harvard-acc / smaug

SMAUG: Simulating Machine Learning Applications Using Gem5-Aladdin

https://harvard-acc.github.io/smaug_docs

harvard-acc/smaug Issues

Errors when bulid gem5-aladdin with --debug-aladdin flag
Updated a year ago
Can I simulate the accelerator whose clock frequency is larger than 1GHz
Updated a year ago3
Fail to replace accelerator spad with cache: `RuntimeError: Unknown array "inputs".`
Updated 2 years ago2
Failed to run Smaug example
Updated 2 years ago
Can we run multiple instances of NVDLA minerva model on smaug .
Updated 2 years ago2
Systolic array loops when feature size is the same of kernel size
Updated 2 years ago1
Simulation with ResNet fails
Updated 3 years ago7
Question about DMA setup time modeling
Updated 3 years ago2
Memory policy of AllCache causes SMAUG to crash in simulation
Closed 3 years ago1
How to port Rinforcement learning algorithm on smaug?
Updated 3 years ago2
Support for Depthwise Conv layer
Updated 3 years ago1
Add a padding operator
Updated 3 years ago
Improve documentation for SMV backend
Updated 3 years ago
Prepare tutorial showing how to simulate a network
Closed 4 years ago3
Prepare README for alpha open source release
Closed 4 years ago
Update Dockerfile and image with doc generator prereqs
Closed 4 years ago1
Clean up public APIs that may not need to be public or exist at all
Updated 4 years ago
Workspace is not a singleton
Updated 4 years ago1
Workspace::addTensor may leak memory if a Tensor name is reused
Updated 4 years ago
Clean up files from SMAUGv1
Updated 4 years ago
Simplify Python module naming for easier usage
Closed 4 years ago
Catch2 unit tests OOM when using the JUnit reporter
Updated 4 years ago
Flaky tests caused by floating point precision loss
Closed 4 years ago2
Move proto types into separate protos.
Updated 4 years ago
Prepare a docker image
Closed 4 years ago
Migrate python error handling from assertions to exceptions
Closed 4 years ago
Add a reference implementation with simple tiling for demonstration purposes
Updated 5 years ago