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

• 229ada5fe5 Gpt2 benchmark with HALF and BEAM (#2636) chenyuandGitHub 2023-12-05 22:15:16 -05:00
• a73579919f mlx benchmark, a lil slower than tg geohot 2023-12-05 19:00:43 -08:00
• 7c427d738c don't apply padding on script call (#2585) Oleg RybalkoandGitHub 2023-12-06 03:34:10 +03:00
• 9d7ead84e1 hotfix: no need for model cache in examples/coder.py geohot 2023-12-05 16:27:36 -08:00
• be09cc87c1 Bitcast support / fast bf16 load (#2011) 3 people 2023-12-05 19:19:28 -05:00
• 232ed2af3f more test cleanups (#2631) George HotzandGitHub 2023-12-05 16:17:57 -08:00
• a63f48d3db gpt2 half for kvcache and output logits (#2630) chenyuandGitHub 2023-12-05 16:54:56 -05:00
• 0be5d16950 only 62 gflops (#2629) George HotzandGitHub 2023-12-05 13:28:24 -08:00
• 6d58c19736 binaryops xor (#2627) wozeparrotandGitHub 2023-12-05 16:21:42 -05:00
• c53e854687 cast image doesn't work on nvidia (#2626) George HotzandGitHub 2023-12-05 12:48:19 -08:00
• 8c67eb1c92 GPT bugfixes (#2624) George HotzandGitHub 2023-12-05 11:42:28 -08:00
• 8903a40541 update the onnx test so cuda local run passes (#2623) chenyuandGitHub 2023-12-05 14:04:17 -05:00
• ec594cf03c hotfix: tasteful ctrl-c in parallel beam geohot 2023-12-05 18:20:10 +00:00
• 35b5e95097 parallel beam search (#2610) George HotzandGitHub 2023-12-05 10:09:45 -08:00
• 9996f1adf9 no document prs (#2622) chenyuandGitHub 2023-12-05 13:05:36 -05:00
• dd8b4632a4 regression test for reshape fix #2616 (#2620) chenyuandGitHub 2023-12-05 11:46:33 -05:00
• c257a0dd99 minor reshape cleanups (#2619) chenyuandGitHub 2023-12-05 11:23:17 -05:00
• a6b68e8e40 fix for false merge (#2616) Amrit SahuandGitHub 2023-12-05 21:17:18 +05:30
• fc00da538d helper functions for test_indexing.py (#2615) geohotstanandGitHub 2023-12-05 15:00:41 +08:00
• 7322ab8dfd onnx tests with different dtypes (#2612) chenyuandGitHub 2023-12-05 00:04:08 -05:00
• f12bcccb87 [ready] refactor getitem round 2 :D (#2568) geohotstanandGitHub 2023-12-05 11:36:32 +08:00
• 6ba6349c97 JIT=0 llama.py should not jit (#2609) chenyuandGitHub 2023-12-04 20:21:07 -05:00
• 41d696145d hotfix: forking works okay in HIP now geohot 2023-12-04 21:59:18 +00:00
• 09b6e254a3 hip compile speed (#2606) George HotzandGitHub 2023-12-04 13:47:40 -08:00
• 19a0a839db fix used resources in metal graph (#2604) nimlgenandGitHub 2023-12-05 00:45:51 +03:00
• fde44aed76 update hip_matmul with new abstraction (#2605) Yixiang GaoandGitHub 2023-12-04 15:37:10 -06:00
• 5540f6e966 hotfix: make_half4 George Hotz 2023-12-04 09:58:34 -08:00
• e8d6a6ef2e view.reshape without symbolic (#2218) 3 people 2023-12-04 23:16:53 +05:30
• 664475f247 vals is an argument (#2599) George HotzandGitHub 2023-12-03 21:50:43 -08:00
• fcd0b2ee6c fix multigpu on tinybox (#2595) George HotzandGitHub 2023-12-03 16:48:07 -08:00
• 61c0113928 test external_multi_gpu.py (and works in CUDA) George Hotz 2023-12-03 15:57:13 -08:00
• bbeba8ec85 use default dict for external_model_benchmark (#2592) George HotzandGitHub 2023-12-03 15:25:43 -08:00
• 550817389a enable test_sample for all backend (#2593) chenyuandGitHub 2023-12-03 17:20:27 -05:00
• a58736fdf1 print DEBUG=2 stats after stats update (#2590) chenyuandGitHub 2023-12-03 17:13:37 -05:00
• bc012f26b9 hotfix, disable model inference benchmark on NVIDIA geohot 2023-12-03 13:52:41 -08:00
• 4380ccb169 Non fp32 math (#2264) 3 people 2023-12-03 16:45:49 -05:00
• 1ac958a058 update pytest marks and CI test filters (#2587) chenyuandGitHub 2023-12-03 15:20:44 -05:00
• 88a5c368d4 fix metal graph with var_vals (#2583) nimlgenandGitHub 2023-12-03 20:24:36 +03:00
• f180cac8f0 wgsl renderer cleanup: use the same const render, reuse cast render logic (#2579) qazalandGitHub 2023-12-03 12:24:00 -05:00
• ab2d4d8d29 Fix cl import in the copy_speed test and cifar example (#2586) qazalandGitHub 2023-12-03 12:22:07 -05:00
• 3226b3d96b enable the jit random test (#2580) chenyuandGitHub 2023-12-02 20:25:23 -05:00
• 09c9794f3f clean external_test_opt.py (#2578) chenyuandGitHub 2023-12-02 19:51:08 -05:00
• 171543fc8d cleanups to save lines and files (#2577) George HotzandGitHub 2023-12-02 16:29:56 -08:00
• a9a76639c8 that's not needed (#2574) George HotzandGitHub 2023-12-02 16:01:29 -08:00
• 875c34bfc4 minor lazy tweak before rewrite (#2573) chenyuandGitHub 2023-12-02 18:23:33 -05:00
• fa1d4dd14b implement MAX in other dtypes (#2572) qazalandGitHub 2023-12-02 18:21:59 -05:00
• 065495e0c9 save a few lines in ops_gpu (#2564) 3 people 2023-12-03 02:05:22 +03:00
• 5e87083783 Whisper + LLAMA + VITS (#2332) Oleg RybalkoandGitHub 2023-12-03 02:03:46 +03:00
• 47cec4caf3 int operations shouldn't have a fast math flag (#2571) qazalandGitHub 2023-12-02 17:53:36 -05:00
• d6b404ac11 No dtype alloc (#2570) George HotzandGitHub 2023-12-02 13:29:40 -08:00
• c8774713c5 lazy cleanup (#2567) chenyuandGitHub 2023-12-02 13:21:43 -05:00
• 5068e99d18 refactor to remove extra kernel params (#2563) George HotzandGitHub 2023-12-02 00:32:25 -08:00
• 27481b9206 Switch ops_gpu -> gpuctypes (#2532) George HotzandGitHub 2023-12-01 22:30:21 -08:00
• 99ee2ec37a Refactor code_for_op to accept a dtype (#2555) 3 people 2023-12-02 01:05:28 -05:00
• 82fd932921 Lower schedule 2 (#2561) George HotzandGitHub 2023-12-01 20:25:49 -08:00
• 217cda81ba hotfix: no metalgraph if there's weird ops geohot 2023-12-01 19:32:55 -08:00
• 6733425095 lower schedule (#2559) George HotzandGitHub 2023-12-01 19:17:46 -08:00
• 077567f62d Remove as_buffer for TORCH (#2554) sirhcmandGitHub 2023-12-01 18:51:38 -08:00
• 05a5357dd9 fix handcode_resnet50_opt.py (#2558) chenyuandGitHub 2023-12-01 20:51:21 -05:00
• 86fbd413f3 update test_real_world configs (#2557) chenyuandGitHub 2023-12-01 20:03:52 -05:00
• 00523d5656 New fix accessing elements created by padding (#2529) 3 people 2023-12-02 02:08:10 +02:00
• bfdce1f0e7 hotfix: make openpilot test deterministic geohot 2023-12-01 15:37:23 -08:00
• 9c306be282 better name for fast path geohot 2023-12-01 15:32:22 -08:00
• 67f4e03724 rewrite 0 size loadop into a CONST (#2556) chenyuandGitHub 2023-12-01 18:29:06 -05:00
• 4447188051 gate METAL_FAST_LOAD geohot 2023-12-01 15:28:15 -08:00
• e9426f4fe4 simpler get_contraction (#2552) chenyuandGitHub 2023-12-01 18:02:52 -05:00
• eb595588bb device.py cleanups for -1 line geohot 2023-12-01 14:59:33 -08:00
• f9b1de598f hotfix: metal fastpath on sonoma geohot 2023-12-01 14:55:34 -08:00
• f5de21e753 fast path for copy (#2548) George HotzandGitHub 2023-12-01 11:34:47 -08:00
• 28183c7438 feat: reword (#2549) wozeparrotandGitHub 2023-12-01 13:56:18 -05:00
• 4c984bba7e bump version to 0.8.0, clean CI, remove requests (#2545) George HotzandGitHub 2023-12-01 10:42:50 -08:00
• ff47be3a01 ruff check whitespaces (#2547) nimlgenandGitHub 2023-12-01 21:42:20 +03:00
• 8fd8399437 remove flake8 (#2544) George HotzandGitHub 2023-12-01 09:48:41 -08:00
• d8175a4380 simple fix (#2543) George HotzandGitHub 2023-12-01 09:42:15 -08:00
• 04483f8187 refactor llvm consts (#2537) qazalandGitHub 2023-12-01 12:39:40 -05:00
• badc97f824 hip & cuda to gpuctypes (#2539) nimlgenandGitHub 2023-12-01 20:25:27 +03:00
• 0fb4ff30c8 share duplicate renders with cstyle (#2538) qazalandGitHub 2023-12-01 11:10:36 -05:00
• 7fec966b5e bye bye NOOP (#2534) chenyuandGitHub 2023-12-01 02:10:35 -05:00
• fa549d198d Remove type: ignore comments (#2533) Joe DonovanandGitHub 2023-12-01 01:15:55 -05:00
• 12fa846122 zero copy (#2531) George HotzandGitHub 2023-11-30 18:38:41 -08:00
• 5394a05b9d Fix: Get item from ndarray before casting to int (#2525) Matthias KronbergandGitHub 2023-12-01 03:34:31 +01:00
• 2c363b5f0b new style device (#2530) George HotzandGitHub 2023-11-30 17:07:16 -08:00
• e56511b59a more type annotation for tensor and lazy (#2528) chenyuandGitHub 2023-11-30 17:50:22 -05:00
• ddeec24fa8 Cleanup & fix llama.py (#2524) Davi SilvaandGitHub 2023-12-01 04:00:17 +07:00
• 7d26452305 call ruff with --preview (#2522) chenyuandGitHub 2023-11-30 13:59:00 -05:00
• 5db0cdfbd3 support list of ints (or other Tensorable) in tensor indices (#2520) chenyuandGitHub 2023-11-30 12:46:33 -05:00
• bd941a0df1 first version of test_indexing (#2515) chenyuandGitHub 2023-11-30 00:03:59 -05:00
• d210f6a786 minor device.py cleanups (#2510) chenyuandGitHub 2023-11-29 18:16:25 -05:00
• 370cfbb957 Cleanup vectorized hip renders (#2497) qazalandGitHub 2023-11-29 17:02:12 -05:00
• abfc99187d cleanup realize (#2505) George HotzandGitHub 2023-11-29 11:38:38 -08:00
• 3dedeaae74 rebalance tests (#2504) George HotzandGitHub 2023-11-29 11:18:22 -08:00
• 065aff747e make webgpu test reliable (#2502) George HotzandGitHub 2023-11-29 10:02:24 -08:00
• 6707f2588e use copyin (#2500) George HotzandGitHub 2023-11-29 09:34:00 -08:00
• 947711a532 split metal and webgpu tests (#2501) George HotzandGitHub 2023-11-29 09:32:09 -08:00
• 3eb3c74675 metal ci tests everything (#2499) chenyuandGitHub 2023-11-29 12:04:37 -05:00
• 889acefe85 Support weird loads in Image (#2498) George HotzandGitHub 2023-11-29 08:30:46 -08:00
• e333672675 realize cleanup (#2496) George HotzandGitHub 2023-11-28 21:08:39 -08:00
• 5629fc368c Use Buffer.STORE at the end of ASTs (#2494) George HotzandGitHub 2023-11-28 20:11:37 -08:00
• cf0c9096a9 Removing METAL Skips as CI works (#2488) LiamandGitHub 2023-11-29 04:46:59 +01:00
• 5588922884 Update cuda_matmul.py (#2495) JakeandGitHub 2023-11-28 22:46:01 -05:00