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

autogen-comgr-2

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

ci-deps

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

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

explicit-dev

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

gitea-test-ci

gl_dims

gldims

glm_flash

grad_outerworld

grok_x86_novec

group_div_rules

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-chat-template

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

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

onnx-cpu-count

outer_range

outerworld_work

pad_invalid

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

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-lp

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

view_after_const

view_in_codegen

viz_cbroadcast

warp_fun

wgpu-f16-packed

working_postopt

x86_moves

x86_numel

#1

#2

#3

#4

#5

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

• 80116be9a5 for loop to generate hip math functions for different floats (#3967) chenyuandGitHub 2024-03-27 23:24:29 -04:00
• 03d129baa8 inputs -> membufs (#3964) qazalandGitHub 2024-03-28 02:34:39 +02:00
• 16a1d43f6f llama: prevent device initialization outside of __main__ (#3966) Francis LamandGitHub 2024-03-27 16:19:38 -07:00
• 7c5729a3bd wmma: refactor to remove wmma_func and create TC funcs as needed (#3945) Francis LamandGitHub 2024-03-27 13:43:09 -07:00
• 88b24df40a touchup remove float() in cstyle render_const for float64 (#3962) chenyuandGitHub 2024-03-27 16:08:28 -04:00
• 27af37f2ad misc: remove unused env vars (#3963) qazalandGitHub 2024-03-27 22:08:15 +02:00
• 60639cccac hotfix: RuntimeError for assign geohot 2024-03-27 11:18:48 -07:00
• 9fb573d73c DAG cycle asserts (#3955) qazalandGitHub 2024-03-27 20:09:59 +02:00
• bd3a7d068c correct device for validation test in model benchmark CI (#3960) geohotstanandGitHub 2024-03-28 01:40:06 +08:00
• eec2b00edc change kernel name if it's multioutput (#3958) George HotzandGitHub 2024-03-27 08:42:57 -07:00
• d1c957a471 copy back to clang (#3951) George HotzandGitHub 2024-03-27 08:13:01 -07:00
• 332c82893a Remove redundant check on device (#3957) P4ssengerandGitHub 2024-03-27 11:54:33 -03:00
• 6c7df1445b enforce UOps.CONST arg has python type based on dtype (#3952) chenyuandGitHub 2024-03-27 01:41:38 -04:00
• 91f3326c0b hotfix: increase recursion limit geohot 2024-03-26 21:26:54 -07:00
• 68ca4d4276 split to schedule.py (#3949) George HotzandGitHub 2024-03-26 21:02:46 -07:00
• da07f31fd4 hotfix: remove bf16 test entirely geohot 2024-03-26 20:50:27 -07:00
• 0d5845fb5b hotfix: jit is flaky on mac geohot 2024-03-26 20:44:05 -07:00
• 150ea2eb76 create engine folder and move code (#3948) George HotzandGitHub 2024-03-26 20:38:03 -07:00
• 629cbc5587 only abstractions 2 (#3947) George HotzandGitHub 2024-03-26 20:02:18 -07:00
• 77589bc7a5 rename Scalar to ConstType and cast_scalar to as_const (#3946) chenyuandGitHub 2024-03-26 22:39:58 -04:00
• d6d902afe9 wtf (#3944) uuuvnandGitHub 2024-03-27 02:49:28 +02:00
• 5530b0cbed fuzz_linearizer: reduce debug verbosity and make easier for CI usage (#3942) Francis LamandGitHub 2024-03-26 13:25:24 -07:00
• 8df6587c41 hotfix 97.3 for beautiful_mnist (#3941) chenyuandGitHub 2024-03-26 15:02:53 -04:00
• b1e3817e18 correctly handle Tensor.rand whwn default_float = bf16 (#3940) chenyuandGitHub 2024-03-26 14:56:16 -04:00
• f6ff76be21 check only upcast int amount in upcasted_axis (#3938) chenyuandGitHub 2024-03-26 12:54:57 -04:00
• e2d6f76723 _alloc and _free with options (#3934) nimlgenandGitHub 2024-03-26 19:11:41 +03:00
• 739f47eb0f check on cuEventSynchronize (#3933) nimlgenandGitHub 2024-03-26 16:14:38 +03:00
• 778d17fbd3 intel matmul (#3830) George HotzandGitHub 2024-03-25 22:37:20 -07:00
• ef537672bf bf16 support in metal (#3929) chenyuandGitHub 2024-03-25 23:17:36 -04:00
• 72d617a37d opencl on OSX does not support fp16 extension (#3931) chenyuandGitHub 2024-03-25 19:50:17 -04:00
• cb6e7b57a6 examples: Fix parameter bandwidth accounting for quantized LLama (#3930) Arseny KapoulkineandGitHub 2024-03-25 15:41:05 -07:00
• 4ecd5789ab #include <tgmath.h> in ops_clang (#3927) chenyuandGitHub 2024-03-25 17:48:57 -04:00
• 514c43201d Fix issues with pointer provenance in load/store through ALU (#3916) Arseny KapoulkineandGitHub 2024-03-25 14:41:05 -07:00
• d651835ef5 verify beautiful_mnist.py eval acc and put into benchmark ci (#3926) chenyuandGitHub 2024-03-25 16:47:49 -04:00
• dc508022a9 clean up clang src header (#3925) chenyuandGitHub 2024-03-25 15:18:35 -04:00
• 2080325e8d output_buffer isn't used anymore (#3919) uuuvnandGitHub 2024-03-25 15:03:56 +02:00
• f2a9ea4ea9 lru allocator for copyin host buffers (#3918) nimlgenandGitHub 2024-03-25 15:57:18 +03:00
• e0e234bf94 hotfix, str compare version for cuda geohot 2024-03-24 20:35:24 -07:00
• 715850aef9 Fix sm89 PTX=1 compilation (#3915) Arseny KapoulkineandGitHub 2024-03-24 20:32:29 -07:00
• 83f39a8ceb env var to change default float (#3902) chenyuandGitHub 2024-03-24 20:33:57 -04:00
• 03899a74bb increase atol on reset train geohot 2024-03-24 15:17:31 -07:00
• d8fafca13a assign regression (#3907) 3 people 2024-03-25 00:12:31 +02:00
• 2d0bfdf01c ptx cleanup (#3893) Szymon OżógandGitHub 2024-03-24 22:54:45 +01:00
• 2e39f57594 move lines around in ops_python wmma (#3911) chenyuandGitHub 2024-03-24 17:14:26 -04:00
• e27129a798 Fix linearizer failure 26 test (#3906) 3 people 2024-03-24 16:34:13 -04:00
• 10673d1447 tiny search cleanup (#3910) chenyuandGitHub 2024-03-24 14:20:55 -04:00
• 9a9cac58f9 add lars to nn (#3750) 3 people 2024-03-24 11:43:12 -04:00
• 8c8b57fd5f cleanup ops python (#3908) chenyuandGitHub 2024-03-24 11:36:31 -04:00
• 2c69888654 include negative float in test_dtype (#3884) chenyuandGitHub 2024-03-24 02:39:15 -04:00
• e22d78b3d2 training cifar with BF16 on CUDA (#3905) chenyuandGitHub 2024-03-24 01:37:47 -04:00
• 0145366323 wmma: fix the AMD TC threads to split the first 16 threads (#3904) Francis LamandGitHub 2024-03-23 18:17:42 -07:00
• 7c3632fd1e add --minimal flag to nvrtc (#3899) sekstiniandGitHub 2024-03-24 00:38:31 +01:00
• a2b2597fc2 replace dtype.name str with render_dtype (#3903) chenyuandGitHub 2024-03-23 19:25:48 -04:00
• 24d004a89b hotfix check ckpts before writing achieved model (#3901) chenyuandGitHub 2024-03-23 17:16:38 -04:00
• 4d566f12b1 touchup einsum (#3900) chenyuandGitHub 2024-03-23 16:46:39 -04:00
• 556dcfb8f2 Fix the result permutation in einsum (#3895) 3 people 2024-03-23 15:48:19 -04:00
• 4e18dd78d3 faster program start in llvm (#3897) nimlgenandGitHub 2024-03-23 15:20:15 +03:00
• 46a3501cec nv ioctl sniffer (#3892) George HotzandGitHub 2024-03-23 00:29:30 -07:00
• 18e0cef14d cheap less lines in ptx (#3890) chenyuandGitHub 2024-03-23 01:12:31 -04:00
• f0c4e06ffd fix cuda sync (#3888) George HotzandGitHub 2024-03-22 19:02:30 -07:00
• 2d3ce53348 touchup test_dtype.test_gradient_dtype (#3887) chenyuandGitHub 2024-03-22 20:56:45 -04:00
• fc11808a79 initialize Tensor grad same type as self (#3613) 3 people 2024-03-22 17:33:18 -07:00
• 8db7a6bbcc debug: add optional detailed BEAM_LOG logging (#3883) Francis LamandGitHub 2024-03-22 16:23:31 -07:00
• f7f67e0cc5 simple fix llama shard with quantize (#3882) chenyuandGitHub 2024-03-22 18:15:37 -04:00
• ee502c8055 fixup to_movement_ops and add back to CI (#3881) chenyuandGitHub 2024-03-22 18:14:49 -04:00
• 16e31f7f0d init multidevice cuda graph (#3858) 3 people 2024-03-22 23:49:48 +03:00
• 0c197b9cf3 hotfix: hip bfloat formatting geohot 2024-03-22 11:52:05 -07:00
• 54dc48aa47 fix assign (#3878) George HotzandGitHub 2024-03-22 11:48:48 -07:00
• 5587594a00 fuzz_linearizer: add --ast and --file params to read kernels (#3877) Francis LamandGitHub 2024-03-22 11:27:40 -07:00
• c5467e5bd6 diverse test value in test_dtype DATA based on dtype (#3864) chenyuandGitHub 2024-03-22 14:22:06 -04:00
• 86ee36e697 preschedule all (#3875) George HotzandGitHub 2024-03-22 11:20:06 -07:00
• d8c3f1894a Use UOpGraph in test (#3876) Szymon OżógandGitHub 2024-03-22 19:12:38 +01:00
• 1c51d586ea replace raise Exception with specific errors (#3874) chenyuandGitHub 2024-03-22 12:32:21 -04:00
• 8ef5490ec8 cuda tranfer + async copyin (#3873) nimlgenandGitHub 2024-03-22 19:01:37 +03:00
• 624bc89910 PTX - implement float 4, ptr arithmetics and other speed improvements (#3775) 3 people 2024-03-22 16:54:02 +01:00
• f4055439dc don't include hip common (#3851) George HotzandGitHub 2024-03-22 08:50:50 -07:00
• 4a27ce6ec9 tiny version of amd_hip_bfloat16 (#3868) qazalandGitHub 2024-03-22 17:37:30 +02:00
• 82ce60e172 use JIT_BATCH_SIZE=4 for GPT2 3090 benchmark (#3870) chenyuandGitHub 2024-03-22 00:40:06 -04:00
• fe6ceff15f proposal: multioutput JIT spec (#3856) qazalandGitHub 2024-03-22 06:28:30 +02:00
• a26090d404 search: change to use "spawn" and limit the number of tasks per child (#3862) Francis LamandGitHub 2024-03-21 21:23:36 -07:00
• dca69df197 hot fix use DEBUG >= 3 for allreduce message (#3869) chenyuandGitHub 2024-03-21 23:40:44 -04:00
• 6729f20aab Ring allreduce try 2 (#3852) 5 people 2024-03-22 01:17:51 +02:00
• 3c0478bfab fuzz_linearizer: add additional DEBUG info for comparison errors (#3866) Francis LamandGitHub 2024-03-21 15:58:10 -07:00
• bc482729d0 lower hlb_cifar acc to 93.3 (#3865) chenyuandGitHub 2024-03-21 17:58:53 -04:00
• e50b7abe4f diversed buf inputs based on dtype in fuzz_linearizer (#3863) chenyuandGitHub 2024-03-21 16:23:11 -04:00
• c40f78499f reuse fuzz_linearizer.compare_linearizer in test_linearizer_failures (#3861) chenyuandGitHub 2024-03-21 14:23:37 -04:00
• 30fa03243e reuse fuzz_linearizer.compare_linearizer in test_linearizer_failures (#3861) chenyuandGitHub 2024-03-21 14:12:27 -04:00
• 33dd99acf4 remove helper_add_store from test_linearizer_failures (#3860) chenyuandGitHub 2024-03-21 12:53:31 -04:00
• 6bf0b82267 alloc new output in fuzz_linearizer between baseline and real one (#3859) chenyuandGitHub 2024-03-21 11:36:05 -04:00
• b78352b423 do not create structs every call in CUDAProgram (#3855) nimlgenandGitHub 2024-03-21 17:51:40 +03:00
• e5745c1a0d fix nan on multigpus cuda (#3854) nimlgenandGitHub 2024-03-21 15:21:55 +03:00
• 4e0819e40b fixing the benchmark not printing in handcode resnet50 opt example (#3850) Anurag LamsalandGitHub 2024-03-21 10:40:31 +05:45
• 85691c8e20 fix hsa sync issue (#3847) nimlgenandGitHub 2024-03-21 04:00:30 +03:00
• f271cd682b user _resolve_dim in argmax (#3846) chenyuandGitHub 2024-03-20 20:17:30 -04:00
• 5c4cf62d2c fix View.pad arg type (#3845) chenyuandGitHub 2024-03-20 19:36:02 -04:00
• 6d5dec2fef log optimized kernels and a script to compare with non-optimized ones (#3829) Francis LamandGitHub 2024-03-20 16:22:08 -07:00
• 9d1d08fbb0 show llama bandwith with timing (#3844) chenyuandGitHub 2024-03-20 17:19:15 -04:00
• 7ff47e45a1 cifar TARGET_EVAL_ACC_PCT=93.5 (#3843) chenyuandGitHub 2024-03-20 16:56:51 -04:00
• 92c5067439 conceptual small refactor (#3842) qazalandGitHub 2024-03-20 22:46:14 +02:00
• 519336cfea factor out partial in SumNode div int (#3841) chenyuandGitHub 2024-03-20 16:34:33 -04:00