MCPcopy Create free account

hub / github.com/HazyResearch/HipKittens / functions

Functions1,340 in github.com/HazyResearch/HipKittens

⨍ Functions 1,340◇ Types & classes 190↳ Endpoints 1
↓ 1 callersMethoddynamic_shared_memory
kernels/gemm/bf16fp32/micros/hint_based/kernel.cpp:36
↓ 1 callersMethoddynamic_shared_memory
kernels/gemm/bf16fp32/micros/192x256/kernel.cpp:36
↓ 1 callersMethoddynamic_shared_memory
kernels/gemm/bf16fp32/micros/192x256/kernelv2.cpp:36
↓ 1 callersMethoddynamic_shared_memory
kernels/gemm/bf16fp32/archive/256_256_64_32/256_256_64_32.cpp:35
↓ 1 callersMethoddynamic_shared_memory
kernels/gemm/bf16fp32/archive/256_256_64_16/256_256_64_16.cpp:35
↓ 1 callersMethoddynamic_shared_memory
kernels/gemm/bf16fp32/archive/micros/test_accum_layout/kernel.cpp:23
↓ 1 callersMethoddynamic_shared_memory
kernels/gemm/bf16fp32/archive/micros/64_16/kernel.cpp:24
↓ 1 callersMethoddynamic_shared_memory
kernels/gemm/bf16fp32/archive/micros/32_16/kernel.cpp:24
↓ 1 callersMethoddynamic_shared_memory
kernels/gemm/bf16fp32/archive/micros/256_64/kernel.cpp:34
↓ 1 callersMethoddynamic_shared_memory
kernels/gemm/bf16fp32/archive/micros/test_load_to_lds/kernel.cpp:23
↓ 1 callersMethoddynamic_shared_memory
kernels/gemm/bf16fp32/archive/micros/transpose_matmul/256_256_64_16.cpp:31
↓ 1 callersMethoddynamic_shared_memory
kernels/layernorm/kernel.cpp:81
↓ 1 callersMethoddynamic_shared_memory
kernels/softmax/kernel.cpp:27
↓ 1 callersMethoddynamic_shared_memory
kernels/attn/gqa_backwards/attn_bkwd_non_causal.cpp:44
↓ 1 callersMethoddynamic_shared_memory
kernels/attn/gqa_backwards/attn_fwd_non_causal.cpp:77
↓ 1 callersMethoddynamic_shared_memory
kernels/attn/gqa_backwards/archive/GQA_bkwd_asm.cpp:30
↓ 1 callersMethoddynamic_shared_memory
kernels/attn/gqa_backwards/archive/GQA_bkwd_4warps.cpp:40
↓ 1 callersMethoddynamic_shared_memory
kernels/attn/gqa_backwards/archive/GQA_bkwd_8warps.cpp:41
↓ 1 callersMethoddynamic_shared_memory
kernels/attn/gqa_backwards/archive/GQA_fwd.cpp:51
↓ 1 callersMethoddynamic_shared_memory
kernels/attn/gqa_causal_backwards/attn_bkwd_causal.cpp:45
↓ 1 callersMethoddynamic_shared_memory
kernels/attn/gqa_causal_backwards/attn_fwd_causal.cpp:157
↓ 1 callersMethoddynamic_shared_memory
kernels/attn/gqa/kernel.cpp:99
↓ 1 callersMethoddynamic_shared_memory
kernels/attn/gqa/kernel_d64.cpp:99
↓ 1 callersMethoddynamic_shared_memory
kernels/attn/gqa_causal/kernel.cpp:178
↓ 1 callersMethoddynamic_shared_memory
kernels/attn/gqa_causal/kernel_d64.cpp:178
↓ 1 callersMethoddynamic_shared_memory
kernels/rotary/kernel.cpp:41
↓ 1 callersMethoddynamic_shared_memory
training/llama/csrc/attn_bkwd_causal_HBN.cpp:45
↓ 1 callersMethoddynamic_shared_memory
training/llama/csrc/attn_fwd_causal.cpp:157
↓ 1 callersMethoddynamic_shared_memory
training/llama/csrc/attn_bkwd_causal_HNB.cpp:45
↓ 1 callersFunctionefficiency
Calculate efficiency in TFLOPS.
analysis/baselines/attn/attn_bwd_baselines.py:37
↓ 1 callersFunctionevaluate
Example of inference with trained model. It loads trained image classification model from checkpoint. Then it loads example image and predicts
training/llama/train/eval.py:47
↓ 1 callersFunctionextract_assembly_from_json
Parse the JSON file and extract only the assembly code. The JSON structure appears to be: {"code": [["instruction", ...], ...]}
docs/profiling/extract_asm_from_rocprof_json.py:5
↓ 1 callersFunctionfind_makefiles
Find all Makefiles and return (directory_path, relative_path) tuples. Excludes micros and archived kernels.
analysis/paper_experiments/compile_time/measure_compile_time.py:15
↓ 1 callersFunctionflops
(batch, seqlen, nheads, headdim, causal, mode="bwd")
analysis/baselines/attn/attn_bwd_baselines.py:32
↓ 1 callersFunctionflops
Calculate FLOPs for attention operation.
analysis/baselines/attn/attn_fwd_baselines.py:25
↓ 1 callersFunctionflops
Calculate FLOPs for LayerNorm operation.
analysis/layernorm/mi350x/test_python.py:38
↓ 1 callersFunctionflops
(batch, seqlen, nheads, headdim, causal, mode="bwd")
analysis/attn/bkwd/benchmark/test_python.py:46
↓ 1 callersFunctionflops
Calculate FLOPs for attention operation.
analysis/attn/fwd/benchmark/test_python.py:36
↓ 1 callersFunctionflops
Calculate FLOPs for LayerNorm operation.
analysis/rotary/mi350x/test_python.py:22
↓ 1 callersFunctionflops
Calculate FLOPs for LayerNorm operation.
kernels/layernorm/test_python.py:20
↓ 1 callersFunctionflops
(batch, seqlen, dim)
kernels/softmax/test_python.py:12
↓ 1 callersFunctionflops
(batch, seqlen, nheads, headdim, causal, mode="bwd")
kernels/attn/gqa_backwards/test_python.py:45
↓ 1 callersFunctionflops
(batch, seqlen, nheads, headdim, causal, mode="bwd")
kernels/attn/gqa_backwards/archive/test_python.py:45
↓ 1 callersFunctionflops
(batch, seqlen, nheads, headdim, causal, mode="bwd")
kernels/attn/gqa_causal_backwards/test_python.py:45
↓ 1 callersFunctionflops
Calculate FLOPs for attention operation.
kernels/attn/gqa/test_python.py:35
↓ 1 callersFunctionflops
Calculate FLOPs for attention operation.
kernels/attn/gqa_causal/test_python.py:35
↓ 1 callersFunctionflops
Calculate FLOPs for LayerNorm operation.
kernels/rotary/test_python.py:20
↓ 1 callersFunctionflops
(batch, seqlen, nheads, headdim, causal, mode="bwd")
training/llama/csrc/test.py:45
↓ 1 callersFunctiongenerate_inputs
()
analysis/baselines/attn/attn_bwd_baselines.py:150
↓ 1 callersFunctiongenerate_inputs
()
analysis/attn/bkwd/benchmark/test_python.py:154
↓ 1 callersFunctiongenerate_inputs
()
kernels/attn/gqa_backwards/test_python.py:153
↓ 1 callersFunctiongenerate_inputs
()
kernels/attn/gqa_backwards/archive/test_python.py:153
↓ 1 callersFunctiongenerate_inputs
()
kernels/attn/gqa_causal_backwards/test_python.py:153
↓ 1 callersFunctiongenerate_inputs
()
training/llama/csrc/test.py:153
↓ 1 callersFunctionget_autotune_configs
()
analysis/baselines/attn/triton_baseline_v02.py:425
↓ 1 callersFunctionget_cdna_autotune_configs
()
analysis/baselines/attn/triton_baseline_v02.py:390
↓ 1 callersFunctionget_cos_sin
(x, rotary_emb_base=10000, rotary_emb_dim=D, dtype=torch.bfloat16)
analysis/rotary/mi350x/test_python.py:38
↓ 1 callersFunctionget_cos_sin
(x, rotary_emb_base=10000, rotary_emb_dim=D, dtype=torch.bfloat16)
kernels/rotary/test_python.py:36
↓ 1 callersFunctionget_cuda_autotune_config
()
analysis/baselines/gemm/triton_gemm_v01.py:167
↓ 1 callersMethodget_device_view
Get device view for kernel usage
distributed-kernels/iris_py.cpp:104
↓ 1 callersFunctionget_hip_autotune_config
()
analysis/baselines/gemm/triton_gemm_v01.py:205
↓ 1 callersFunctionget_predictions
(model: nn.Module, data_loader: DataLoader)
training/bert/tasks.py:405
↓ 1 callersFunctionget_rdna_autotune_configs
()
analysis/baselines/attn/triton_baseline_v02.py:405
↓ 1 callersFunctionget_strides_from_layout
(q, k, v, o, metadata)
analysis/baselines/attn/triton_baseline_v02.py:1063
↓ 1 callersMethodgrid
distributed-kernels/bf16_gemm/kernel.cpp:83
↓ 1 callersMethodgrid
analysis/paper_experiments/phases/ds_write_b64/kernel.cpp:13
↓ 1 callersMethodgrid
analysis/paper_experiments/phases/ds_read_b128/kernel.cpp:13
↓ 1 callersMethodgrid
analysis/paper_experiments/phases/ds_read_b96/kernel.cpp:13
↓ 1 callersMethodgrid
analysis/paper_experiments/grid_micro/kernel_8192_w0.cpp:32
↓ 1 callersMethodgrid
analysis/paper_experiments/grid_micro/kernel_8192_w32.cpp:32
↓ 1 callersMethodgrid
analysis/paper_experiments/grid_micro/kernel_8192_w2.cpp:32
↓ 1 callersMethodgrid
analysis/paper_experiments/grid_micro/kernel_8192_w8.cpp:32
↓ 1 callersMethodgrid
analysis/paper_experiments/grid_micro/kernel_8192_w4.cpp:32
↓ 1 callersMethodgrid
analysis/layernorm/mi350x/kernel_4096.cpp:79
↓ 1 callersMethodgrid
analysis/layernorm/mi350x/kernel_1024.cpp:80
↓ 1 callersMethodgrid
analysis/layernorm/mi350x/kernel_2048.cpp:79
↓ 1 callersMethodgrid
analysis/layernorm/mi350x/kernel_16384.cpp:79
↓ 1 callersMethodgrid
analysis/layernorm/mi350x/kernel_8192.cpp:79
↓ 1 callersMethodgrid
analysis/attn/bkwd/benchmark/attn_bkwd_causal_HBN.cpp:43
↓ 1 callersMethodgrid
analysis/attn/bkwd/benchmark/attn_bkwd_non_causal_8_warps.cpp:52
↓ 1 callersMethodgrid
analysis/attn/bkwd/benchmark/attn_fwd_causal.cpp:155
↓ 1 callersMethodgrid
analysis/attn/bkwd/benchmark/attn_bkwd_causal_HNB.cpp:43
↓ 1 callersMethodgrid
analysis/attn/bkwd/benchmark/attn_fwd_non_causal.cpp:75
↓ 1 callersMethodgrid
analysis/attn/bkwd/benchmark/attn_bkwd_non_causal_4_warps.cpp:51
↓ 1 callersMethodgrid
analysis/attn/bkwd/benchmark/attn_bkwd_non_causal_HBN.cpp:42
↓ 1 callersMethodgrid
analysis/attn/bkwd/benchmark/attn_bkwd_non_causal_HNB.cpp:42
↓ 1 callersMethodgrid
analysis/attn/fwd/benchmark/attn_fwd_causal.cpp:176
↓ 1 callersMethodgrid
analysis/attn/fwd/benchmark/attn_fwd_causal_d64.cpp:176
↓ 1 callersMethodgrid
analysis/attn/fwd/benchmark/attn_fwd_non_causal.cpp:96
↓ 1 callersMethodgrid
analysis/attn/fwd/benchmark/attn_fwd_non_causal_d64.cpp:97
↓ 1 callersMethodgrid
analysis/rotary/mi350x/kernel_4096.cpp:39
↓ 1 callersMethodgrid
analysis/rotary/mi350x/kernel_1024.cpp:39
↓ 1 callersMethodgrid
analysis/rotary/mi350x/kernel_2048.cpp:39
↓ 1 callersMethodgrid
analysis/rotary/mi350x/kernel_16384.cpp:39
↓ 1 callersMethodgrid
analysis/rotary/mi350x/kernel_8192.cpp:39
↓ 1 callersMethodgrid
analysis/bf16_gemm/mi325x/kernel_4096.cpp:29
↓ 1 callersMethodgrid
analysis/bf16_gemm/mi325x/kernel_1024.cpp:29
↓ 1 callersMethodgrid
analysis/bf16_gemm/mi325x/kernel_2048.cpp:29
↓ 1 callersMethodgrid
analysis/bf16_gemm/mi325x/kernel_16384.cpp:29
↓ 1 callersMethodgrid
analysis/bf16_gemm/mi325x/kernel_8192.cpp:29
← previousnext →401–500 of 1,340, ranked by callers