iddm

Error

ERR_NVGPUCTRPERM: Permission issue with Performance Counters

Solutions for this issue

NVIDIA has a page that addresses the issue. Unfortunately, the instructions didn't solve the problem for me. However, I found a solution from rameshgunjal in the NVIDIA forums.

Option 1: Running the program as admin

You might have noticed that sudo nv-nsight-cu results in an error. Instead, you have to provide the path: sudo /usr/local/cuda/bin/nv-nsight-cu. The same applies for other applications as well: sudo /usr/local/cuda/bin/nsight-sys.

iddm

Missing documentation found empirically for the aforementioned variables.

We can traverse only one TLAS at at time. While traversing it via traceRayEXT, we can hit just one BLAS at at a time. Hence, all the *ID references here are relative to the BLAS we hit (except gl_InstanceID):

1. gl_GeometryIndexID - Geometry index within the BLAS this ray has hit.
2. gl_InstanceID - BLAS index within the TLAS.
3. gl_PrimitiveID - Triangle (or other primitive) index within the BLAS.

iddm

// Credits to "Robk" https://stackoverflow.com/users/13382603/robk
// From https://stackoverflow.com/questions/4991707/how-to-find-my-current-compilers-standard-like-if-it-is-c90-etc
static void _printversion() {
    if (!__STDC__ && !__STDC_VERSION__) printf("The C compiler does not comply with the C89 or later standard!\nIt likely complies with the 1978 K&R C standard (informally known as C78).\n");
    else if (__STDC_VERSION__ >= 201710L) printf("The C compiler complies with the C17 standard.\n");
    else if (__STDC_VERSION__ >= 201112L) printf("The C compiler complies with the C11 standard.\n");
    else if (__STDC_VERSION__ >= 199901L) printf("The C compiler complies with the C99 standard.\n");
    else if (__STDC_VERSION__ >= 199409L) printf("The C compiler complies with the amended C90 standard (also known as C95).\n");
    else if (__STDC__) printf("The C compiler complies with the ANSI C89 / ISO C90 standard.\n");
   

iddm

for line in yourString:gmatch("([^\n]*)\n?") do
  -- do something
end

iddm

# First get your monitor's name from ddc.
❯ ddcutil detect
Display 1
   I2C bus:             /dev/i2c-4
   EDID synopsis:
      Mfg id:           VSC
      Model:            XG270
      Serial number:    VXR194200328
      Manufacture year: 2019
      EDID version:     1.4

iddm

openssl rand -base64 32

iddm

docker ps | awk {' print $1 '} | tail -n+2 > tmp.txt; for line in $(cat tmp.txt); do docker kill $line; done; rm tmp.txt

iddm

#!/bin/sh

# This will read 4GB+ data from /dev/zero and write to a test file in current directory.
# If you need to test the speed of your SSD drive, go to any partition with it with a filesystem
# with write support, make sure you have rights to do this and then run the command below (or just this script).
# As a sort of customisation, the block size ideally should be identical to the block size of your device:
# https://unix.stackexchange.com/questions/52215/determine-the-size-of-a-block-device
# However, the `64k` value worked absolutely fine for my setup.
dd if=/dev/zero of=test_$$ bs=64k count=64k conv=fdatasync && rm -f test_$$

iddm

export VERSION=26.2

yum install gcc make ncurses-devel giflib-devel libjpeg-devel libtiff-devel gnutls-devel
curl -O http://ftp.gnu.org/gnu/emacs/emacs-$VERSION.tar.xz > emacs.tar.xz
tar xJf emacs.tar.xz
cd emacs-$VERSION
./configure --without-x --without-selinux
make && sudo make install

iddm

1. Delete all containers

    $ docker ps -q -a | xargs docker rm
   

-q prints only the container IDs -a prints all containers

Notice that it uses xargs to issue a remove container command for each container ID

2. Delete all untagged images