Our paper on tensor core modeling in GPGPU-Sim was accepted (conditionally) to appear in ISPASS 2019, the 2019 IEEE International Symposium on Performance Analysis of Systems and Software! Here is a link to the arXiv version of the paper.

We will also present two posters at ISPASS summarizing our recent work enabling cuDNN and PyTorch to run on top of GPGPU-Sim, and updating the memory system model in GPGPU-Sim to better match recent NVIDIA GPUs.

The code for the above three works are available in the dev branch of GPGPU-Sim on github!