TensorRT-LLM icon indicating copy to clipboard operation
TensorRT-LLM copied to clipboard

Same GPU build, same files, but got the error: The engine plan file is generated on an incompatible device, expecting compute 9.0 got compute 8.9, please rebuild.

Open JoJoLev opened this issue 9 months ago • 1 comments

System Info

Running an an H200, engine was built on an H200 as outlined by the config.json file. When deploying to H200, I received this error: The engine plan file is generated on an incompatible device, expecting compute 9.0 got compute 8.9, please rebuild.

Who can help?

No response

Information

  • [ ] The official example scripts
  • [ ] My own modified scripts

Tasks

  • [ ] An officially supported task in the examples folder (such as GLUE/SQuAD, ...)
  • [ ] My own task or dataset (give details below)

Reproduction

Install TensorRT-LLM and follow quick guide setup compile rank.engine files deploy with triton

Expected behavior

Expected behavior was for the engine file to run since it was built on the same exact GPU architecture

actual behavior

I receive this error: The engine plan file is generated on an incompatible device, expecting compute 9.0 got compute 8.9, please rebuild.

additional notes

Not sure the exact issue around this, worked fine on the engine build step so it's odd to see a mistmatch.

JoJoLev avatar Mar 24 '25 13:03 JoJoLev

@JoJoLev

Hi, can you share the concrete reproducing steps to reproduce the issue?

Thanks June

juney-nvidia avatar Mar 26 '25 06:03 juney-nvidia