Searched refs:LDS (Results 1 - 21 of 21) sorted by relevance

/xsrc/external/mit/MesaLib/dist/src/amd/compiler/
H A DREADME-ISA.md116 ## `m0` with LDS instructions on Vega and newer
118 The Vega ISA doc (both the old one and the "7nm" one) claims that LDS instructions
121 In reality, only the `_addtid` variants of LDS instructions use `m0` on Vega and
123 LLVM also doesn't emit any initialization of `m0` for LDS instructions, and this
H A DREADME.md172 * LS and HS share the same LDS space, so LS can store its output to LDS, where HS can read it
185 * HW LS and HS stages are merged, and the merged shader still uses LDS in the same way as before
186 * HW ES and GS stages are merged, so ES outputs can go to LDS instead of VRAM
/xsrc/external/mit/MesaLib/dist/docs/relnotes/
H A D17.3.8.rst66 - radv: get correct offset into LDS for indexed vars.
H A D13.0.3.rst118 - radeonsi: wait for outstanding LDS instructions in memory barriers if
H A D20.3.1.rst146 - radv: don't count unusable vertices to the NGG LDS size
H A D20.3.5.rst275 - radv/llvm: Fix reporting LDS stats of tess control shaders.
277 - aco: Fix LDS statistics of tess control shaders.
H A D20.2.5.rst149 - radv: don't count unusable vertices to the NGG LDS size
H A D12.0.4.rst250 - radeonsi: fix 64-bit loads from LDS
H A D20.1.0.rst2080 - r600: Handle texcoord semantics in LDS index evaluation
2083 - r600/sfn: Add LDS IO instructions to r600 IR
2084 - r600/sfn: Add LDS instruction to assembly conversion
2088 - r600/sfn: Emit some LDS instructions
2094 - r600/sfn: Handle LDS output in VS
3846 - radv/gfx10: adjust the LDS size used to compute waves
4168 - aco: Extract LDS alignment calculation to a separate function.
4169 - aco: Remove esgs_itemsize from LDS alignment calculation.
4191 - aco: Only write TCS outputs to LDS when they are read by the TCS.
4192 - aco: Don't store TCS outputs to LDS whe
[all...]
H A D21.0.0.rst2086 - radeonsi: don't allocate LDS for TCS outputs if they are not read
2087 - radeonsi: limit HS LDS usage per workgroup to 16K to allow at least 2 WGs/CU
2096 - radeonsi: don't allocate LDS for TCS inputs if it's not used
2966 - radv: don't count unusable vertices to the NGG LDS size
3159 - aco: Skip TCS s_barrier when VS outputs are not stored in the LDS.
3168 - radv/llvm: Fix reporting LDS stats of tess control shaders.
3170 - aco: Fix LDS statistics of tess control shaders.
H A D20.3.0.rst3314 - radeonsi: don't count unusable vertices to the NGG LDS size
3316 - radeonsi: remove the NGG hack decreasing LDS usage to deal with overflows
3352 - radeonsi: don't execute LDS stores for TCS outputs that are never read
3529 - radeonsi: optimize out LDS bank conflicts in the NGG culling shader
3531 - radeonsi: write VS/TES system values into LDS after culling
3532 - radeonsi: pack LDS better for NGG culling
4378 - radv/llvm: reduce LDS size for tess by using NIR IO assigned locations
4526 - aco/ngg: Create LDS layout for NGG GS.
4532 - aco/ngg: Use more efficient LDS layout to help reduce bank conflicts.
4541 - aco: Add a few assertions about LDS usag
[all...]
H A D19.3.0.rst2859 - aco: take LDS into account when calculating num_waves
3066 - radv/gfx10: compute the LDS size for exporting PrimID for VS
3100 - radv/gfx10: declare a LDS symbol for the NGG emit space
3109 - radv/gfx10: adjust the LDS size for VS/TES NGG streamout
3110 - radv/gfx10: fix unnecessary LDS overallocation for NGG GS
H A D20.0.0.rst2330 - radeonsi/gfx10: don't declare any LDS for NGG if it's not used
2851 - aco: improve readfirstlane after uniform LDS loads
3049 - ac: handle pointer types to LDS in ac_get_elem_bits()
3148 - radv/gfx10: do not declare LDS for NGG if useless
H A D21.1.0.rst4755 - aco: consider that GFX10.3 allocates LDS in 1024 byte blocks
4759 - radv,aco: allow unaligned LDS access on GFX9+
4771 - Revert "radv,aco: allow unaligned LDS access on GFX9+"
5131 - ac/rgp: report LDS size in CU mode on GFX10+
5353 - aco: Fix LDS statistics of tess control shaders.
5354 - radv/llvm: Fix reporting LDS stats of tess control shaders.
5387 - radv: Calculate tess patches and LDS use outside the backend compilers.
H A D20.2.0.rst2249 - r600/sfn: use modern c++ in printing LDS read instruction
3392 - radeonsi: don't count unusable vertices to the NGG LDS size
3394 - radeonsi: remove the NGG hack decreasing LDS usage to deal with overflows
4475 - radv: align the LDS size in calculate_tess_lds_size()
4476 - radv: set LDS TCS size at shaders creation for GFX9+
H A D21.3.0.rst2702 - radeonsi: enable NGG passthrough when LDS is used, document the real constraints
4138 - aco: Consider LDS usage by PS inputs in MaxWaves calculation.
4171 - aco: Don't write m0 register for LDS instructions on GFX9+.
H A D19.1.0.rst4481 - radv: take LDS into account for compute shader occupancy stats
H A D21.2.0.rst4452 - radv: improve LDS alignment check for load/store vectorization
/xsrc/external/mit/MesaLib/dist/src/gallium/drivers/nouveau/codegen/
H A Dnv50_ir_target_gv100.cpp171 OPINFO(LDS , NONE, NONE, NONE, NONE, NONE, NONE);
/xsrc/external/mit/MesaLib.old/dist/src/mesa/x86/
H A Dassyntax.h509 #define LDS(a, b) CHOICE(ldsl ARG2(a,b), lds ARG2(a,b), lds ARG2(b,a)) macro
1226 #define LDS(a, b) lds b, P_(a) macro
/xsrc/external/mit/MesaLib/dist/src/mesa/x86/
H A Dassyntax.h509 #define LDS(a, b) CHOICE(ldsl ARG2(a,b), lds ARG2(a,b), lds ARG2(b,a)) macro
1226 #define LDS(a, b) lds b, P_(a) macro

Completed in 74 milliseconds