summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* isl: Kill off isl_format_layout::bsJason Ekstrand2016-07-137-22/+21
| | | | Reviewed-by: Chad Versace <[email protected]>
* isl: Take bpb rather than bs in tiling_get_infoJason Ekstrand2016-07-132-6/+6
| | | | Reviewed-by: Chad Versace <[email protected]>
* isl: Use bpb in a few places where it's more natural than bsJason Ekstrand2016-07-135-8/+8
| | | | Reviewed-by: Chad Versace <[email protected]>
* isl: Use bpb for determining YUV image paddingJason Ekstrand2016-07-131-1/+1
| | | | | | | When we initially dropped bpb in favor of bs, we accidentally didn't change this one line properly. This brings it back to what it should be. Reviewed-by: Chad Versace <[email protected]>
* isl: Bring back isl_format_layout::bpbJason Ekstrand2016-07-132-2/+4
| | | | | | | | A while ago we got rid of the bits-per-block because we thought we didn't need it. We're about to introduce some very useful 1 and 2-bit formats so we really should be able to handle them again. Reviewed-by: Chad Versace <[email protected]>
* isl: Change the physical size of a W-tile to 128x32Jason Ekstrand2016-07-134-19/+15
| | | | Reviewed-by: Chad Versace <[email protected]>
* isl: Rework the way we define tile sizes.Jason Ekstrand2016-07-132-81/+137
| | | | | | | | | | | This is based on a very long set of discussions between Chad and myself about how we should properly represent HiZ and CCS buffers. The end result of that discussion was that a tiling actually has two different sizes, a logical size in elements, and a physical size in bytes and rows. This commit reworks ISL's pitch and size calculations to work in terms of these two sizes. Reviewed-by: Chad Versace <[email protected]>
* isl: Rework the way we handle surface paddingJason Ekstrand2016-07-131-27/+25
| | | | Reviewed-by: Chad Versace <[email protected]>
* isl: Use ARRAY_PITCH_SPAN_FULL for depth/stencil surfaces on gen7Jason Ekstrand2016-07-131-1/+1
| | | | | | | We helpfully inserted a PRM quotation about how we need to use ARRAY_PITCH_SPAN_FULL and then set it to COMPACT. Oops... Reviewed-by: Chad Versace <[email protected]>
* isl: Stop multiplying height by block sizeJason Ekstrand2016-07-131-2/+2
| | | | | | | | The row pitch already specifies the size of a row of elements. Multiplying by the block height simply causes us to allocate as muc as 12 times more memory than needed for compressed textures. Reviewed-by: Chad Versace <[email protected]>
* isl: Get rid of tiling_get_extentJason Ekstrand2016-07-132-17/+0
| | | | | | It was unused Reviewed-by: Chad Versace <[email protected]>
* nir/spirv: Don't multiply the push constant block size by 4Jason Ekstrand2016-07-131-1/+1
| | | | | | | | | I have no idea why we were multiplying by 4 before. The offsets we get from SPIR-V are in bytes and so is nir->num_uniforms so there's no need to do any adjustment whatsoever. Signed-off-by: Jason Ekstrand <[email protected]> Cc: "12.0" <[email protected]>
* anv/pipeline: Assert that the number of uniforms from NIR fitsJason Ekstrand2016-07-131-0/+1
|
* radeonsi: report accurate SGPR and VGPR spillsMarek Olšák2016-07-132-5/+15
| | | | Reviewed-by: Nicolai Hähnle <[email protected]>
* radeonsi: add a workaround for a compute VGPR-usage LLVM bugMarek Olšák2016-07-131-0/+35
| | | | | | | v2: use abort(), describe which LLVM version is affected Cc: 12.0 <[email protected]> Reviewed-by: Nicolai Hähnle <[email protected]>
* radeonsi: use LLVMGetTypeKind to tell if an input is an array of descriptorsMarek Olšák2016-07-131-19/+11
| | | | | | just a cleanup Reviewed-by: Nicolai Hähnle <[email protected]>
* radeonsi: replace !tbaa with !invariant.loadMarek Olšák2016-07-131-12/+5
| | | | | | no change in generated code thanks to dereferenceable(n) Reviewed-by: Nicolai Hähnle <[email protected]>
* radeonsi: set dereferenceable attribute on descriptor arraysMarek Olšák2016-07-131-4/+11
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This allows moving the loads arbitrarily in the Sinking pass. 26002 shaders in 14643 tests Totals: SGPRS: 2080160 -> 2080160 (0.00 %) VGPRS: 798875 -> 797826 (-0.13 %) Spilled SGPRs: 108485 -> 79165 (-27.03 %) Spilled VGPRs: 327 -> 327 (0.00 %) Scratch VGPRs: 1656 -> 1652 (-0.24 %) dwords per thread Code Size: 36127192 -> 35559780 (-1.57 %) bytes LDS: 767 -> 767 (0.00 %) blocks Max Waves: 212464 -> 212672 (0.10 %) Wait states: 0 -> 0 (0.00 %) PERCENTAGES / App Shaders SGPRs VGPRs SpillSGPR SpillVGPR Scratch CodeSize MaxWaves Waits (unknown) 4 . . . . . . . . 0ad 6 . . . . . . . . alien_isolation 2938 . 0.04 % -8.53 % . . -0.71 % -0.06 % . anholt 10 . . . . . . . . batman_arkham_origins 589 . -0.58 % -79.54 % . . -6.72 % 0.57 % . bioshock-infinite 1769 . -0.65 % -89.32 % . . -4.73 % 0.48 % . borderlands2 3968 . -0.31 % -51.21 % . . -4.09 % 0.22 % . brutal-legend 338 . -0.03 % -2.95 % . . -0.06 % . . civilization_beyond.. 116 . . -14.17 % . . -0.88 % . . counter_strike_glob.. 1142 . . . . . . . . dirt-showdown 541 . -0.56 % -40.14 % . -3.45 % -1.82 % 0.35 % . dolphin 22 . . . . . 0.16 % . . dota2 1747 . . . . . 0.01 % . . europa_universalis_4 76 . -0.23 % -42.11 % . . -0.96 % . . f1-2015 774 . -0.09 % -28.89 % . . -2.60 % 0.09 % . furmark-0.7.0 4 . . . . . . . . gimark-0.7.0 10 . . . . . . . . glamor 16 . . . . . . . . humus-celshading 4 . . . . . . . . humus-domino 6 . . . . . . . . humus-dynamicbranching 24 . 0.71 % . . . 0.29 % -0.45 % . humus-hdr 10 . . . . . . . . humus-portals 2 . . . . . . . . humus-volumetricfog.. 6 . . . . . . . . left_4_dead_2 1762 . . . . . . . . metro_2033_redux 2670 . -0.10 % -7.15 % . . -0.03 % . . nexuiz 80 . . . . . . . . pixmark-julia-fp32 2 . . . . . . . . pixmark-julia-fp64 2 . . . . . . . . pixmark-piano-0.7.0 2 . . . . . . . . pixmark-volplosion-.. 2 . . . . . . . . plot3d-0.7.0 8 . . . . . . . . portal 474 . . . . . . . . sauerbraten 7 . . . . . . . . serious_sam_3_bfe 392 . . -13.20 % . . -1.81 % . . supertuxkart 4 . . . . . . . . talos_principle 324 . -0.21 % -18.39 % . . -2.73 % 0.14 % . team_fortress_2 808 . . . . . . . . tesseract 430 . 0.08 % -68.57 % . . -0.45 % . . tessmark-0.7.0 6 . . . . . . . . thea 172 . . . . . 0.03 % . . ue4_effects_cave 299 . -0.04 % -10.15 % . . -0.25 % 0.04 % . ue4_elemental 586 . -0.02 % -13.93 % . . -0.13 % 0.02 % . ue4_lightroom_inter.. 74 . -0.17 % -70.00 % . . -1.27 % . . ue4_realistic_rende.. 92 . . -32.58 % . . -0.35 % . . unigine_heaven 322 . 0.12 % -54.17 % . . -1.42 % -0.12 % . unigine_sanctuary 264 . . . . . . . . unigine_tropics 210 . . . . . . . . unigine_valley 278 . -0.15 % -40.74 % . . -2.00 % 0.09 % . unity 72 . . . . . 0.03 % . . warsow 176 . . . . . . . . warzone2100 4 . . . . . 0.13 % . . witcher2 1040 . -0.03 % -86.28 % . . -0.28 % 0.01 % . xcom_enemy_within 1236 . -0.24 % -63.54 % . . -0.93 % 0.18 % . yofrankie 82 . -0.61 % -100.00 % . . -0.83 % 0.41 % . ----------------------------------------------------------------------------------------------------------- Total 26002 . -0.13 % -27.03 % . -0.24 % -1.57 % 0.10 % . Reviewed-by: Nicolai Hähnle <[email protected]>
* gallivm: add helper lp_add_attr_dereferenceableMarek Olšák2016-07-132-0/+14
| | | | | | | | | Not sure if this is the right way to do it, but it seems to work. v2: make it a no-op on LLVM <= 3.5 Reviewed-by: Roland Scheidegger <[email protected]> Reviewed-by: Nicolai Hähnle <[email protected]>
* radeonsi: clean up shader value metadata codeMarek Olšák2016-07-131-15/+19
| | | | | | | No change in behavior. BTW, tbaa_md_kind == 1, which was the magic number in the code. Reviewed-by: Nicolai Hähnle <[email protected]>
* radeonsi: remove LLVMNoUnwindAttribute usesMarek Olšák2016-07-131-36/+31
| | | | | | always set by gallivm Reviewed-by: Nicolai Hähnle <[email protected]>
* radeonsi: fix a typo in SI_PARAM_LINEAR_* handlingMarek Olšák2016-07-131-1/+1
| | | | | | introduced in 476e9cee1d0cbe321c401277214e6c36ce5b18c9 Reviewed-by: Nicolai Hähnle <[email protected]>
* gallium/radeon: normalize the code styleMarek Olšák2016-07-132-338/+286
| | | | | | no change in behavior Reviewed-by: Nicolai Hähnle <[email protected]>
* radeonsi: just save buffer sizes instead of buffers while recording IBsMarek Olšák2016-07-135-10/+5
| | | | | | whole buffer objects are not needed Reviewed-by: Nicolai Hähnle <[email protected]>
* Add c99_alloca.h include to fix compilation on CygwinJon Turney2016-07-131-0/+1
| | | | | | | | Fix compilation on Cygwin, since 50b22354, by adding c99_alloca.h include, which should know how to portably make the alloc() prototype available. Signed-off-by: Jon Turney <[email protected]> Reviewed-by: Marek Olšák <[email protected]>
* i965/blorp: Cleanup leftovers from push constant disablingTopi Pohjolainen2016-07-132-65/+12
| | | | | | | | | Setup for pixel shader push constants is the same as for other stages. Note that on gen8+ the if-else branches were identical and the generation check for packet size redundant. Signed-off-by: Topi Pohjolainen <[email protected]> Reviewed-by: Kenneth Graunke <[email protected]>
* i965/blorp/gen7+: Bring back push constant setupTopi Pohjolainen2016-07-132-0/+52
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This is partial revert of commit cc2d0e64. It looks that even though blorp disables a stage the corresponding 3DSTATE_CONSTANT_XS packet is needed to be programmed. Hardware seems to try to fetch the constants even for disabled stages. Therefore care needs to be taken that the constant buffer is set up properly. Blorp will continue to trash it into non-existing such as before. It is possible that this could be omitted on SKL where the constant buffer is considered when the corresponding binding table settings are changed. Bspec: "The 3DSTATE_CONSTANT_* command is not committed to the shader unit until the corresponding (same shader) 3DSTATE_BINDING_TABLE_POINTER_* command is parsed." However, as CONSTANT_XS packet itself does not seem to stall on its own, it is safer to emit the packets for SKL also. Possible alternative to blorp trashing could have been to setup defaults in the beginning of each batch buffer. However, hardware doesn't seem to tolerate these packets being programmed multiple times per primitive. Bspec for IVB: "It is invalid to execute this command more than once between 3D_PRIMITIVE commands." Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=96878 Signed-off-by: Topi Pohjolainen <[email protected]> Reviewed-by: Kenneth Graunke <[email protected]>
* radeonsi: silence Coverity warningNicolai Hähnle2016-07-132-0/+4
| | | | | | | | Coverity's analysis is too weak to understand that r600_init_flushed_depth(_, _, NULL) only returns true when flushed_depth_texture was assigned a non-NULL value. Reviewed-by: Marek Olšák <[email protected]>
* i965/fs: do d2x lowering before simd splittingSamuel Iglesias Gonsálvez2016-07-131-5/+5
| | | | | | | So that we can have gen7 split large writes produced by this lowering pass. Signed-off-by: Samuel Iglesias Gonsálvez <[email protected]> Reviewed-by: Francisco Jerez <[email protected]>
* i965/fs: do pack lowering before simd splittingIago Toral Quiroga2016-07-131-5/+5
| | | | | | So that we can have gen7 split large writes produced by the pack lowering. Reviewed-by: Francisco Jerez <[email protected]>
* i965/fs: do not require force_writemask_all with exec_size 4Samuel Iglesias Gonsálvez2016-07-131-1/+1
| | | | | | | | | | | | So far we only used instructions with this size in situations where we did not operate per-channel and we wanted to ignore the execution mask, but gen7 fp64 will need to emit code with a width of 4 that needs normal execution masking. v2: - Modify the assert instead of deleting it (Curro) Reviewed-by: Francisco Jerez <[email protected]>
* i965/fs/gen7: split instructions that run into exec masking bugsIago Toral Quiroga2016-07-131-0/+29
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | In fp64 we can produce code like this: mov(16) vgrf2<2>:UD, vgrf3<2>:UD That our simd lowering pass would typically split in instructions with a width of 8, writing to two consecutive registers each. Unfortunately, gen7 hardware has a bug affecting execution masking and as a result, the second GRF register write won't work properly. Curro verified this: "The problem is that pre-Gen8 EUs are hardwired to use the QtrCtrl+1 (where QtrCtrl is the 8-bit quarter of the execution mask signals specified in the instruction control fields) for the second compressed half of any single-precision instruction (for double-precision instructions it's hardwired to use NibCtrl+1, at least on HSW), which means that the EU will apply the wrong execution controls for the second sequential GRF write if the number of channels per GRF is not exactly eight in single-precision mode (or four in double-float mode)." In practice, this means that we cannot write more than one consecutive GRF in a single instruction if the number of channels per GRF is not exactly eight in single-precision mode (or four in double-float mode). This patch makes our SIMD lowering pass split this kind of instructions so that the split versions only write to a single register. In the example above this means that we split the write in 4 instructions, each one writing 4 UD elements (width = 4) to a single register. v2 (Curro): - Make explicit that the thing about hardwiring NibCtrl+1 for the second compressed half is known to happen in Haswell and the issue with IVB might not be exactly the same. - Assign max_width instead of returning early so that we can handle multiple restrictions affecting to the same instruction. - Avoid division by 0 if the instruction does not write any registers. - Ignore instructions what have WE_all set. - Use the instruction execution type size instead of the dst type size. v3 (Curro): - Move the implementation down so it is not placed in the middle of another workaround. - Declare channels_per_grf as const. - Don't break the loop early if we find a BAD_FILE source. - Fix the number of channels that the hardware shifts for the second half of a compressed instruction to be 8 in single precision and 4 in double precision. Reviewed-by: Francisco Jerez <[email protected]>
* i965/fs: use the new helper function to create double immediatesIago Toral Quiroga2016-07-131-3/+4
| | | | Reviewed-by: Kenneth Graunke <[email protected]>
* i965/fs: add a helper function to create double immediatesIago Toral Quiroga2016-07-132-0/+40
| | | | | | | | | | | | | | | | | | | Gen7 hardware does not support double immediates so these need to be moved in 32-bit chunks to a regular vgrf instead. Instead of doing this every time we need to create a DF immediate, create a helper function that does the right thing depending on the hardware generation. v2: - Define setup_imm_df() as an independent function (Curro) - Create a specific builder to get rid of some instruction field assignments (Curro). v3: - Get devinfo from builder (Kenneth) Signed-off-by: Samuel Iglesias Gonsálvez <[email protected]> Reviewed-by: Kenneth Graunke <[email protected]>
* vc4: Validate QPU uniform pointer updates.Eric Anholt2016-07-121-0/+22
|
* vc4: Add support for NIR loops and break/continue.Eric Anholt2016-07-122-3/+79
|
* vc4: Add support for emitting NIR IF nodes.Eric Anholt2016-07-121-1/+91
|
* vc4: Add support for storing to NIR registers in a non-SSA fashion.Eric Anholt2016-07-122-85/+144
| | | | | | | Previously, there were occasionally NIR registers in our programs, but they were always actually used SSA-only. Now that we're trying to support control flow, we need to actually conditionally move to registers based on whether channels are active or not.
* vc4: Add a flag in the screen to track control flow support.Eric Anholt2016-07-123-1/+14
| | | | | For now it's still always false, but I need it in place for kernel backwards compat support as I extend the backend for control flow.
* vc4: Define a QIR branch instructionEric Anholt2016-07-124-9/+61
| | | | | | This uses the branch condition code in inst->cond to jump to either successor[0] (condition matches) or successor[0] (condition doesn't match).
* vc4: Add kernel support for branching in shader validation.Eric Anholt2016-07-123-17/+280
| | | | | | | | | | | | | | | | | | | | | We're already checking that branch instructions are within the contents of the shader and the proper PROG_END sequence is present. The other thing we need in the presence of branching is to verify that the shader doesn't overflow past the end of the uniforms stream. To do that, we require that at the start of any basic block reading uniforms have the following instructions: load_imm temp, <offset within uniform stream> add unif_addr, temp, unif The instructions are generated by userspace, and the kernel verifies that the load_imm is of the expected offset, and that the add adds it to a uniform. We track which uniform in the stream that is, and at draw call time fix up the uniform stream to have the address of the start of the shader's uniforms for that draw call. Signed-off-by: Eric Anholt <[email protected]>
* vc4: Add a bitmap of branch targets in kernel validation.Eric Anholt2016-07-123-2/+133
| | | | | | This isn't used yet, it's just a first step toward loop validation. During the main parsing of instructions, we need to know when we hit a new basic block so that we can reset validated state.
* vc4: Track the current instruction into the validation_state.Eric Anholt2016-07-121-24/+30
| | | | | This reduces how much we need to pass around as arguments, which was becoming more of a problem with looping validation.
* vc4: Add QPU support for generating BRANCH instructions.Eric Anholt2016-07-125-1/+85
|
* vc4: Print live variable start/ends during QIR dumping.Eric Anholt2016-07-121-0/+45
| | | | | This only happens when live variables are set up, which is not in the normal dump, but is set up when we've failed to register allocate.
* vc4: Implement live intervals using a CFG.Eric Anholt2016-07-126-39/+393
| | | | | Right now our CFG is always a trivial single basic block, but that will change when enable loops.
* vc4: Make vc4_qir_schedule handle each block in the program.Eric Anholt2016-07-121-14/+23
| | | | | | | | Basically we just treat each block independently. The only inter-block scheduling I can think of that would be be interesting would be to move texture result collection to after a short loop/if block that doesn't do texturing. However, the kernel disallows that as part of its security validation.
* vc4: Convert uniforms lowering to work with multiple blocks.Eric Anholt2016-07-121-29/+44
| | | | | | | | | We still decide which uniform to lower based on how many instructions-that-need-lowering use that uniform, but now we emit a new temporary uniform load in each of the basic blocks containing an instruction being lowered. This commit is best reviewed with diff -b.
* vc4: Convert vc4_opt_peephole_sf to work with control flow.Eric Anholt2016-07-121-4/+18
| | | | | | We need to apply the peephole pass to each of the blocks in the program. We don't do dataflow analysis for SF across blocks, but we also don't generate code that would need us to do so.
* vc4: Create a basic block structure and move the instructions into it.Eric Anholt2016-07-126-20/+122
| | | | | | | The optimization passes and scheduling aren't actually ready for multiple blocks with control flow yet (as seen by the "cur_block" references in them instead of iterating over blocks), but this creates the structures necessary for converting them.
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-25 05:00:02 +0000