mratsim

## main.nim - uses a loop and a split
import strutils

let inFile = open("input.txt", fmRead)
let outFile = open("output.txt", fmWrite)

var ln: TaintedString = ""
var parts: seq[string]

while inFile.readLine(ln):

mratsim

## From https://forum.nim-lang.org/t/2856

import macros

macro chaining(code: untyped): untyped =
  const chainIdent = "chain"
  const combineIdent = "combine"
  proc inspect(depth: int, n: NimNode): NimNode =
    case(n.kind)
    of nnkIdent, nnkStrLit:

mratsim

proc filter[T](it: (iterator : T), f: proc(x: T): bool): (iterator: T) =
  return iterator(): T =
    while (let x = it(); not finished(it)):
      if f(x):
        yield x

mratsim

git rev-list --objects --all \
| git cat-file --batch-check='%(objecttype) %(objectname) %(objectsize) %(rest)' \
| awk '/^blob/ {print substr($0,6)}' \
| sort --numeric-sort --key=2 \
| cut --complement --characters=13-40 \
| numfmt --field=2 --to=iec-i --suffix=B --padding=7 --round=nearest

mratsim

#!/usr/bin/env bash

# Colours picked from https://robinpowered.com/blog/best-practice-system-for-organizing-and-tagging-github-issues/

###
# Label definitions
###
declare -A LABELS

# Platform

mratsim

import nimcuda/[cuda_runtime_api, driver_types, nimcuda]
import sequtils, future

type GpuArray[T: SomeReal] = object
  data: ref[ptr T]
  len: int

{.compile: "./square.cu".}
proc cuda_square(bpg, tpb: cint, y: ptr cfloat, x: ptr cfloat) {.importc, header:"../square.cuh".}
#../square.cuh is a workaround because header is not copied to nimcache

mratsim

__global__
void  GPUCode( int* input, int* output, int length)
{
               int idx = __umul24( blockDim.x, blockIdx.x) + threadIdx.x;
 
               if ( idx < length )
                {
                      output[ idx ] = input[ idx ] + 2 * input[ idx + 1 ];
               }
}

mratsim

__global__
void saxpy(int n, float a, float *x, float *y)
{
    for (int i = blockIdx.x * blockDim.x + threadIdx.x; 
         i < n; 
         i += blockDim.x * gridDim.x) 
      {
          y[i] = a * x[i] + y[i];
      }
}

mratsim

## Even easier, without the cu / cuh
## Note VScode properly syntax highlight the emit part, yeah!

import nimcuda/[cuda_runtime_api, driver_types, nimcuda]
import sequtils, future

type GpuArray[T: SomeReal] = object
  data: ref[ptr T]
  len: int

mratsim

// Note: this doesn't work (yet)

// Example program
#include <iostream>
#include <string>

int cuda_getIndexOfElementID(
    const int rank,
    const int * __restrict__ shape,
    const int * __restrict__ strides,