Commit Graph

53 Commits

Author SHA1 Message Date
nmzik fbdc1e8714 shader: implement RDNA2 scalar quad-mask reduction 2026-08-21 03:32:48 +02:00
Senaxx 56a15c2036 Implement RDNA2 per-invocation scratch memory 2026-08-21 03:32:47 +02:00
nmzik 85cf965d54 refactor(shader): centralize typed resource metadata 2026-08-21 03:32:47 +02:00
nmzik 87166eed94 refactor(shader): reduce dispatcher state overhead 2026-08-21 03:32:47 +02:00
nmzik 6e9f162bd4 refactor(shader): compact paired 64-bit lowering 2026-08-21 03:32:47 +02:00
nmzik 8063822243 refactor(shader): preserve native scalar memory operations 2026-08-21 03:32:47 +02:00
nmzik c236ba713d refactor(shader): preserve native wide shared operations 2026-08-21 03:32:47 +02:00
nmzik c9e942a972 refactor(shader): compact whole-quad mask lowering 2026-08-21 03:32:47 +02:00
nmzik 5383972e96 refactor(shader): preserve native wide buffer operations 2026-08-21 03:32:47 +02:00
nmzik 3ea2099245 refactor(shader): simplify buffer access emission 2026-08-21 03:32:47 +02:00
nmzik 49145c1902 refactor(shader): emit structured phis natively 2026-08-21 03:32:47 +02:00
nmzik e1bcc298a0 refactor(shader): remove register-state compatibility paths 2026-08-21 03:32:47 +02:00
Senaxx 7e42513faf perf(shader): cache analysis (#259)
perf(shader): cache SPIR-V requirement analysis

Co-authored-by: nmzik <Nmzik@mail.ru>
2026-08-19 00:50:13 +02:00
nmzik 377fb8f50a shader: use ISA-style decoder opcode names 2026-08-18 03:53:45 +02:00
nmzik 330889e50b shader: implement RDNA2 frexp mantissa 2026-08-18 03:53:44 +02:00
nmzik 1be9568a02 shader: implement RDNA2 frexp exponent 2026-08-18 03:53:44 +02:00
nmzik e7c8bcb3a4 recompiler: deduplication cleanup 2026-08-18 03:53:44 +02:00
nmzik 0f3540b1ff shader: unify stage input metadata 2026-08-18 03:53:44 +02:00
nmzik 44302cbf96 shader: implement RDNA2 V_MED3_I16 2026-08-18 03:53:44 +02:00
nmzik 2a1a8e72e3 shader: preserve exec-masked FLAT address roots 2026-08-18 03:53:44 +02:00
nmzik 1375d650a7 graphics: support RDNA2 dependency waits 2026-08-18 03:53:43 +02:00
nmzik d7f1bcaa19 graphics: support signed 16-bit integer packing 2026-08-18 03:53:43 +02:00
nmzik 430de9bd15 graphics: support dynamic indirect image tables 2026-08-18 03:53:43 +02:00
nmzik 443bbcbf90 graphics: separate scalar address bindings 2026-08-18 03:53:43 +02:00
nmzik 83ee040b4d graphics: specialize invariant indirect image tables 2026-08-18 03:53:43 +02:00
Senaxx 980d89f739 graphics: support dynamic storage mip writes 2026-08-18 03:53:43 +02:00
nmzik 5e27a28ab0 shader: fix missing homogeneous semantics 2026-08-18 03:53:42 +02:00
nmzik 7556a74f19 shader: implement s_bfe_i32 2026-08-18 03:53:42 +02:00
nmzik f3cdf217ea shader: evaluate unsigned min in runtime descriptors 2026-08-18 03:53:42 +02:00
nmzik 28a913894d graphics: implement RDNA2 64-bit first-set-bit 2026-08-18 03:53:42 +02:00
nmzik 0ffacf22e6 graphics: accept phi-mediated SRT cycles 2026-08-18 03:53:42 +02:00
nmzik 13e87a8e58 graphics: implement RDNA2 alignbyte shader opcode 2026-08-18 03:53:42 +02:00
nmzik 09fec6815f shader: handle disabled RDNA2 traps 2026-08-18 03:53:42 +02:00
nmzik 4a7bc55afa shader: implement RDNA2 subrev borrow 2026-08-18 03:53:42 +02:00
nmzik 9c316e6988 shader: implement RDNA2 v_cmp_gt_u64 2026-08-18 03:53:42 +02:00
nmzik b42380ced6 shader_recompiler: add SSA 2026-08-18 03:53:41 +02:00
nmzik 21aa7b66ce graphics: replace raw GPU enum values with typed formats 2026-08-11 16:12:38 +02:00
nmzik 67cfec9063 shader: remove redundant runtime descriptor provenance walks (perf) 2026-08-07 18:08:40 +02:00
nmzik ac64d202af shader: implement BUFFER_ATOMIC_FMAX 2026-08-05 02:59:28 +02:00
nmzik df44ad3fac shader: implement BUFFER_ATOMIC_FMIN, unhandled IR opcode now fails 2026-08-05 02:59:28 +02:00
nmzik 0a2f4814ac refactor(shader): centralize opcode metadata and lookup paths 2026-08-05 02:59:28 +02:00
nmzik 822ca3c10b shader: implement S_CMP_EQ_U64 2026-08-05 02:59:27 +02:00
nmzik 2b6fd699e7 shader: implement V_CMP_EQ_I64 2026-08-05 02:59:27 +02:00
nmzik 53feb6cb19 graphics: log image descriptor specialization mismatches 2026-08-03 04:02:35 +02:00
Claxten 43f30d3ab2 graphics: shader: ignore unused sampler border state
* Sampler dword 3 only matters when a clamp mode uses border color
  (values >= 4). When no border mode is active, dword 3 is unused
  but can still vary across loop iterations due to wave-lane spills.
  This makes resource tracking think the descriptor is dynamic and
  fail with "unsupported GPU selection".

* Fix by zeroing dword 3 when all clamp modes are non-border.

Signed-off-by: Claxten <claxten10@gmail.com>
2026-08-02 02:42:15 +02:00
nmzik d68a477276 renderer: broaden compatibility 2026-07-31 22:24:56 +02:00
nmzik 846002c5eb shader: fix invalid texture descriptors and add missing format 2026-07-31 17:30:31 +02:00
nmzik d8a4c83cc7 format src and tests with clang-format 2026-07-31 11:36:12 +02:00
nmzik 68be13345a fix(shader): support multisampled depth image loads 2026-07-31 11:36:12 +02:00
nmzik 167da0abe0 shader: fix readlane/writelane for inactive host lanes 2026-07-31 11:36:12 +02:00