| Age | Commit message (Collapse) | Author |
|---|
|
|
|
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.
|
|
|
|
Small perf improvement in ipers.
radeon_drm_get_cs_handle is exactly what this commit tries to avoid
in every write_reloc.
|
|
isn't a problem yet, but have issues in hiz branch.
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
|
|
|
|
I have had a look at the libdrm sources and they just contain more or less
the same checking we do in macros, and begin_cs may realloc the CS buffer
if we overflow it, which never happens with r300g. So these are pretty
much useless.
There is a small but measurable performance increase by dropping the two
functions.
|
|
|
|
|
|
The idea is to build a hardware command buffer for every CSO and memcpy
the buffer to a command stream at bind time (or dirty-state-emission time,
to be precise).
|
|
The fewer macros, the better.
|
|
|
|
I'd like the CS macros to be as lightweight as possible for performance
reasons.
|
|
|
|
|
|
|
|
Whoo!
|
|
|
|
|
|
This creates a cleaner winsys and drop the simple screen stuff.
It makes r300g use pb_bufmgr structs.
It also tries to avoid overheads from mapping too often.
v5: clean warnings
v6: break out of cache check on first buffer - since most likely
the first busy one implies all after it are busy.
v7: cleanup a bit
v8-merged: drop cman for now to just get all the interface changes in first.
rework to changes that happened upstream
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
This reverts commit fff5be8e7b4557c221f2425dcafc2e7cbbba76ba.
Probably went too soon with this, dileX reported OA not working for him
it works here fine, but the optimisations I wanted aren't working properly
yet so I'll fix that now.
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
This creates a cleaner winsys and drop the simple screen stuff.
It makes r300g use pb_bufmgr structs and adds usage of the cached
bufmgr for vertex/index buffers.
It also avoids mapping too often.
I'm not 100% sure this is perfect but it won't find its own bugs.
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
|
|
|
|
|
|
There's like five good reasons for this, I swear.
|
|
|
|
|
|
So that debugging is no longer a full-spam-or-nothing approach, you are now
supposed to set the RADEON_DEBUG environment flag just like for classic Mesa.
The available debug flags are different, however. Just running an OpenGL
application with RADEON_DEBUG set to an arbitrary string will print out
helpful information.
Everything must be compiled with -DDEBUG for any of this to work
|
|
Most of it is no longer interesting.
|
|
This fixes some silent problems in current libdrm_radeon.
surface_copy still locks up hard.
|
|
Make things more consistent, prepare for more function hooks.
|
|
This fixes hardlocks with anything using elts.
|
|
*pulls paper bag down over head*
|
|
Guess it was a mistake in the first place. Oops.
|
|
Deck chairs on the Hindenburg. :3
|
|
Makes it a bit more manageable to read through the console logs.
|
|
Man, util/u_math just gets better by the day.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This might work for other people too.
|
|
This commit is only to protect against data loss, so please skip it when
bisecting. Thanks.
|
|
|
|
Deck chairs on the Titanic.
|
|
|
