nimlgen/tinygrad
1
0
Fork 0
forked from tinygrad/tinygrad
Code Pull Requests Activity
11,925 Commits 334 Branches 15 Tags
amd_refac
Commit Graph
854 Commits
Author SHA1 Message Date
George HotzandGitHub 88bc5ee212 assembly/amd: rename to better names (#14384) 
* assembly/amd: rename to better names

* might help fuzzing segfault

* emu2 -> emu
 2026-01-28 10:00:54 +08:00
chenyuandGitHub cd22ee9ed0 add InvalidType to ConstType [pr] (#14373) 
* add InvalidType to ConstType [pr]

TYPED=1 python test/test_tiny.py passes.
added PyConst = float|int|bool for some Tensor level input types

* hcq
 2026-01-27 14:09:34 -05:00
chenyuandGitHub db010a31be IGNORE_OOB -> CHECK_OOB [pr] (#14374) 
flip the meaning
 2026-01-27 12:20:59 -05:00
sirhcmandGitHub c9c533fc78 libclang path is homebrew on macos (#14357) 
* libclang path is homebrew macos

* typo

* ugh

* typo

* regen

* no LIBCLANG_PATH
 2026-01-26 17:32:09 -05:00
qazalandGitHub 2d91fe6310 use amdgpu dsl in mmapeak (#14342) 
* use amdgpu dsl in mmapeak

* don't rely on llvm for vgpr counting

* llvm roundtrip assert

* rm it, add ci

* vgpr_count

* move emulated test to amd, it needs comgr

* env

* arch

* inst._fields -> inst.operands

* vgpr offset
 2026-01-26 22:03:43 +09:00
qazalandGitHub b2e2ace85b viz: remove ci check, it's VIZ=-1/-2 (#14343) 2026-01-26 20:36:23 +09:00
George HotzandGitHub be23776ba7 assembly/amd: replace pcode with ucode (#14002) 
* a bunch of todos for my boy claude

* uops have types

* lil cleanups

* simpler ucode

* isNAN

* calls

* move more

* cleanup pcode_parse

* cvt functions

* fix parser bugs

* no void

* minmax

* more pcode parse

* pretty print

* transform

* comments

* move to transform

* assign/declare

* simpler norm

* single PM

* just Uops

* simpler

* more typed

* all rewrite

* less verbose

* work

* spec

* transform

* work

* simpler spec

* less spec

* bitcast

* simpler

* simp ucode

* work

* more in pcode_transform

* remove junk

* more functions

* bug

* no void assign

* load/store

* wave

* fixes

* move denorm

* move more functions

* tests

* cat is shape None

* uop syntax

* move a few more

* program_spec

* cat stuff

* assign fix clear

* unused

* nans

* fp bits

* works with simplify

* remove junk

* special

* meh

* more

* more

* update test pcode parse

* improve parser

* parse some for loops

* merge master

* dead files

* tests pass

* emu2

* better emu2

* test_plus works

* uselessly write more instructions

* use pcode

* something

* something

* bench_emu

* progress

* ds works

* work

* work

* more passing

* run compare

* bench_emu

* more pcode

* a few more

* bugfixes

* bugfix

* test fixes

* tests pass without USE_HW

* all hw tests pass

* add more hw tests

* new hw tests

* bit

* less handcode

* parse more

* consolidate pcode

* fixes

* rsrc

* lane pcode

* cleanups

* simpler

* emu bugs

* one cmp test fails

* fix decode and upd name

* fix name and test harness

* _ftz_f32

* fix denorm

* fix VOPD and use load

* fix carry bug

* no load where / just invalid

* clean

* simpler

* merge sops

* refactoring

* simplifications

* bugfixes

* new tests

* f16 sin fix

* assertion and hw tests

* cvt functions

* one more failure

* bugfixes

* bugfix + regression

* more tests

* fmac

* no manual unrolling

* ordering

* LLVM backend is a lot faster

* compile inst

* more bugs

* f16

* bugfix

* fix regression

* one clang call

* 1M inst

* scratch works

* do scratch correctly

* cleanup

* regression

* cmp

* fmamk fixes

* merge

* fix vcmpx

* unify memory

* remove unused code

* ignore oob for test

* cleanups

* fix mbs

* unify cmp

* test

* minor cleanups

* bump timeout

* fix tests

* revert the CMPLE stuff

* remove opt

* less diff

* simpler

* revert

* support multiple backends

* memset is a lot faster

* split out in bench emu

* improve timing

* timing

* cache that

* cache that

* simpler and faster

* tokenize

* binop table

* simpler

* move to parser

* tok for lambda

* refactor

* expr_parser

* delete emu2_pcode

* import cleanup

* lil

* if parse

* work

* simpler

* no v

* trig preop is faster

* durations for tests

* fix cmp bug

* sdst

* remove scartch_size hack

* null behavior

* _MXCSRContext

* bugfixes

* DEBUG >= 3

* test smem crashes my gpu

* debug

* test

* test smem

* profiler

* full inst

* bugfix

* rtag(1)

* pc is 64-bit and word

* pc is real code now

* dynamic

* more dynamic

* fix oob access

* fix crash, more dyn

* all dyn

* really all dyn

* correct null mask

* lit + format

* 21s on the tests

* 13s on the tests

* canonical name

* simm16

* more dyn

* 14s

* proper saddr dedup

* dyn

* debug 5

* better 5

* revert dynamic stuff

* that can be dyn

* negative offsets

* dyn wmma

* f16 wmma support / ops / dtype / dtype_alu

* symbolic changes not needed

* ConstFloat

* more uop.const

* __eq__

* uop tests

* fix f16

* bf16 tensor cores

* whitespace

* remove cast roundtrip

* Revert "remove cast roundtrip"

This reverts commit c5bb0381c3.

* just the fix

* remove dead paths

* llvm runs
 2026-01-26 18:04:29 +08:00
nimlgenandGitHub 21ab23ae18 nv: add pma for ada (#14328) 
* nv: add pma for ada

* um

* fix

* shorter

* mock
 2026-01-25 17:33:37 +03:00
chenyuandGitHub 7e41da1ae8 fix generate_dataset.sh (#14324) 
added `set -e` so wrong pathes would fail the script, then fixed the path
 2026-01-24 16:47:10 -05:00
chenyuandGitHub e04767e39e run pre-commit in ci (#14253) 
* run pre-commit in ci

prevents pre-commit regression

* IGNORE_OOB=1

* pytest

* unit test

* split
 2026-01-20 12:24:33 -05:00
qazalandGitHub b1c5a242b7 Revert "move is_dtype_supported logic to renderer (#14188)" (#14237) 
This reverts commit 161fee9a48.
 2026-01-20 12:19:14 +09:00
sirhcmandGitHub 161fee9a48 move is_dtype_supported logic to renderer (#14188) 
* move is_dtype_supported logic to renderer

* fix CPU_COUNT

* mypy happy

* early import libclang too with llvm

* run with debug

* skip autogen tests if MTLCompiler or llvm is loaded

* run autogen tests separately in CI

* lint
 2026-01-18 22:37:04 -05:00
sirhcmandGitHub 1eb110cd7d fix memory corruption in NIR, reenable process replay (#14204) 2026-01-18 02:05:12 -05:00
qazalandGitHub feaa804158 skip lvp process replay in CI [pr] (#14202) 2026-01-18 13:25:04 +09:00
George HotzandGitHub e9ce12028e assembly/amd: amdxml cleanups, remove broken SDWA/DPP, merge in pdf.py (#14154) 
* assembly/amd: amdxml cleanups, remove broken SDWA/DPP

* remove buf junk

* simplify

* simplify

* lil cleanup

* dead fixes

* strip non pcode extraction from pdf

* merge pdf.py into amdxml.py

* only amdxml
 2026-01-15 09:23:19 +09:00
chenyuandGitHub 986e865830 fix TINY_BACKEND=1 cumsum (#14138) 
* fix TINY_BACKEND=1 cumsum

old hack was wrong, need to apply contiguous on the input

* test time

* test_linalg_svd is slow
 2026-01-14 09:54:49 -05:00
George HotzandGitHub 2ab18ea7e3 assembly/amd: use xml instead of pdf (#14118) 
* assembly/amd: use xml instead of pdf

* use amdxml to generate info about op sizes

* fix many tests with invalid instructions

* fix info generation

* chad xml fixes many bugs

* rename to operands

* simplify

* amdxml

* bug fix
 2026-01-14 10:03:37 +09:00
George HotzandGitHub 8b1b15aec0 assembly/amd: SQTT support (#14099) 
* assembly/amd: SQTT support

* simpler

* cmp wave

* instruction compare

* rocprof decode

* simpler

* no llvm

* no strcmp
 2026-01-12 05:07:17 +09:00
chenyuandGitHub 92246ea731 update tests, WEBGPU=1 pytest . passes (#14089) 
* update tests, `WEBGPU=1 pytest .` passes

* minor update
 2026-01-10 00:03:02 -05:00
George HotzandGitHub 20653d2996 assembly/amd: make pdf.py code shine (#14029) 
* assembly/amd: make pdf.py code shine

* no merge

* pdf2 is the future

* something

* regen enums

* test

* work

* remove junk

* write

* pcode extraction

* pdf2 passes all tests

* simplify

* simpler pdf

* late filter

* remove hacks

* simplify pdf2.py

* field type

* remove defaults

* don't export srcenum

* simple pdf.py

* simpler

* cleaner

* less hack in PDF
 2026-01-05 18:49:40 -08:00
sirhcmGitHubClaude Sonnet 4.5
 		b2a0b9c551 autogen: dump patch in CI (#14010) 
* autogen: don't fast-fail, produce patch artifact on differences

All verification steps now use continue-on-error to run completely.
Each job generates a patch artifact containing all differences found.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <[email protected]>

* add gen from header test

* fix tests

* fail if diff

* add forward decl autogen test

* remove confusing/wrong comments

* macos unittests set LIBCLANG_PATH

---------

Co-authored-by: Claude Sonnet 4.5 <[email protected]>
 2026-01-04 22:38:12 -05:00
George HotzandGitHub 8328511808 assembly/amd: make the emu.py code shine (#13996) 
* assembly/amd: make the code shine

* lil clean

* reg back in pcode

* cleanups

* gen fma_mix

* no writelane hacks

* fn cleanup

* dead vgpr_write

* readable

* smem

* cleanup bench_emu

* speedups

* simpler and faster

* direct inst._fn

* split fxn

* Revert "simpler and faster"

This reverts commit e85f6594b3.

* move lds to wavestate

* dispatcher

* pc in dispatch

* literal isn't wavestate

* cleanups + program

* one readlane

* exec_vop3sd in exec_vop

* cleaner exec_vopd

* fully merge VOP3P

* no special paths

* no SliceProxy

* low=0

* no bigint

* failing tests

* fma on python 3.13
 2026-01-03 20:33:09 -08:00
sirhcmandGitHub 35c2870b1f gate image_conv2d pitch hacks on IMAGE==1 (#13995) 
* gate image_conv2d pitch hacks on IMAGE==1

* fix opencl image copies

* cleanup
 2026-01-03 12:27:31 -05:00
sirhcmandGitHub 9dc524536f IMAGE=1 creates "dynamic" images (#13769) 
* remove image from BufferSpec

* cl tiny_gemm (64) works

* mypy

* padding

* openpilot CL

* reshape properly

* remove extra qcom checks

* pad output

* mypy

* update compile test

* move undo

* TestImageCopy valid images

* TestImageRealization valid images

* TestImageDType valid images

* cleanups

* test_renderer_failures

* ruff

* mypy

* simplify ops_qcom

* bump step time

* Revert "bump step time"

This reverts commit 75a037c7d0.

* "dynamic textures" are optional

* a start

* IMAGE=1 works, no FLOAT16

* fast but wrong

* mypy

* some fixes

* better

* works

* refactor

* oops
 2026-01-02 16:22:39 -05:00
George HotzandGitHub dfb813b760 assembly/amd: add pcode ds ops (#13939) 
* assembly/amd: add pcode ds ops

* refactors

* fix ds op

* update autogen

* fix flat bug

* more tests

* fix emu test

* that's a hack

* generic

* fix all tests

* two tests

* fix test failure

* better

* remove __all__
 2026-01-01 16:24:13 -05:00
chenyuandGitHub e2987001ee unify pre-commit mypy and ci mypy (#13940) 2025-12-31 17:51:51 -05:00
chenyuandGitHub a9a7b33404 IGNORE_OOB=0 in CI (#13903) 2025-12-31 12:56:59 -05:00
chenyuandGitHub ba9aa5cd6f skip some PTX IGNORE_OOB validation (#13927) 2025-12-31 12:40:21 -05:00
chenyuandGitHub 4968060ad4 fix IGNORE_OOB=0 for WEBGPU (#13926) 2025-12-31 10:41:28 -05:00
chenyuandGitHub 404755bafd merge ci ruff tests and update ruff version (#13922) 2025-12-31 09:53:49 -05:00
chenyuandGitHub dc27eb48ac remove PYTHONPATH="." from test.yml (#13909) 2025-12-30 17:00:16 -05:00
George HotzandGitHub efc99d0c55 assembly/amd: more refactors (#13907) 
* assembly/amd: more refactors

* more refactors

* more refactors

* simpler emu

* generate.py

* regen all

* cleanups

* more

* work

* more readme

* lil
 2025-12-30 16:13:24 -05:00
George HotzandGitHub 69cdc8066d assembly/amd: add dtype tests to AMD IDE CI (#13899) 
* add dtype tests to AMD IDE CI

* more tests

* add trig preop

* regen done

* split to amd autogen

* simpler
 2025-12-30 11:09:51 -05:00
George HotzandGitHub 2b838dc1d8 assembly/amd: fix AMD_LLVM=1 support in emulator (#13881) 
* fix AMD_LLVM=1 support in emulator

* more llvm with dtype

* work

* more fixes

* fix dtype
 2025-12-30 09:09:57 -05:00
George HotzandGitHub 9d8397be11 add CDNA3+RDNA4 support (#13882) 
* fix CI

* remove junk

* rename lib to dsl

* correct

* cleanups
 2025-12-29 15:51:29 -05:00
George HotzandGitHub 81cf9ea0ab rename to extra.assembly.amd (#13879) 2025-12-29 14:10:55 -05:00
George HotzandGitHub 37f0fa11b6 rdna3 test cleanups (#13878) 
* rdna3 test cleanups

* cleanups

* ugh DONT SKIP
 2025-12-29 13:41:59 -05:00
George HotzandGitHub f1471a3b99 speed up rdna3 unit tests + add to CI (#13871) 
* speed up rdna3 unit tests

* add test to CI

* faster and simpler

* speedups

* bugfixes

* use helper

* fix CI maybe

* test fixes

* llvm-21 on 24.04

* upd

* llvm-21

* fix test

* bring that back

* merge gen into lib

* test generators
 2025-12-29 10:26:48 -05:00
chenyuandGitHub cba05acadf re-enable TYPED=1 import test (#13858) 2025-12-28 11:49:06 -05:00
qazalandGitHub a1c1684b91 set .amdhsa_kernarg_size in asm test (#13826) 2025-12-25 13:08:14 +09:00
chenyuandGitHub 80b84f5267 ruff lint tinykitten (#13762) 
deleted used import and double spaces. a few ignore to not change the real code
 2025-12-19 14:31:00 -05:00
sirhcmandGitHub 97103831c5 Revert "remove image from BufferSpec (#13636)" (#13761) 
This reverts commit 2571a1eb47.
 2025-12-19 13:54:36 -05:00
sirhcmandGitHub 2571a1eb47 remove image from BufferSpec (#13636) 
* remove image from BufferSpec

* cl tiny_gemm (64) works

* mypy

* padding

* openpilot CL

* reshape properly

* remove extra qcom checks

* pad output

* mypy

* update compile test

* move undo

* TestImageCopy valid images

* TestImageRealization valid images

* TestImageDType valid images

* cleanups

* test_renderer_failures

* ruff

* mypy

* simplify ops_qcom

* bump step time
 2025-12-19 13:41:20 -05:00
George HotzandGitHub 4b741e893f remove REMOTE=1 (#13722) 
* remove REMOTE=1

* leave ibverbs
 2025-12-16 15:58:10 -04:00
George HotzandGitHub e5a66ace80 multi custom kernel support (#13716) 
* multi custom kernel support

* custom kernel xfrom

* works

* no SPEC=2 on ck

* panic

* touchups
 2025-12-16 11:36:30 -04:00
George HotzGitHubClaude Opus 4.5
 		316da9f7ff llm: add created/model fields, non-streaming support, and tests (#13660) 
* llm: add created/model fields, non-streaming support, and tests

- Add `created` timestamp and `model` fields to response (required by OpenAI spec)
- Add non-streaming mode support for /v1/chat/completions
- Add `send_data` helper to HTTPRequestHandler for responses with Content-Length
- Refactor viz/serve.py to use send_data
- Add integration tests using real OpenAI client

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <[email protected]>

* add openai to testing

* toml

* Remove 'openai' from dependencies

Removed 'openai' from the dependencies list.

* bump cache

---------

Co-authored-by: Claude Opus 4.5 <[email protected]>
 2025-12-12 14:50:36 -05:00
George HotzandGitHub f0fa9bcd98 openai api for llm (#13648) 
* openai api for llm

* responds to simple request

* schedule cache needs to unbind

* stream works

* share stream code

* 20k

* one print

* cid
 2025-12-12 08:25:33 -05:00
sirhcmandGitHub cb3d756547 NAK compile-only test (#13621) 2025-12-08 15:53:46 -05:00
sirhcmandGitHub a4c3d48aa9 compile-only test for IR3 actually works (#13619) 2025-12-08 15:07:49 -05:00
sirhcmandGitHub 1c16b6e082 Mesa: freedreno (#12746) 
* ir3 init

* got a program

* 1 + 1 works

* use isa_disasm instead of shader_disasm

* wip

* matmul works

* works on py3.14

* fix const loading

* skip QCOM failing tests

* cleanup

* args actually work

* add compile-only tests

* fix typo and install tinymesa

* IR3 NULL backend

* (float32) images work

* autogen fix

* fix compile only test

* typo

* mypy happy

* compile-only uses py3.14

* bump mesa

* unify qcom disassembler

* float16 works

* disasm shows in viz

* save a line

* add real del

* variable workgroup sizes

* simplify diff

* bump line count

* properly set wgsz

* regen mesa

* no preamble

* bump lines
 2025-12-08 14:02:08 -05:00