https://github.com/breandan/lavm
https://github.com/breandan/lavm
Last synced: about 2 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/breandan/lavm
- Owner: breandan
- Created: 2026-04-13T21:09:27.000Z (4 months ago)
- Default Branch: main
- Last Pushed: 2026-06-10T03:53:11.000Z (2 months ago)
- Last Synced: 2026-06-10T05:21:53.071Z (2 months ago)
- Language: TeX
- Size: 664 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# lavm
A massively parallel virtual machine based on the [RASP architecture](https://en.wikipedia.org/wiki/Random-access_stored-program_machine).
### Experiments
To reproduce the JVM experiments, run the following command (tested on M4 Max):
```agsl
./gradlew run
```
This will run a parallel streaming search for heapless programs and print out the longest-running busy beavers.
To reproduce the CUDA experiments, first generate the VMs like so:
```
./gradlew writeVMs
```
Then either (1) run the following commands on hardware with Compute Capability 7.5 or higher, e.g., for an H100:
```agsl
cd cuda
nvcc -O3 -std=c++17 -lineinfo -Xptxas=-v -gencode arch=compute_90,code=sm_90 rasp_cuda_benchmark_soa.cu -o rasp_cuda_benchmark_soa
./rasp_cuda_benchmark_soa rasp_vms.bin
```
or (2) deploy remotely. We provide a Modal harness [here](cuda/rasp_cuda_modal_cu_soa.py).
### Paper
* [https://arxiv.org/pdf/2604.12902](https://arxiv.org/pdf/2604.12902)