LRZ fast-clear works on all gens, however blob disables it on
gen1 and gen2. We also elect to disable fast-clear on these gens
because for close to none gains it adds complexity and seem to work
a bit differently from gen3+. Which creates at least one edge case:
if first draw which uses LRZ fast-clear doesn't lock LRZ direction
the fast-clear value is undefined.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6829
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17599>
On-GPU LRZ direction tracking allows LRZ to support secondary cmdbufs,
reusing LRZ between renderpasses, and in future to support LRZ when
VK_KHR_dynamic_rendering is used.
With on-gpu tracking we have to be careful keeping LRZ state in sync
with underlying depth image, which means we should invalidate LRZ
when underlying image is changed or the view of image is different
from previous renderpass.
All of this resulted in LRZ logic being thinly spread through the code,
making it hard to understand. So most of it was moved to tu_lrz.c.
For more details on past and new LRZ features see comment at the
top of tu_lrz.c.
Note about blob:
- Blob is much more happy to do LRZ_FLUSH, it flushes at the start
of the renderpass, after binning, and at the end of the renderpass.
- Blob seem not to care about changes in depth image done via
vkCmdCopyImage.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6347
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16251>
- gen4 - has dp4acc and dp2acc, dp4acc is used to implement
4x8 dot product.
- gen3 - has dp2acc, in OpenCL blob uses dp2acc for dot product
on both get3 and gen4.
- gen2 - unknown, lower everything.
- gen1 - no dp2acc, lower everything. OpenCL blob doesn't advertise
cl_qcom_dot_product8 but still generates code for it.
The assembly is more verbose and uses yet to be documented
mad32.u16 instruction.
Passes:
dEQP-VK.spirv_assembly.instruction.compute.opsdotkhr.*
dEQP-VK.spirv_assembly.instruction.compute.opudotkhr.*
dEQP-VK.spirv_assembly.instruction.compute.opsudotkhr.*
dEQP-VK.spirv_assembly.instruction.compute.opsdotaccsatkhr.*
dEQP-VK.spirv_assembly.instruction.compute.opudotaccsatkhr.*
dEQP-VK.spirv_assembly.instruction.compute.opsudotaccsatkhr.*
Only packed 4x8 unsigned and mixed versions are accelerated.
However in theory we should be able to do better for signed version
than current NIR lowering.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13986>
Inferring from blob's cmdstream the size of shader instruction
cache for:
- a630 is 64
- a650 is 128
- a660 is 128
On a650 and a660 gpu could hang if we exceed the limit. Though
it is not reproducible with computerator or a single amber
test. Also while blob limits the size to 128 - Turnip still
hangs with it but does not hang with the limit of 127.
On a630 there seem to be no hang when limit is exceeded.
Fixes the hang of compute shader in Alien Isolation on a650/a660.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14044>
On some GPUs when:
- depth bounds test is enabled
- depth test is disabled
- depth attachment uses UBWC in sysmem mode
GPU hangs. As a workaround we should enable z test. That's what blob
is doing for a630. And since we enable z test we should make it always pass.
Blob doesn't emit this workaround on a650 and a660. Untested on a640.
Fixes:
dEQP-VK.pipeline.extended_dynamic_state.two_draws_static.depth_bounds_test_disable
dEQP-VK.pipeline.extended_dynamic_state.two_draws_dynamic.depth_bounds_test_disable
dEQP-VK.dynamic_state.ds_state.depth_bounds_1
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12407>
Move away from using gpu_id as the primary means to identify which
adreno we are running on, as future GPUs (starting with 7c3) stop
providing a gpu_id as a new naming scheme is introduced.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12159>
We only need it in a single .c file, so we can make the device table
static. Also rename the struct for device table entries, as I want
to re-use the name 'fd_dev_id'
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12159>
Newer a6xx devices drop this packet from the sqe firmware, and use
direct (pkt4) register writes instead for the few cases that previously
used CP_REG_WRITE.
The turnip part was adapted from Jonathans patch on !10892
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11790>
A step towards getting rid of checks for gpu_id sprinkled around.
Checking major generation is ok, but checking for == or >= a specific
gpu_id is going to start getting messy as we add more a6xx.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11790>
This way we can make the tables const. At the same time, for a6xx, this
introduces a "sub-generation template" to reduce the copy/paste for
parameters which are keyed to the sub-generation. It also explicitly
lists every supported GPU, to get rid of duplicate lists of supported
gpus between the device-info and drivers.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11790>
Replace it with a calculation which works for all current GPUs.
Duplicated the calculation in both drivers because freedreno_dev_info isn't
meant for derived parameters (and drivers might want to just calculate on
the fly instead).
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11790>
We have to allocate backing storage big enough to hold all the private
memory for all threads that can possibly be in flight, which means that
we have to start filling in some more model-specific information as the
sizes will be different for models with different core counts/ALU
counts.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7386>