5 people 6729f20aab Ring allreduce try 2 (#3852) ...

* Ring allreduce v3

* Configurable size, number of gpus and jit in benchmark

* ScheduleBarrier v0

* GB/s that make sense

* ScheduleBarrier v0.1

* Fallback on 2 GPUs

* ScheduleBarrier v0.2

* ScheduleBarrier v0.3

* ScheduleBarrier v0.3.1

* ScheduleBarrier v0.3.2

* Replace ScheduleBarrier with automatic optimization

* unused import

* fix comment

* typing

* better fallback

* python 3.8

* RING=2 and use ContextVar

* DEBUG >= 2 and change name

* linter

* type

---------

Co-authored-by: George Hotz <[email protected]>
Co-authored-by: chenyu <[email protected]>
Co-authored-by: nimlgen <[email protected]>

2024-03-21 19:17:51 -04:00

..

mlperf_resnet

PolynomialDecayWithWarmup + tests (#3649)

2024-03-07 18:53:36 -05:00

external_benchmark_hip_compile.py

hip compile speed (#2606)

2023-12-04 13:47:40 -08:00

external_benchmark_load_stable_diffusion.py

gate METAL_FAST_LOAD

2023-12-01 15:28:40 -08:00

external_benchmark_multitensor_allreduce.py

Ring allreduce try 2 (#3852)

2024-03-21 19:17:51 -04:00

external_cl_half_max.py

use default dict for external_model_benchmark (#2592)

2023-12-03 15:25:43 -08:00

external_hip_compiler_bug.py

support pickling tensors and others (#3787)

2024-03-17 18:29:14 -07:00

external_jit_failure.py

fix jit realize issue (#3258)

2024-01-26 18:27:35 -08:00

external_llama_eval.py

ruff checks the max line length is 150 (#2734)

2023-12-12 17:34:47 -08:00

external_model_benchmark.py

move graph.py and jit.py into features (#3376)

2024-02-12 17:34:34 +01:00

external_multi_gpu.py

s/lazydata.realized/lazydata.base.realized/g (#2914)

2023-12-22 14:45:13 -05:00

external_osx_profiling.py

move dtypes to dtype.py (#2964)

2024-01-01 14:58:48 -08:00

external_test_embedding.py

make embedding and GPT-2 fast (#1631)

2023-08-22 15:14:38 -07:00

external_test_example.py

remove cpu and torch backends (#3399)

2024-02-15 16:55:39 +01:00

external_test_hip_compile.py

multioutput ScheduleItem (#3699)

2024-03-13 08:59:38 -07:00

external_test_hsa_driver.py

fix hsa sync issue (#3847)

2024-03-21 04:00:30 +03:00

external_test_image.py

fixes (#1893)

2023-09-22 07:20:27 +08:00

external_test_jit_on_models.py

move graph.py and jit.py into features (#3376)

2024-02-12 17:34:34 +01:00

external_test_mnist_data_select.py

add quick external data select test

2024-03-02 05:38:32 -08:00

external_test_onnx_backend.py

move is_dtype_supported to test.helpers (#3762)

2024-03-15 14:33:26 -04:00

external_test_opt.py

move graph.py and jit.py into features (#3376)

2024-02-12 17:34:34 +01:00

external_test_optim.py

PolynomialDecayWithWarmup + tests (#3649)

2024-03-07 18:53:36 -05:00

external_test_speed_llama.py

cleanup tests Device[Device.DEFAULT] is always Compiled (#3645)

2024-03-07 11:15:42 -05:00

external_test_uops_graphing.py

move create schedule and delete old API (#3377)

2024-02-12 18:10:45 +01:00

external_test_whisper_librispeech.py

whisper: support audio >30s (#2378)

2023-11-21 14:37:51 -08:00

external_test_yolo.py

move to new cached fetch (#2493)

2023-11-28 17:36:55 -08:00

external_test_yolov8.py

ruff checks the max line length is 150 (#2734)

2023-12-12 17:34:47 -08:00

fuzz_linearizer.py

fuzz_linearizer: add additional DEBUG info for comparison errors (#3866)

2024-03-21 18:58:10 -04:00

fuzz_shapetracker_math.py

touchup canonicalize empty mask (#3308)

2024-02-03 21:05:10 -05:00

fuzz_shapetracker.py

minor improvements (#2845)

2023-12-18 22:09:08 -08:00

fuzz_symbolic.py

Variable.num -> NumNode (#2354)

2023-11-18 15:45:52 -05:00

graph_batchnorm.py

with Tensor.train() (#1935)

2023-09-28 18:02:31 -07:00

speed_beam_v_hcopt.py

script to benchmark beam v hcopt (#3737)

2024-03-14 15:04:03 -04:00

speed_compare_cuda_ptx.py

support pickling tensors and others (#3787)

2024-03-17 18:29:14 -07:00

verify_kernel.py

log optimized kernels and a script to compare with non-optimized ones (#3829)

2024-03-20 19:22:08 -04:00