Commit Graph

Select branches

Hide Pull Requests

abstractions_4

add_local_buffer

after_qol

algebraic_upat

all_bv_no_idx

all_sink

amd_isel

amd_refac

amd_smu_reset_patch

amd_sqtt

amd_uop

amd_uop_cleanups

assign_copy

assign_try_3

before2

better_spec

better_syn

bitcast_spec

broadcast_shape_2

bt_ei

buf_heur

bump_nv_2404

cache_reshape

call_inline

call_is_kernels

callify

cast_vec_as_cast

cat_mop

changes_from_rangify

check_clsetarg

check_dims

cifar_30

cifar_30_clean

clean_load

clean_spec

clone_tg

codegen_try_2

contig_on_after

continue_mat

deepsek

default_late_allreduce

delete_kernel_py

delete_ones

delete_slow_rangeify

delta_attention

delta_attention_scan

demote_op

disable_rules

dsp_search

dsp_search_merged

dtype_shape

earlier_gater

emu_cdna

emulate_ctx

end_all_ranges

endrange

er_prioity

execution_unit

failing_assign

fancy_usb

fast_llm

fix_fa_fwd_clang_22

fix_fuse

fix_llvm_cast

fix_range_merging_sd

fix_rangeify_tests

fix_rc

fix_sched_copy

fix_slice_store

fix_src_spec

fixup_fix_kernel

full_spec

fun_w_egraphs

gate-on-load-store

gate_rewrite

gemma4_gpt

gen_pdf_fast

gh-pages

gl_dims

gldims

glm_flash

grad_outerworld

grok_x86_novec

group_div_rules

hcq2_vram_copies

heur_4096

heur_group

hk_gemm

http_threading

hwtest_fixes

id_store

if_range_cleanups

ignore_oob

image_idx_xy

image_in_mc

image_no_vec

image_refactors_stack

index_to_slice

invalid_try_3

jinja_template

kernel4

kernel_is_call

kernelless

kimi_slop

kitten_matmul

late_Locals

late_add_load

late_allreduce

late_codegen

late_decomp_is_late

late_lower_idx

late_view

late_weakint

lds_double_buffering

less_wmma_arg

lil_dg_change

lil_rangeify_td

llama_trainer

llm_262k_minimal

llm_flash_attn

llm_qwen

llm_serve_fixes

load_view_late

local_cache

locals_have_device

loop_end_op

loop_is_rng

lowerer_hacks

mac_pytest

manual_kernel_3

master

mega_lowerer

min_new_lin

min_outer_range

minor_views

mixins7

mlperf_training_v4.1

mlperf_training_v5.0

mlperf_training_v6.0

modern_scan

more_apps

more_early_comps

more_folding_rules

more_kernelcount

more_pcode

more_rangeify_pm_tests

more_ren_clean

more_uop_gc_test

more_uop_mm

more_vmap

move_devec

move_dtype_decomps

move_expander

move_gates_to_load_store

move_mop_cleanup

move_schedule

move_simp_views

moveleftright

mt_typing

multi_range

multioutput

new_ck

new_devec

new_invalid_and_image

new_linearizer

new_shape

new_test

no_assign_2

no_buf_0

no_decimals

no_late_simplify_marg

no_merge_ends

no_merge_views

no_mstack_in_allreduce

no_one_ranges

no_sip_usbgpu

no_tsink_base

no_tuplize

no_user_ck_contig

no_vec_x86_4

no_weakint_variable

noop_index

novecimg

num_batches_tracked_shape

olmo3_support

one_compiler

one_hcopt

only_assign_buffers

only_loop_backends

only_reg_emu

only_reg_emu2

outer_range

outerworld_work

pad_invalid_work

pad_mixin

param_call_is_function

parse_sqtt_more

pci_iommu

port_nir

postrange_hcopts

precompile_backward

prefill_qwen_fast

print_special_ops

ps_clean

pyremu_try

python_prg

python_speed

q4k_support

qcom-openpilot-13ms

qcom_830

qcom_mmapeak

qcom_speed_codex

qol_string

qwen36_27b_amd_900

qwen36_27b_amd_minimal

qwen36_amd_bugfixes

qwen_mergable

r_cleanups

r_new_opt

ranged_call

rangeify_2

rangeify_const_folding

rdna3_vibes

rdna4_asm

rdna4_gemm

real_pyrender

recurse_sub

reg_coalese

relu_pattern

remove_assign

remove_cast_hack

remove_ck_contig

remove_define_var

remove_forced_re

remove_on_stack

remove_programspec

remove_pylint

remove_slow_tests

remove_vec_2

remove_vec_3

renderers_use_shapes_2

replace_if_with_range

reprocess_node_experiments

reshape_trait

resnet-ci

revert-11547-fix_mismatch_red

revert-11715-revert-11711-amd_llvm_tc_cast

revert-11809-no_ki_in_gpudims

revert-11811-better_div_nesting

revert-11876-memplan_vbig

revert-12104-simplify_valid_in_symbolic

revert-12308-bufferize_early

revert-12731-jitlink_init_check

revert-12813-after_cleanups

revert-13046-better_cleanup_arange_buffers

revert-13311-prof_link_sqtt

revert-13583-rocr_desc_2

revert-14734-tload

revert-17141-full_rewrite_to_sink

revert-17408-_broadcasted-weak

revert-9438-simple_matmul_np_rand

revert_wrong_ck_fix

rewrite_rangeify2

rtoposort

scan_assign

scan_op_8

sched_cache

sched_lin

sd_load_simple

self_type

set_testpath

shrink_in_render

simp_spec

simple_gidxs

simple_pool

simple_priority

simple_rangeify

simpler-sd-compile

simpler_end

simpler_fusion

simpler_image

simpler_postrange

simpler_reduce

slice_like_shrink

spec_3

spec_cleanup

spec_ren

speed_one_linear

split_buf_idx

split_decomp_pass

split_sub

stack_0

store_is_void

string_ranges

support_opts_in_contig

swizzle_in_kernel

sym_llm

sym_work

symbolic-chunk-split

szdiff_bot_test

tc_is_heur

tensor_graph_single_sink

test_amd_crash

test_fa

test_from_postopt

test_hgemm

test_unit

tiny_scan

tmr

tool_call_support

topoprop

ttomsa-new_linearizer

tuplegtuple

two_stage_remove

typed_checks

typing_work

unused_permute_arg

uop_prg

upd_test4

update_benchmark

update_benchmark_search

update_mlperf

user_failure

vecless_x86_glm

verify_sink_dims_clean

vf_ish

view_after_const

view_in_codegen

viz_cbroadcast

warp_fun

wgpu-f16-packed

working_postopt

x86_moves

x86_numel

v0.10.0

v0.10.1

v0.10.2

v0.10.3

v0.11.0

v0.12.0

v0.13.0

v0.4.0

v0.5.0

v0.6.0

v0.7.0

v0.8.0

v0.9.0

v0.9.1

v0.9.2

Mono Color

• 4dd10d03b7 move is_increasing to ops [pr] (#9134) George HotzandGitHub 2025-02-17 19:27:48 +08:00
• 22c571d3cb add kernel axis colors to viz [pr] (#9129) qazalandGitHub 2025-02-17 13:21:35 +02:00
• 1bf66d62cf symbolic gets its own file [pr] (#9132) George HotzandGitHub 2025-02-17 18:55:21 +08:00
• bd694faf6c factor out the expander logic [pr] (#9131) George HotzandGitHub 2025-02-17 18:09:48 +08:00
• 5bdf0c7951 Bitcast constant folding 2.0 (#9089) 3 people 2025-02-17 11:08:20 +01:00
• 2be4529f14 Test broken const folding wraparound behavior (#9080) 3 people 2025-02-17 10:44:56 +01:00
• 7eea9b639d hotfix: add replay_pkl debugging env geohot 2025-02-17 17:34:58 +08:00
• af9d8d39d2 dsp matchers + bump line count to 11300 (#9130) George HotzandGitHub 2025-02-17 17:31:54 +08:00
• 638d925e4e Prevent const folding in test_payne_hanek_reduction (#9088) quortusandGitHub 2025-02-17 10:31:10 +01:00
• 9289425170 add ast to ProgramSpec + pre matcher [pr] (#9128) George HotzandGitHub 2025-02-17 16:39:14 +08:00
• fe260ac4d7 viz/server cleanups [pr] (#9127) qazalandGitHub 2025-02-17 09:59:41 +02:00
• a38b47e026 hotfix: DSP doesn't use that path geohot 2025-02-17 10:45:29 +08:00
• edf7213f34 Make bitcast to the same dtype noop (#9121) quortusandGitHub 2025-02-17 02:28:44 +01:00
• 59fe45f947 Solve get_grouped_dims does not split issue (#9085) 3 people 2025-02-17 01:57:29 +01:00
• 84dc331dd1 Refactor async (#9126) Ahmed HarmoucheandGitHub 2025-02-16 23:47:15 +01:00
• 6a9e5598f9 small viz touchups [pr] (#9123) qazalandGitHub 2025-02-16 21:07:40 +02:00
• b3127f38e6 faster viz data fetching with streaming [pr] (#9122) qazalandGitHub 2025-02-16 20:31:11 +02:00
• 8926bac00a am: profiling working (#9119) uuuvnandGitHub 2025-02-16 20:51:08 +05:00
• 97cb9cb1ed always viz the first graph + non blocking matches fetch [pr] (#9117) qazalandGitHub 2025-02-16 14:39:51 +02:00
• 1fda98d14f fix import time_linearizer [pr] (#9118) chenyuandGitHub 2025-02-15 21:33:28 -05:00
• c1dfe5c00d compact get_late_rewrite_patterns [pr] (#9116) chenyuandGitHub 2025-02-15 20:33:09 -05:00
• 2e97022e5e remove extra block in viz [pr] (#9115) qazalandGitHub 2025-02-16 02:38:09 +02:00
• fd95543ff1 user scatter_reduce in scatter [pr] (#9114) chenyuandGitHub 2025-02-15 18:21:01 -05:00
• c954419bc8 minor tweak to transcendental pow (#9112) chenyuandGitHub 2025-02-15 18:03:25 -05:00
• 8dfa0024f0 raise in scatter if self and src have different dtype [pr] (#9109) chenyuandGitHub 2025-02-15 11:21:34 -05:00
• d129ccda4c add RAWAST back to DEBUG=3 [pr] (#9107) chenyuandGitHub 2025-02-15 09:12:51 -05:00
• 2e19976d03 assert views in tensor uops [pr] (#9106) qazalandGitHub 2025-02-15 13:27:55 +02:00
• 81f5a7af7d improve DEBUG=3 [pr] (#9105) George HotzandGitHub 2025-02-15 18:44:56 +08:00
• 41d143d27c new order to prepare for becomes_map = tensor_map [pr] (#9104) qazalandGitHub 2025-02-15 11:37:36 +02:00
• 4672d9af73 actual tests for the dsp backend [pr] (#9102) George HotzandGitHub 2025-02-15 15:17:56 +08:00
• 7e09057afa fixup clang devectorize (#9099) George HotzandGitHub 2025-02-15 09:29:47 +08:00
• 8824f7e9df Make logcumsumexp numerically stable (#9050) 3 people 2025-02-15 01:25:17 +01:00
• 81597ddd96 increase lr for bert (#9098) chenyuandGitHub 2025-02-14 19:10:35 -05:00
• 3ad39b247b refactor LLVMRenderer (#9090) 3 people 2025-02-15 08:00:31 +08:00
• 1f1362fd27 add truncate_bf16 (#9078) 3 people 2025-02-15 07:59:09 +08:00
• 2dc8f1867c Synchronize webgpu (#9093) Ahmed HarmoucheandGitHub 2025-02-14 22:52:10 +01:00
• b58e7b1898 zero out the weight in bert init run (#9076) chenyuandGitHub 2025-02-14 08:40:41 -05:00
• 82ad0d2e65 keep CONST/BUFFER uops in tensor_map [pr] (#9083) qazalandGitHub 2025-02-14 14:50:08 +02:00
• 65297066c2 move buffer refcount increment to the toposort [pr] (#9081) qazalandGitHub 2025-02-14 13:54:22 +02:00
• 73af42aeab fix pow backward when base is 0 (#9075) chenyuandGitHub 2025-02-13 21:06:01 -05:00
• 2d04a75a40 start tracking bottom_up_rewrite in viz [pr] (#9071) qazalandGitHub 2025-02-14 01:28:10 +02:00
• 5ef48bbe0a swap order in rsqrt (#9069) chenyuandGitHub 2025-02-13 16:51:21 -05:00
• e83905696e Show install instructions when dawn library is missing (#9059) Ahmed HarmoucheandGitHub 2025-02-13 22:30:20 +01:00
• 9e91898941 bert eval at the end of training (#9070) chenyuandGitHub 2025-02-13 16:29:44 -05:00
• e02e3b94c3 remove SQRT hack in llvm (#9067) chenyuandGitHub 2025-02-13 15:42:34 -05:00
• 947c97e6ff add test_sqrt to test_speed_v_torch (#9066) chenyuandGitHub 2025-02-13 15:25:54 -05:00
• 49abc09f77 remove the reshapes in test_arange_2_reduce [pr] (#9063) chenyuandGitHub 2025-02-13 12:33:25 -05:00
• 2573d0621a Tensor.scatter_reduce touchup [pr] (#9060) chenyuandGitHub 2025-02-13 10:01:14 -05:00
• 1f9d2442b9 Add Tensor.scatter_reduce (#8947) 3 people 2025-02-13 09:08:54 -05:00
• 2b9ce1235a simple failing case for reorder expand + keep views in tensor_map [pr] (#9057) qazalandGitHub 2025-02-13 12:22:55 +02:00
• 765a936b81 getenv(CC) for clang (#9054) George HotzandGitHub 2025-02-13 15:30:01 +08:00
• 33a1151f2f Revert "match torch rmsnorm implementation (#6799)" (#9052) George HotzandGitHub 2025-02-13 14:42:45 +08:00
• a66b8250e0 match torch rmsnorm implementation (#6799) 3 people 2025-02-13 15:02:51 +10:00
• 19ae829bd1 test float uop in sym_infer (#7456) ggandGitHub 2025-02-12 23:55:28 -05:00
• 095504d094 mulacc_unrolled should happen even with no DEVECTORIZE (#9029) 3 people 2025-02-13 05:54:15 +01:00
• 74742c018f hotfix: setup_mock_nv_osx geohot 2025-02-13 12:25:15 +08:00
• d2ff55e9c6 OSX GPUOcelot (#8209) 4 people 2025-02-13 05:24:29 +01:00
• f4f56d7c15 move time_linearizer to extra.optimization.helpers [pr] (#9048) chenyuandGitHub 2025-02-12 15:49:58 -05:00
• c15486cf39 remove contiguous in test_subbuffer_used [pr] (#9046) chenyuandGitHub 2025-02-12 14:41:16 -05:00
• b3eab03055 Three things to get Windows CI working correctly: (#9047) rmtewandGitHub 2025-02-13 08:41:00 +13:00
• f53b819648 UOps. -> Ops. [pr] (#9044) chenyuandGitHub 2025-02-12 12:53:23 -05:00
• 6811688d29 disallow VIEW(BUFFER) in tensor [pr] (#9041) qazalandGitHub 2025-02-12 18:27:35 +02:00
• 7b5ac2c15e free_intermediates in bert (#9040) chenyuandGitHub 2025-02-12 10:00:39 -05:00
• 916d5e7f08 WebGPU f16 support (f16 bounty part 2) (#8653) Ahmed HarmoucheandGitHub 2025-02-12 12:46:53 +01:00
• aaed315fee add AMX support to LLVM (#8957) 3 people 2025-02-12 05:01:18 -03:00
• 0c97c10814 TestOps: silence pytorch std()/var() degrees of freedom warnings (#9034) Josh MooreandGitHub 2025-02-12 01:49:18 -05:00
• d581afd873 skipdata capstone (#9026) Ignacio SicaandGitHub 2025-02-11 21:11:14 -03:00
• 2845f8797a failed test cases for rsqrt at 0 and similar ones (#9035) chenyuandGitHub 2025-02-11 17:50:16 -05:00
• 101652a55c hcq: thread fence (#8991) nimlgenandGitHub 2025-02-11 18:09:37 +03:00
• 45aae8a6bc hotfix: add External Benchmark Schedule to CI geohot 2025-02-11 22:06:17 +08:00
• 17fa6e7619 disk: better error desc when not opened (#9028) nimlgenandGitHub 2025-02-11 16:31:04 +03:00
• 166670a2f2 nv: fill grid/block sizes (#9025) nimlgenandGitHub 2025-02-11 16:30:30 +03:00
• c80603285e bring back some things from the fix_kernel_ops diff [pr] (#9027) qazalandGitHub 2025-02-11 14:20:31 +01:00
• 9209b85c91 add UOps.CAT (#9022) George HotzandGitHub 2025-02-11 19:50:37 +08:00
• a521260b7a dont reduce the ptr size, sz is base for unaligned [pr] (#9023) George HotzandGitHub 2025-02-11 19:50:23 +08:00
• d0d58a6771 add CUSTOM support to cstyle (#9020) George HotzandGitHub 2025-02-11 18:02:58 +08:00
• fb698920f1 revert scheduler change (#9019) George HotzandGitHub 2025-02-11 11:34:12 +08:00
• 16e9e4db37 make llvm opt the default (#9017) George HotzandGitHub 2025-02-11 10:08:45 +08:00
• bec4f59ce8 workaround f16 cast ambiguity (#8935) divinity76andGitHub 2025-02-11 02:38:56 +01:00
• b741a9aae7 update doc of Tensor.tolist (#9016) chenyuandGitHub 2025-02-10 16:51:23 -05:00
• 04e64765c4 Minor typo in ReadMe (#9015) JoelandGitHub 2025-02-10 21:30:20 +01:00
• 6c39aa4a6b adjust cuda ci test targets (#9014) chenyuandGitHub 2025-02-10 15:29:59 -05:00
• dfc9d6827f am_smi: print power state (#9013) nimlgenandGitHub 2025-02-10 23:07:39 +03:00
• bf0bcb2d5a cleanup ast rewriter [pr] (#9012) qazalandGitHub 2025-02-10 19:07:59 +01:00
• 586e48d696 a few more backward tests now pass (#9010) chenyuandGitHub 2025-02-10 12:46:21 -05:00
• f9898f7554 update gpuocelot commit (#9011) chenyuandGitHub 2025-02-10 12:18:44 -05:00
• c52cd2b437 kernel op cleanups + use ScheduleItem [pr] (#9009) qazalandGitHub 2025-02-10 17:54:30 +01:00
• 25fa5e4d5f enable backward tests in test_std_one_in_axis [pr] (#9007) 3 people 2025-02-10 10:44:05 -05:00
• d426f1ad6e don't open devices in lowering (#9008) qazalandGitHub 2025-02-10 15:28:51 +01:00
• cfd3db7862 construct the schedule sink 2 (#8925) qazalandGitHub 2025-02-10 15:23:02 +01:00
• 3e005ca0c2 am: resize bar0 to max supported (#9006) nimlgenandGitHub 2025-02-10 16:48:44 +03:00
• 07cb7e701c am: fix gfx usage at 100% (#9003) nimlgenandGitHub 2025-02-10 16:48:23 +03:00
• f91409f038 am: fix proclogs (#9004) nimlgenandGitHub 2025-02-10 16:38:58 +03:00
• cd77e51810 fix tensor realization bug in #8975 (#8984) qazalandGitHub 2025-02-10 13:51:30 +01:00
• b17ec42b56 remove const_arg (#9002) qazalandGitHub 2025-02-10 12:45:11 +01:00
• 0568720a68 delete revectorize (#9000) George HotzandGitHub 2025-02-10 18:32:35 +08:00
• fd9f9ec772 realized base tensors become RESHAPE(BUFFER) [pr] (#8994) qazalandGitHub 2025-02-10 10:17:54 +01:00
• 910ae260cd dsp float4 fold + revectorize [pr] (#8995) George HotzandGitHub 2025-02-10 12:14:32 +08:00
• e618efce22 COMMUTATIVE flipping is only for ints (#8996) George HotzandGitHub 2025-02-10 12:01:28 +08:00
• 2983285315 use HEX_REG_QEMU_INSN_CNT from qemu as a DSP timer [pr] (#8993) George HotzandGitHub 2025-02-10 11:07:35 +08:00