-
Notifications
You must be signed in to change notification settings - Fork 186
NvTensorRtRtx dependency on CUDA device name removed #1485
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
NvTensorRtRtx dependency on CUDA device name removed #1485
Conversation
NvTensorRtRtx dependency on using ORT CUDA EP removed.
- Fixing formatting for CICD failure - Adding final inheritance for CUDA and NvTensorRtRtx EP interfaces impl structs. - Removing Graph Capture for NvTensorRtRtx EP for now.
Fixing compilation error for linux builds
…ub.com/BLSharda/onnxruntime-genai into bhaviksharda/nvep_removes_cuda_ep_deps
@kunal-vaishnavi and @baijumeswani , can you help review and merge this change? |
- Removing cuda graph knob for now until we test it for functional and perf aspect.
@baijumeswani, can you please help with review and merge this change? |
Could you elaborate or share the relevant onnxruntime pull-request so we can track the reason for this chance? |
In PR 24656 CUDA dependency is removed (USE_CUDA) for NvTensorRtRtx EP so we need to switch device name from "cuda" to "NvTensorRtRtx". |
GenAI EP NvTensorRtRtx dependency on using CUDA device removed. This is to align with latest changes in ORT and NV EP.
Also, still requires onnxruntime-genai-cuda.dll but device name is NvTensorRtRtx instead of CUDA. And the memory type is still "Cuda" for ORT API.