all the functions
is O(1) in nearly all scenarios, thanks to hashing).
(libdrm's radeon_bo_gem is still being used in the driver.)
It works like this:
cs_add_reloc(cs, buf, read_domain, write_domain) adds a new relocation and
also adds the size of 'buf' to the used_gart and used_vram winsys variables
based on the domains, which are simply or'd for the accounting purposes.
The adding is skipped if the reloc is already present in the list, but it
accounts any newly-referenced domains.
cs_validate is then called, which just checks:
used_vram/gart < vram/gart_size * 0.8
The 0.8 number allows for some memory fragmentation. If the validation
fails, the pipe driver flushes CS and tries do the validation again,
i.e. it validates only that one operation. If it fails again, it drops
the operation on the floor and prints some nasty message to stderr.
cs_write_reloc(cs, buf) just writes a reloc that has been added using
cs_add_reloc. The read_domain and write_domain parameters have been removed,
because we already specify them in cs_add_reloc.
The space checking has been tested by putting small values in vram/gart_size
variables.
/external/mesa3d/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
|