Skip to content

Instantly share code, notes, and snippets.

View cheshire's full-sized avatar

George Karpenkov cheshire

179 followers · 29 following
View GitHub Profile
@cheshire
cheshire / gist:f81fd036f07ea9267e3abb8c14c12aec
Created August 31, 2022 08:09
ROCm issues
tensorflow/core/profiler/backends/gpu/device_tracer_rocm.cc:1102:27: warning: ignoring returned value of type 'tensorflow::Status', declared with attribute nodiscard [-Wunused-result]
tensorflow/compiler/xla/stream_executor/rocm/rocm_dnn.cc:599:58: warning: ignoring returned value of type 'tensorflow::Status', declared with attribute nodiscard [-Wunused-result]
tensorflow/compiler/xla/stream_executor/rocm/rocm_dnn.cc:605:55: warning: ignoring returned value of type 'tensorflow::Status', declared with attribute nodiscard [-Wunused-result]
tensorflow/compiler/xla/stream_executor/rocm/rocm_dnn.cc:609:56: warning: ignoring returned value of type 'tensorflow::Status', declared with attribute nodiscard [-Wunused-result]
tensorflow/compiler/xla/stream_executor/rocm/rocm_dnn.cc:4134:25: warning: ignoring returned value of type 'hipError_t', declared with attribute nodiscard [-Wunused-result]
tensorflow/compiler/xla/stream_executor/rocm/rocm_dnn.cc:4163:40: warning: ignoring returned value of type 'hipError_t', declar
@cheshire
cheshire / fusion2_optimized.hlo
Created August 18, 2021 17:10
HloModule module
%scalar_add_computation (scalar_lhs: f32[], scalar_rhs: f32[]) -> f32[] {
%scalar_lhs = f32[] parameter(0)
%scalar_rhs = f32[] parameter(1)
ROOT %add.23852 = f32[] add(f32[] %scalar_lhs, f32[] %scalar_rhs)
}
%primitive_computation_max.289663 (parameter.289664: f32[], parameter.289665: f32[]) -> f32[] {
%parameter.289664 = f32[] parameter(0)
@cheshire
cheshire / slow.ptx
Created May 21, 2021 02:59
This file has been truncated, but you can view the full file.
//
// Generated by LLVM NVPTX Back-End
//
.version 6.0
.target sm_70
.address_size 64
// .globl fusion_2287
.shared .align 4 .b8 shared_cache_0[128];
@cheshire
cheshire / cudnn.log
Created May 21, 2021 00:38
I! CuDNN (v7604) function cudnnSetFilterNdDescriptor() called:
i! dataType: type=cudnnDataType_t; val=CUDNN_DATA_HALF (2);
i! format: type=cudnnTensorFormat_t; val=CUDNN_TENSOR_NHWC (1);
i! nbDims: type=int; val=4;
i! filterDimA: type=int; val=[256,256,3,3];
i! Time: 2021-05-20T12:38:10.745481 (0d+0h+0m+24s since start)
i! Process=1524391; Thread=1524637; GPU=NULL; Handle=NULL; StreamId=NULL.
I! CuDNN (v7604) function cudnnCreateConvolutionDescriptor() called:
@cheshire
cheshire / cudnn.log
Created May 20, 2021 19:40
I! CuDNN (v7604) function cudnnSetFilterNdDescriptor() called:
i! dataType: type=cudnnDataType_t; val=CUDNN_DATA_HALF (2);
i! format: type=cudnnTensorFormat_t; val=CUDNN_TENSOR_NHWC (1);
i! nbDims: type=int; val=4;
i! filterDimA: type=int; val=[256,256,3,3];
i! Time: 2021-05-20T12:38:10.745481 (0d+0h+0m+24s since start)
i! Process=1524391; Thread=1524637; GPU=NULL; Handle=NULL; StreamId=NULL.
I! CuDNN (v7604) function cudnnCreateConvolutionDescriptor() called:
@cheshire
cheshire / cublas.log
Created May 19, 2021 20:57
Loading:
Loading: 0 packages loaded
Analyzing: target //third_party/tensorflow/compiler/xla/client/lib:svd_test_gpu (0 packages loaded, 0 targets configured)
INFO: Analyzed target //third_party/tensorflow/compiler/xla/client/lib:svd_test_gpu (0 packages loaded, 0 targets configured).
INFO: Found 1 target...
[0 / 1] [Prepa] action 'BuildInfo build-info.txt'
Target //third_party/tensorflow/compiler/xla/client/lib:svd_test_gpu up-to-date:
blaze-bin/third_party/tensorflow/compiler/xla/client/lib/svd_test_gpu
INFO: Elapsed time: 0.365s, Critical Path: 0.09s, Remote (0.00% of the time): [queue: 0.00%, setup: 0.00%, process: 0.00%]
INFO: Build completed successfully, 1 total action
@cheshire
cheshire / patch.diff
Created April 28, 2021 03:41
diff --git a//tensorflow/compiler/jit/BUILD b//tensorflow/compiler/jit/BUILD
--- a//tensorflow/compiler/jit/BUILD
+++ b//tensorflow/compiler/jit/BUILD
@@ -151,6 +151,7 @@ cc_library(
":jit_compilation_passes",
":xla_device",
":xla_kernel_creator", # buildcleaner: keep
+ ":xla_device_no_jit_rewrite_registration",
"//third_party/absl/memory",
"//third_party/absl/strings",
@cheshire
cheshire / gist:151ef00fc078ef1e641661437e24774c
Created April 14, 2021 18:46
This file has been truncated, but you can view the full file.
HloModule Module
ENTRY computation {
tmp_0 = u8[32,224,224,3]{3,2,1,0} parameter(0)
tmp_1 = f32[] constant(1)
tmp_2 = f32[32,7,7,352]{3,2,1,0} broadcast(f32[] tmp_1), dimensions={}
tmp_3 = f32[32,7,7,352]{3,2,1,0} constant({ { /*i0=0*/ { /*i1=0*/ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
@cheshire
cheshire / gist:de6331415e4829722effcead954619da
Created April 14, 2021 18:45
hello
@cheshire
cheshire / IntegerTheoryFuzzer.java
Last active July 8, 2016 15:39
Fuzzer over the theory of integers
package org.sosy_lab.solver.test;
import org.sosy_lab.common.UniqueIdGenerator;
import org.sosy_lab.solver.api.FormulaManager;
import org.sosy_lab.solver.api.IntegerFormulaManager;
import org.sosy_lab.solver.api.NumeralFormula.IntegerFormula;
import java.util.Random;
import java.util.stream.IntStream;