ERROR: failed to solve: nvidia/cuda:8.0-devel: docker.io/nvidia/cuda:8.0-devel: not found

[nyck33]

nyck33@lenovo-gtx1650:/mnt/d/LLVM/llvm-project/llvm/utils/docker/nvidia-cuda$ docker build -t llvm-cuda-env .
[+] Building 1.7s (5/5) FINISHED        docker:default
 => [internal] load build definition from Docker  0.0s
 => => transferring dockerfile: 1.43kB            0.0s
 => [internal] load .dockerignore                 0.0s
 => => transferring context: 2B                   0.0s
 => ERROR [internal] load metadata for docker.io  1.7s
 => CANCELED [internal] load metadata for docker  1.7s
 => [auth] nvidia/cuda:pull token for registry-1  0.0s
------
 > [internal] load metadata for docker.io/nvidia/cuda:8.0-devel:
------
Dockerfile:31
--------------------
  29 |     
  30 |     # Stage 2. Produce a minimal release image with build results.
  31 | >>> FROM nvidia/cuda:8.0-devel
  32 |     LABEL maintainer "LLVM Developers"
  33 |     # Copy clang installation into this container.
--------------------
ERROR: failed to solve: nvidia/cuda:8.0-devel: docker.io/nvidia/cuda:8.0-devel: not found
nyck33@lenovo-gtx1650:/mnt/d/LLVM/llvm-project/llvm/utils/docker/nvidia-cuda$ docker build -t llvm-cuda-env .
[+] Building 0.0s (2/2) FINISHED        docker:default
 => [internal] load build definition from Docker  0.0s
 => => transferring dockerfile: 1.49kB            0.0s
 => [internal] load .dockerignore                 0.0s
 => => transferring context: 2B                   0.0s
ERROR: failed to solve: circular dependency detected on stage: builder
nyck33@lenovo-gtx1650:/mnt/d/LLVM/llvm-project/llvm/utils/docker/nvidia-cuda$ docker build -t llvm-cuda-env .
[+] Building 1.8s (8/11)                docker:default
 => [internal] load .dockerignore                 0.0s
 => => transferring context: 2B                   0.0s
 => [internal] load build definition from Docker  0.0s
 => => transferring dockerfile: 1.48kB            0.0s
 => [internal] load metadata for docker.io/nvidi  1.6s
 => CANCELED [builder 1/6] FROM docker.io/nvidia  0.0s
 => => resolve docker.io/nvidia/cuda:11.5.2-deve  0.0s
 => => sha256:d14c2904ba34c18ae4 2.63kB / 2.63kB  0.0s
 => => sha256:3df12c4530b84128 16.45kB / 16.45kB  0.0s
 => => sha256:ff53b082160dddbb203677 743B / 743B  0.0s
 => [internal] load build context                 0.0s
 => => transferring context: 2B                   0.0s
 => CACHED [builder 2/6] RUN apt-get update &&    0.0s
 => ERROR [builder 3/6] ADD checksums /tmp/check  0.0s
 => ERROR [builder 4/6] ADD scripts /tmp/scripts  0.0s
------
 > [builder 3/6] ADD checksums /tmp/checksums:
------
------
 > [builder 4/6] ADD scripts /tmp/scripts:
------
Dockerfile:19
--------------------
  17 |         rm -rf /var/lib/apt/lists/*
  18 |     
  19 | >>> ADD checksums /tmp/checksums
  20 |     ADD scripts /tmp/scripts
  21 |     
--------------------
ERROR: failed to solve: failed to compute cache key: failed to calculate checksum of ref 882ec67d-e2c8-4c89-ac67-6fa13f2809ff::cn5mxcrbrk4nisva3ory4vaak: "/checksums": not found
nyck33@lenovo-gtx1650:/mnt/d/LLVM/llvm-project/llvm/utils/docker/nvidia-cuda$ docker build -t llvm-cuda-env .
[+] Building 0.8s (3/3) FINISHED        docker:default
 => [internal] load .dockerignore                 0.0s
 => => transferring context: 2B                   0.0s
 => [internal] load build definition from Docker  0.0s
 => => transferring dockerfile: 1.48kB            0.0s
 => ERROR [internal] load metadata for docker.io  0.8s
------
 > [internal] load metadata for docker.io/nvidia/cuda:8.0-devel:
------
Dockerfile:31
--------------------
  29 |     
  30 |     # Stage 2. Produce a minimal release image with build results.
  31 | >>> FROM nvidia/cuda:8.0-devel
  32 |     #FROM nvidia/cuda:11.5.2-devel-ubuntu20.04
  33 |     
--------------------
ERROR: failed to solve: nvidia/cuda:8.0-devel: docker.io/nvidia/cuda:8.0-devel: not found
nyck33@lenovo-gtx1650:/mnt/d/LLVM/llvm-project/llvm/utils/docker/nvidia-cuda$ 

and for the debian

nyck33@lenovo-gtx1650:/mnt/d/LLVM/llvm-project/llvm/utils/docker/debian10$ docker build -t llvm-cuda-env .
[+] Building 1.7s (10/14)               docker:default
 => [internal] load .dockerignore                 0.0s
 => => transferring context: 2B                   0.0s
 => [internal] load build definition from Docker  0.0s
 => => transferring dockerfile: 2.17kB            0.0s
 => [internal] load metadata for launcher.gcr.io  1.6s
 => [internal] load build context                 0.0s
 => => transferring context: 2B                   0.0s
 => CANCELED [builder 1/8] FROM launcher.gcr.io/  0.0s
 => => resolve launcher.gcr.io/google/debian10:l  0.0s
 => => sha256:c583011c5370f62f33116f 529B / 529B  0.0s
 => => sha256:dcfdcd76b18436200c 1.46kB / 1.46kB  0.0s
 => CACHED [builder 2/8] RUN grep deb /etc/apt/s  0.0s
 => CACHED [builder 3/8] RUN apt-get update &&    0.0s
 => CACHED [builder 4/8] RUN wget "https://githu  0.0s
 => ERROR [builder 5/8] ADD checksums /tmp/check  0.0s
 => ERROR [builder 6/8] ADD scripts /tmp/scripts  0.0s
------
 > [builder 5/8] ADD checksums /tmp/checksums:
------
------
 > [builder 6/8] ADD scripts /tmp/scripts:
------
Dockerfile:28
--------------------
  26 |         rm ninja-linux.zip
  27 |     
  28 | >>> ADD checksums /tmp/checksums
  29 |     ADD scripts /tmp/scripts
  30 |     
--------------------
ERROR: failed to solve: failed to compute cache key: failed to calculate checksum of ref 882ec67d-e2c8-4c89-ac67-6fa13f2809ff::7pxxp8zpsjzkvpup3pnnptukv: "/checksums": not found
nyck33@lenovo-gtx1650:/mnt/d/LLVM/llvm-project/llvm/utils/docker/debian10$ 

I'm on Windows 11 WSL2. From https://llvm.org/docs/Docker.html

Is there a workaround?

[Artem-B]

That docker container is, AFAICT, obsolete. CUDA-8 used in the Dockerfile is ancient.

I'm not exactly sure what the purpose of that docker image is. CUDA SDK is not needed at all in order to build clang.

What are you trying to do with it?

If you do want to make the container buildable, replacing nvidia/cuda:8.0-devel with an up to date image from https://hub.docker.com/r/nvidia/cuda/tags should help, though I suspect other steps in the build file may need updating, too.

[nyck33]

I wanted to get an image with MLIR but thought I needed to build an image of LLVM first and then install MLIR on top of LLVM? (LLVM is a dependency for MLIR)? Then run the toy tutorial for MLIR so I learn the basics and then start learning to optimize CUDA kernels in neural networks.
I learned about NVIDIA Cutlass only recently but it seems like the workflow would be something like (for custom neural networks):
Write CUDA kernels using Cutlass, optimize config parameters for Cutlass for these kernels and specific matmuls in the neural network, then use LLVM or MLIR to dig into the IR to try to optimize even further. Am I totally off here? I've only taken an intro to software testing and analysis course using LLVM.
I have a GTX 1650 which is not great but good enough for testing my conceptual understanding so I was trying to build the nvidia-cuda image and tried using a newer tag from the nvidia cuda Docker Hub account but the build fails. I have either Cuda 11.5 on my WSL2 or 12.2 on native Ubuntu that I can use.
I was opting for Docker because I heard installing/building LLVM can be quite difficult and actually in that course we were given a VM image but the LLVM version was quite old so thought I needed a newer version for MLIR. I apologize for the wall of words.