This changes allow us to support HOST_COHERENT, HOST_CACHED and
HOST_COHERENT + HOST_CACHED memory types for platforms that has
the PAT uAPI.
Be aware that Xe KMD will not be able to support cached only memory
types, anv_xe_physical_device_init_memory_types() will reflect that
but internal usage should not allocate
VK_MEMORY_PROPERTY_HOST_CACHED_BIT only memory, hence the assert
added.
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25462>
Insert a dummy blit prior to MI_ARB_CHECK, MI_SEMAPHORE_WAIT,
MI_FLUSH_DW submitted on the copy engine.
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26209>
This provides a basic implementation of VK_AMD_buffer_marker: we can
write the 32-bit markers from within a command buffer. Unfortunately,
our hardware has several limitations that make this difficult to
implement well:
1. We don't have insight into when specific stages finish (i.e.
all geometry shaders are done, but pixel rasterization may
still be occurring).
2. We cannot perform pipelined writes of 32-bit values to arbitrary
memory locations. PIPE_CONTROL::Write Immediate Value would be
the obvious way to implement this, but it only supports 64-bit
values, and the extension doesn't allow us to do that. We instead
use MI_STORE_DATA_IMM to write 32-bit values, but this requires
hard stalls.
Despite those limitations, the extension may still be useful for tools
to debug GPU hangs. We hope to offer another extension in the future
which offers similar functionality but is more efficient on our GPUs.
v2: Updated by Lionel Landwerlin to fix a number of flushing and
cache coherency issues with these writes.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14924>
Per Google Test FAQ recommendation, prefer consutrctors and destructors
unless there's a need to use SetUp/TearDown.
We will take advantage of this later to initialize an fs_builder.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26301>
A "dance" is required with this uAPI, first we need to ask KMD what is
the size of the giving query id, then memory needs to be allocated to
match that size and then query again with the memory address set and
at this time Xe KMD will copy the query data to memory.
This dance was being duplicated in xe_engine_get_info() and
anv_xe_physical_device_get_parameters() and the next patch will also
use it in Iris, so here adding it common/xe and re-using as much
as possible.
There is one more implementation of this function in intel/dev but
due to how libs are linked intel/dev can't depend on to intel/common.
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26325>
This heuristic doesn't show much difference when you have a beafy
processor but on lower end skus, it increase the number of buffers in
the execbuffer ioctl, adding significant overhead in i915.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 4cdd3178fb ("anv: Meet CCS alignment reqs with dedicated allocs")
Tested-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26335>
This solves a problem when you have a big memory chunk of which some
regions are bound to images. If the image is destroyed, currently the
aux-tt mapping stays and prevent any new image aux-tt mapping within
that region, until the memory is freed.
This maps & unmaps the aux-tt region at respectively bind & destroy
time, so that the memory chunks can be map through aux-tt.
If there is aliasing of memory to 2 different images, then the first
one "wins" the aux mapping and gets compression support. The second
one doesn´t.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: ee6e2bc4a3 ("anv: Place images into the aux-map when safe to do so")
Tested-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26335>
The BO address is not really a good criteria since we can bind an
image at an offset inside a BO.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: ee6e2bc4a3 ("anv: Place images into the aux-map when safe to do so")
Tested-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26335>
To implement this feature, we need to do CPU side tracking of all
L3/L2/L1 entries. This does add a little bit of CPU allocations, but
the advantage is that the traversal of the page table tree is faster.
No more need for the linear seach of find_buffer().
With this feature, we can have multiple VkImage bind to the same main
memory address, as long as they share exact same mapping parameters.
The AUX mapping will be removed when the last VkImage is destroyed.
As previously, if the L1 mapping entry parameters don't match, the
mapping fails. Anv handles this nicely by just disabling AUX on the
image.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Tested-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26335>
Once NIR code is lowered and a few optimization passes have run, there
might be flag register interactions between instructions quite far
away from one another.
In the following case :
f0 = and r0, r1
...
fs_interpolate r2, r3
...
if f0
...
endif
If we lower fs_inteporlate while using the f0 register, we completely
garble the value meant for the if block.
To fix this, emit the predication for fs_interpolate in brw_fs_nir.cpp
when doing the NIR translation to the backend IR. This will guarantee
that the flag register interactions are visible to the optimization
passes, avoiding the problem above.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 68027bd38e ("intel/fs: implement dynamic interpolation mode for dynamic persample shaders")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/9757
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26306>
`\$` is interpreted before being passed to `re.search()`, but luckily
for us the escape is also invalid and because of that, python 3.12+
warns us about it.
Use a raw string instead, so that the `\` is passed untouched to
`re.search()`.
Fixes: aa04b47c6e ("intel/perf: add support for GtSlice/GtSliceXDualsubsliceY variables")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26355>
Here renaming the PAT entries to a name that better express each
entry.
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25447>
Anybody will be tempted to factor out the if-else block code since it
looks like duplication but else block actually handles the ycbcr images
where the aspect masks are compatible but don't need to be the same.
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26294>
Set some state and implement dummy draws whenever viewport pointer
is being reprogrammed.
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25987>
Add a helper that only emits hw_state, this makes it easier to modify
dirty state and call helper to emit only wanted state.
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25987>
Those are not reused, so this will be the first and only allocation, so
no need to use the "realloc" variants.
For the fs_reg arrays, there's currently no particular reason to keep
them uninitialized, so zero-initialize them too -- not ideal but better
than random values.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26302>
This reworks the intel_compute_pixel_hash_table_nway() pixel pipe
hashing table computation helper to handle cases where some pixel
pipes have processing power different from the others, this is helpful
for Gfx12.7+ platforms where there are pixel pipes with 1 DSS as well
as pixel pipes with 2 DSSes, which currently can lead to a serious
performance bottleneck in the pixel pipes with lower processing power.
In order to avoid such a load imbalance the
intel_compute_pixel_hash_table_nway() function will now take two pixel
pipe bitsets instead of one: Pixel pipes enabled on both bitsets will
appear with twice the frequency on the table as pixel pipes which only
appear on one bitset. See the comments below for more details on the
algorithm used to construct a pixel hashing table with the desired
properties.
With this change rendering performance improves by about 25% on a
fused MTL platform -- The list of specific configs this is expected to
show an improvement on is not included here since the list is rather
long and some of the configs may still be embargoed or may never be
productized, but in order to find out whether your Gfx12.7+ device
could be affected by this you can check the output of the
intel_dev_info tool from the Mesa tree and see if there are multiple
"pixel pipe" entries with different DSS count. That isn't expected to
occur on any DG2 configuration, only on MTL+ platforms, so this change
should have no effect at all on DG2 (it's easy to convince oneself
that it won't since for DG2 mask1 should equal mask2 so mask2 will be
set to zero at the beginning of intel_compute_pixel_hash_table_nway()
and the new swzx[] permutation will be set to the identity).
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26266>
Sync xe_drm.h with commit 3b8183b7efad ("drm/xe/uapi: Be more specific
about the vm_bind prefetch region").
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26238>
Using the unsized data field is incorrect. The data is located behind
the entire drm_i915_query_perf_config structure.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: mesa-stable
Reviewed-by: Rohan Garg <rohan.garg@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26285>
DRM_XE_EXEC_QUEUE_SET_PROPERTY is the offset,
while DRM_IOCTL_XE_EXEC_QUEUE_SET_PROPERTY is the real number.
Signed-off-by: Shuicheng Lin <shuicheng.lin@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26253>
Regarding supporting these formats, the spec says:
"A sparse image created using VK_IMAGE_CREATE_SPARSE_RESIDENCY_BIT
supports all non-compressed color formats with power-of-two element
size that non-sparse usage supports. Additional formats may also be
supported and can be queried via
vkGetPhysicalDeviceSparseImageFormatProperties.
VK_IMAGE_TILING_LINEAR tiling is not supported."
Regarding the formats themselves, the spec says:
"VK_FORMAT_B8G8R8G8_422_UNORM specifies a four-component, 32-bit
format containing a pair of G components, an R component, and a B
component, collectively encoding a 2×1 rectangle of unsigned
normalized RGB texel data. One G value is present at each i
coordinate, with the B and R values shared across both G values and
thus recorded at half the horizontal resolution of the image. This
format has an 8-bit B component in byte 0, an 8-bit G component for
the even i coordinate in byte 1, an 8-bit R component in byte 2,
and an 8-bit G component for the odd i coordinate in byte 3. This
format only supports images with a width that is a multiple of two.
For the purposes of the constraints on copy extents, this format is
treated as a compressed format with a 2×1 compressed texel block."
Since these formats are to be considered compressed 2x1 blocks and we
don't necessarily have to support non-compressed formats that
non-sparse support, we can claim them as not supported with sparse.
In addition to all of that, if you look at isl_gfx125_filter_tiling()
you'll see that we don't even support Tile64 for these formats, so
sparse residency (i.e., non-opaque image binds) doesn't really make
sense for them yet.
The Vulkan spec defines 4 other YCBCR "2x1 compressed" formats like
the ones we have in this commit, but we don't support them even
without sparse, so there's no reason to check them here.
A recent change in VK-GL-CTS made tests that use these formats go from
unsupported to failures:
7ecc7716a983 ("Do not use and check for STORAGE image support, when
it is not used in the test")
This commit "fixes" the following VK-GL-CTS failures (by making them
return NotSupported):
dEQP-VK.sparse_resources.image_block_shapes.2d.b8g8r8g8_422_unorm.samples_1
dEQP-VK.sparse_resources.image_block_shapes.2d.g8b8g8r8_422_unorm.samples_1
dEQP-VK.sparse_resources.image_block_shapes.2d_array.b8g8r8g8_422_unorm.samples_1
dEQP-VK.sparse_resources.image_block_shapes.2d_array.g8b8g8r8_422_unorm.samples_1
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
On i915.ko we don't have the vm_bind ioctl, so sparse requires TR-TT.
Unfortunately, on gfx < 20 TR-TT is not compatible with non-render
queues, so we have to disable those when sparse is enabled. Notice
that although we don't have TR-TT for non-render queues on gfx >= 20,
vm_bind is the default, and it doesn't have this restriction.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
We need to wait for the batches to complete before we return the BOs
to the pool. We were previously doing this completely synchronously,
which made the code unnecessarily wait. Now we have a timeline syncobj
that signals completion of the previous BOs, so sometimes we check
where we are in the timeline and then return the BOs that we know are
unused.
This, in addition to the previous patch that made us wait for the
other syncobjs through the execbuf ioctl instead of through the CPU,
makes TR-TT batches at least an order of magnitude faster. Still, I
don't think we'll notice any changes in games's FPS as they don't bind
sparse resources that often.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
At this moment this patch won't buy us anything since we're already
being completely synchronous, but the next patch is going to change
this and so queue->sync will start making sense.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
For now it just wraps the bo and size, so there's really no value to
having it. In the next commit we'll add more elements to the struct.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
Pass them as part of the TR-TT batch. This is what a lot of the
previous commits were building up to.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
When sparse images are being used, applications normally use
non-opaque binds and leave opaque binds just for the miptail part.
Since miptails are always at the end of the array layers, processing
the opaque binds after processing the non-opaque binds increases the
chance that anv_sparse_submission_add() will join the miptail bind
operation with the last non-opaque opreration, especially if the user
is trying to bind the last few non-miptail levels and the miptail in
the same vkQueueBindSparse opration.
In the real world this case does happen, so we're able to save a bind
operation every once in a while in Steam games.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
Move waiting/signaling to the backends so we can fix each backend
separately.
As I write this patch the vm_bind backend is back to using synchronous
vm_binds so we can't pass syncobjs to the synchronous vm_bind ioctl
anymore. We'll need more discussions and possibly some rework before
we go back to asynchronous vm_binds. This commit should allow us to
fix the TR-TT backend in the next commit and leave vm_bind for later.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
Don't pass it as a parameter when it's also part of a struct. Have to
touch 9 files just for that...
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
If we're going to move syncobj waiting/signaling down to the backend
we're going to need a queue to signal as lost in case those operations
fail.
In some places of the stack we don't have a queue available, such as
when we're creating or destroying resources. For those, for vm_bind
cases we don't use the queue for anything so passing it as NULL is
fine. For TR-TT we are already using device->trtt.queue.
For TR-TT specifically this also means we're going to start using the
actual queues from the call stack instead of trtt->queue, but that
shouldn't make any difference since we only ever have one queue.
Still, this is more technigally correct.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
Our ultimate goal is to have the backend functions deal with the wait
and signal syncobjs instead of waiting for them on the CPU inside
anv_queue_submit_sparse_bind_locked(). For that, we'll need waits and
signals parameters to be passed all the way to the backend functions
that actually make the submission, and this is what this patch does,
through struct anv_sparse_submission.
This patch just deals with passing the parameters to the functions,
nothing is using the new variables yet. There should be no functional
changes here. The goal here is to make code review easier.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
Currently, a single vkQueueBindSparse() call may lead to multiple bind
calls in the backend (either a vm_bind ioctl or a command submission
that updates the TR-TT page tables). These operations can be quite
slow so it's better for us if we try to emit as few of them as
possible.
On top of that, this gives our "just extend the last operation's size
if possible" code a little more chance to act and save us real time.
Our ultimate goal here is to also pass submit->waits and
submit->signals to the backend so we can avoid doing CPU waits, so
having a single call to the backend helps simplify things a little
too, and we just created the structure to carry these extra pointers
forward.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
Having it helped us printing the resource offset, which made debugging
some situations easier. The problem is that we want to rework the code
a little bit and we won't have a 'sparse' struct anymore to pass
around. Since it's all debug code drop it for now so it doesn't get in
the way of the rework. If we need it later we can find a way to add it
back, or we find another way to print the value.
Drive-by drop the DEBUG_SPARSE check that's already in the caller.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
Same as the L1 case, but this one deals with 64bit entry addresses and
pte addresses.
Consecutive L3/L2 writes are much rarer than L1 writes since they
require some pretty big buffers, but we can still those cases in the
wild. I just don't think any change will be noticeable though.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
If the addresses are sequential, we can emit only a single
MI_STORE_DATA_IMM instruction. This is a very common case, it should
save us some space: 4 bytes per extra_write.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25512>
