malfet

import Text.Printf
import Control.Exception
import System.CPUTime

-- time function implementation borrowed from 
-- http://www.haskell.org/haskellwiki/Timing_computations 
time :: IO t -> IO t
time a = do
    start <- getCPUTime
    v <- a

malfet

#!/usr/bin/env python3
from elftools.elf.elffile import ELFFile
import struct
import sys


# From https://stackoverflow.com/questions/1094841/reusable-library-to-get-human-readable-version-of-file-size
def sizeof_fmt(num, suffix='B'):
    for unit in ['','Ki','Mi','Gi','Ti','Pi','Ei','Zi']:
        if abs(num) < 1024.0:

malfet

import torch

def hflip(img: torch.Tensor) -> torch.Tensor: return img.flip(-1)
print(torch.jit.script(hflip)(torch.rand(3, 8, 8)))

malfet

# as -o hello.o hello.S ; cc -o hello hello.o -nostdlib
  .text
  .globl _start
  .type  _start, @function
_start:
  movl $1,    %eax  # sys_write(
  movl $1,    %edi  # fd = stdout,
  movl $.LC0, %esi  # buf = LC0,
  movl $12,   %edx     # 12);
  syscall

malfet

// gcc -c -Os -mavx512f -masm=intel
#include <immintrin.h>

float foo(float* con) {
    __mmask16 msk = 0x00ff;
    __m512 a = _mm512_maskz_loadu_ps(msk, con);
    __m512 b = _mm512_set1_ps(con[1]);
    __m512 c = _mm512_mul_ps(a,b);
    return ((float *)&c)[0];
}

malfet

// nvcc -o hello hello.cu; ./hello
#include <stdio.h>

__global__ void kernel() {
  printf("Hello World of CUDA\n");
}

int main() {
  kernel<<<1,1>>>();
  return cudaDeviceSynchronize();

malfet

 	[Inline Frame] torch_cuda.dll!std::_Default_allocator_traits<std::allocator<std::_Tree_node<unsigned int,void *>>>::deallocate(std::allocator<std::_Tree_node<unsigned int,void *>> &) Line 689	C++
 	[Inline Frame] torch_cuda.dll!std::_Tree_node<unsigned int,void *>::_Freenode0(std::allocator<std::_Tree_node<unsigned int,void *>> &) Line 373	C++
 	[Inline Frame] torch_cuda.dll!std::_Tree_val<std::_Tree_simple_types<unsigned int>>::_Erase_head(std::allocator<std::_Tree_node<unsigned int,void *>> &) Line 753	C++
 	[Inline Frame] torch_cuda.dll!std::_Tree<std::_Tset_traits<unsigned int,std::less<unsigned int>,std::allocator<unsigned int>,0>>::{dtor}() Line 1191	C++
>	torch_cuda.dll!torch::jit::fuser::newForReduction(torch::jit::fuser::TensorView * tv, const std::vector<unsigned int,std::allocator<unsigned int>> & axes) Line 438	C++
 	torch_cuda.dll!torch::jit::fuser::reductionOp(torch::jit::fuser::BinaryOpType reduction_op_type, const std::vector<int,std::allocator<int>> & axes, torch::jit::fuser::Val * init, to

malfet

#include <arm_neon.h>
#include <math.h>
#include <stdio.h>


void run_neon_reciproc(float data_in[4], float data_out[4]) {
	float32x4_t input = vld1q_f32(data_in);
	float32x4_t out = vrecpeq_f32(input);
	//out = vmulq_f32(vrecpsq_f32(input, out), out);
	//out = vmulq_f32(vrecpsq_f32(input, out), out);

malfet

#!/usr/bin/env python3
import boto3
import os
import bz2
import json
import subprocess
from datetime import datetime


def get_git_commit_history(path, branch="master"):

malfet

#!/usr/bin/env python3

from datetime import datetime
from typing import Any, Dict, List, Optional, Union
from urllib.request import urlopen, Request
import json
import enum
import os