madebyollin

def add_profiling_markers(model):
    """Monkey-patch profiling markers into an nn.Module.
    
    Args:
        model: an nn.Module
    
    Effect:
        all model.named_module() forward calls get wrapped in their
        own profiling scope, making traces easier to understand.
    """

b10011

mcarilli

# This isn't supposed to run as a bash script, i named it with ".sh" for syntax highlighting.

# https://developer.nvidia.com/nsight-systems
# https://docs.nvidia.com/nsight-systems/profiling/index.html

# My preferred nsys (command line executable used to create profiles) commands
#
# In your script, write
# torch.cuda.nvtx.range_push("region name")
# ...

banyudu

WSL2 Docker释放磁盘空间

Docker使用久了，会残留很多中间生成的镜像，占用磁盘空间。

要清理Docker镜像，本来是挺简单的，一条命令就搞定了：

gavinb

The following error message may be encountered when building CUDA code with CMake:

error LNK2019: unresolved external symbol __cudaRegisterLinkedBinary

It is caused by generated glue code not being linked in. The glue code is generated into files named with the project and a suffix:

myprojectname.device-link.obj

Ensure CUDA is enabled as a language for the project:

TengdaHan

Multi-node-training on slurm with PyTorch

What's this?

• A simple note for how to start multi-node-training on slurm scheduler with PyTorch.
• Useful especially when scheduler is too busy that you cannot get multiple GPUs allocated, or you need more than 4 GPUs for a single job.
• Requirement: Have to use PyTorch DistributedDataParallel(DDP) for this purpose.
• Warning: might need to re-factor your own code.
• Warning: might be secretly condemned by your colleagues because using too many GPUs.

schmmd

#!/usr/bin/env python

import os
import re
import subprocess
import socket
import sys

def pids_of_jid(jid):
    result = subprocess.run(["sstat", "-p", "--format=PID", "-j", jid, "--noheader"], stdout=subprocess.PIPE)

codewings

import json
import os
import plugin_api

#
def IsUsefulInput(viewtype):
    return viewtype == "CBV" or viewtype == "IBV" or viewtype == "VBV" or viewtype == "SRV"

def DumpBufferByDrawcall(folder, call, res, id, texture):
    if not os.path.exists(folder):

silvesthu

GPU Optimization for GameDev

Graphics Pipeline / GPU Architecture Overview

• 2011 - A trip through the Graphics Pipeline 2011
• 2013 - Performance Optimization Guidelines and the GPU Architecture behind them
• 2015 - Life of a triangle - NVIDIA's logical pipeline
• 2015 - Render Hell 2.0
• 2016 - How bad are small triangles on GPU and why?
• 2017 - GPU Performance for Game Artists
• 2019 - Understanding the anatomy of GPUs using Pokémon

tiagoshibata

commit 6fac225de0575b18550f9421196f90a2279c44af
Author: Tiago Koji Castro Shibata <[email protected]>
Date:   Fri Dec 28 23:13:14 2018 -0200

    Fix compilation with latest OpenCV
    
    C APIs were removed and must be replaced with C++ calls

diff --git a/Makefile b/Makefile
index 63e15e6..c148d4b 100644