From: Alex Deucher alexander.deuc...@amd.com
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
lib/Target/R600/Processors.td |3 +++
1 files changed, 3 insertions(+), 0 deletions(-)
diff --git a/lib/Target/R600/Processors.td b/lib/Target/R600/Processors.td
index 81f407e..a0735d4
From: Alex Deucher alexander.deuc...@amd.com
Set env var RADEON_VA=0 to disable VM on Cayman/Trinity.
Useful for debugging.
Note: this is a candidate for the 9.1 branch.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/winsys/radeon/drm/radeon_drm_winsys.c |2 ++
1
From: Alex Deucher alexander.deuc...@amd.com
Note: this is a candidate for the 9.1 branch
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/radeonsi/radeonsi_pipe.c |1 +
src/gallium/drivers/radeonsi/si_state.c |3 +++
From: Alex Deucher alexander.deuc...@amd.com
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/radeonsi/radeonsi_pipe.c |1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/radeonsi_pipe.c
From: Alex Deucher alexander.deuc...@amd.com
Note: this is a candidate for the 9.1 branch
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
include/pci_ids/radeonsi_pci_ids.h |7 +++
1 files changed, 7 insertions(+), 0 deletions(-)
diff --git
From: Alex Deucher alexander.deuc...@amd.com
It shouldn't be needed since the FLUSH_AND_INV_EVENT has already
made sure the destination caches are flushed. Additionally,
we didn't previously emit the surface_sync until this commit:
From: Alex Deucher alexander.deuc...@amd.com
Lighter weight then using streamout. Only evergreen
and newer asics support embedded data as src with
CP DMA.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/evergreen_hw_context.c | 66 +++
From: Alex Deucher alexander.deuc...@amd.com
There are too many cases were we end up with lockups.
Once we sort out the remaining issues on master, they
can be backported and hyperz can be re-enabled on 9.1
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
From: Alex Deucher alexander.deuc...@amd.com
The hw can only access compressed textures as tiled not
linear so we need to do format tricks to handle them
properly. The blitter code already handles this so
just fallback to the blitter for compressed textures.
Fixes:
From: Alex Deucher alexander.deuc...@amd.com
Needs to be set just like other blocks.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/evergreen_state.c | 11 +--
1 files changed, 9 insertions(+), 2 deletions(-)
diff --git
From: Alex Deucher alexander.deuc...@amd.com
The DMA block seems to have alignment issues with large
block sizes. Use the blitter for these surfaces.
v2: cayman/TN only
Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=60802
Note: this is a candidate for the 9.1 branch.
Signed-off-by: Alex
From: Alex Deucher alexander.deuc...@amd.com
Needs to be set for depth, stencil, and fmask just
like other blocks.
v2: drop additional cayman bits for now
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/evergreen_state.c |8 ++--
1 files changed, 6
From: Alex Deucher alexander.deuc...@amd.com
Write to a CB register between draws.
May fix:
https://bugs.freedesktop.org/show_bug.cgi?id=50655
https://bugs.freedesktop.org/show_bug.cgi?id=47116
v2: flush along with workaround.
Note: this is a candidate for the 9.1 branch.
Signed-off-by: Alex
From: Alex Deucher alexander.deuc...@amd.com
This patch set cleans up the flushing on r6xx in what seems to be
a logical manner. The last patch enables CP DMA on r6xx. No piglit
regressions on RS780 which I was testing on.
Alex Deucher (4):
r600g: add missing emit_flush for
From: Alex Deucher alexander.deuc...@amd.com
We set the cp_coher_cntl bits but never emit them.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/r600_hw_context.c |1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git
From: Alex Deucher alexander.deuc...@amd.com
I don't see why we shouldn't be setting these bits on 6xx/7xx
as well. They shouldn't hurt anything and we may be missing
synchronizations with certain blocks by not setting them.
The ddx already sets cp_coher_cntl in a similar manner.
v2: adjust
From: Alex Deucher alexander.deuc...@amd.com
PS_PARTIAL flushes seems to be required in certain
cases to prevent hangs, especially on r6xx.
Note: this is a candidate for the 9.1 branch.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/r600.h|1
From: Alex Deucher alexander.deuc...@amd.com
Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=50655
https://bugs.freedesktop.org/show_bug.cgi?id=47116
v2: flush along with workaround.
v3: just need a flush
v4: try WAIT_UNTIL
v5: switch to PS partial flush
v6: rework patch
Note: this is a
From: Alex Deucher alexander.deuc...@amd.com
May fix:
https://bugs.freedesktop.org/show_bug.cgi?id=58042
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/r600_hw_context.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git
From: Alex Deucher alexander.deuc...@amd.com
Streamout buffers need to be synchronized on r6xx as
well.
v2: Add DEST flush as well.
v3: drop DEST flush
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/r600_hw_context.c |2 +-
1 files changed, 1
From: Alex Deucher alexander.deuc...@amd.com
We set the cp_coher_cntl bits but never emit them.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/r600_hw_context.c |1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git
From: Alex Deucher alexander.deuc...@amd.com
With the previous flushing changes this seems to work
reliably now.
v2: add R600_CONTEXT_FLUSH_AND_INV
v3: just enable CP DMA
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/r600_blit.c |3 +--
From: Alex Deucher alexander.deuc...@amd.com
Can cause lockups in certain cases when
zfunc/zenable/zwrite change without a flush
in between.
Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=60969
This is a candidate for the 9.1 branch.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
From: Alex Deucher alexander.deuc...@amd.com
It shouldn't be needed and older kernels don't support
it.
v2: Replace with PS partial flush as before.
Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=59945
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
From: Alex Deucher alexander.deuc...@amd.com
It shouldn't be needed and older kernels don't support
it.
Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=59945
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/r600_hw_context.c |6 --
1 files changed,
From: Alex Deucher alexander.deuc...@amd.com
Need to add the virtual address.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/r600.h|4 ++--
src/gallium/drivers/r600/r600_hw_context.c | 11 +++
2 files changed, 9 insertions(+), 6
From: Alex Deucher alexander.deuc...@amd.com
Use per asic golden values.
Programming this register doesn't seem to be strictly
necessary on SI, but programming it wrong leads to
rendering issues or reduced performance so just
go ahead and program the golden values explicitly
to avoid any
From: Alex Deucher alexander.deuc...@amd.com
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/radeonsi/si_state.c |3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_state.c
From: Alex Deucher alexander.deuc...@amd.com
Rather than disabling the depth buffer.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/radeonsi/si_state.c |9 +++--
1 files changed, 3 insertions(+), 6 deletions(-)
diff --git
From: Alex Deucher alexander.deuc...@amd.com
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/radeonsi/si_state.c |1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_state.c
From: Alex Deucher alexander.deuc...@amd.com
Clean up a few magic numbers and rework the code a bit.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
Reviewed-by: Christian König christian.koe...@amd.com
Reviewed-by: Michel Dänzer michel.daen...@amd.com
---
From: Alex Deucher alexander.deuc...@amd.com
Rather than disabling the depth buffer.
v2: use INVALID hw format rather than ~0U
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/radeonsi/si_state.c | 13 +
1 files changed, 5 insertions(+), 8
From: Alex Deucher alexander.deuc...@amd.com
v2: use INVALID hw format rather than ~0U
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/radeonsi/si_state.c |5 +++--
1 files changed, 3 insertions(+), 2 deletions(-)
diff --git
From: Alex Deucher alexander.deuc...@amd.com
Clean up a few magic numbers and rework the code a bit.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
Reviewed-by: Christian König christian.koe...@amd.com
Reviewed-by: Michel Dänzer michel.daen...@amd.com
---
From: Alex Deucher alexander.deuc...@amd.com
setting the DB format to Z_INVALID disables the depth buffer.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/radeonsi/si_state.c | 10 +++---
1 files changed, 3 insertions(+), 7 deletions(-)
diff --git
From: Alex Deucher alexander.deuc...@amd.com
setting the CB format to Z_INVALID disables the color buffer.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/radeonsi/si_state.c |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git
From: Alex Deucher alexander.deuc...@amd.com
Clean up a few magic numbers and rework the code a bit.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/radeonsi/si_state.c | 22 ++
src/gallium/drivers/radeonsi/sid.h |2 ++
2 files
From: Alex Deucher alexander.deuc...@amd.com
This patch set cleans up and unifies the common state and CS init
set up between evergreen and cayman and gfx and compute. No piglit
regressions on the gfx side here. This also allows compute to run
on cayman without causing a CS error in the kernel
From: Alex Deucher alexander.deuc...@amd.com
It's required. The CP uses this to properly allocate new
contexts. Also do a CS partial flush since we are updating
CONFIG regs which are single state.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
From: Alex Deucher alexander.deuc...@amd.com
Move gfx specific bits out as the code is shared with
compute.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/evergreen_state.c | 73 +---
1 files changed, 33 insertions(+), 40
From: Alex Deucher alexander.deuc...@amd.com
we were previously only setting 8 of them.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/evergreen_state.c | 21 +++--
1 files changed, 19 insertions(+), 2 deletions(-)
diff --git
From: Alex Deucher alexander.deuc...@amd.com
These are common to both evergreen and cayman, but were
not emitted on cayman.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/evergreen_state.c | 54
1 files changed, 54
From: Alex Deucher alexander.deuc...@amd.com
And use it for compute. This should improve compute support
on cayman.
Signed-off-by: Alex Deucher alexander.deuc...@amd.com
---
src/gallium/drivers/r600/evergreen_compute.c |8 -
src/gallium/drivers/r600/evergreen_state.c | 39
43 matches
Mail list logo