ram-coffers
https://github.com/scottcjn/ram-coffers
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
not yet supported1 Subscribers
Add a CodeTriage badge to ram-coffers
Help out
- Issues
- docs: the benchmark machine had two memory configurations, not one
- Vulkan/HIP FP8 E4M3FN decoders map NaN codes 0x7F/0xFF to ±480 (CPU/Python return NaN)
- README states five different RAM capacities for the POWER8 S824 (320/512/544/557/768 GB)
- fix: resolution for issue #721 in benchmarks/BLOG_RAM_COFFERS_VS_DEEPSEEK.md
- display_topology_text shows non-contiguous NUMA CPU sets as a range ([0,2,4,6] → "0-6")
- fix(topology): support memory-only NUMA nodes with empty cpulist in sysfs parser (#704)
- feat(benchmark): add machine-readable topology preflight scripts/topology_preflight.py (#705)
- gen9-cluster: DeepSeek-V4.1-Flash profile (CSA2/CED, Engram, DSpark)
- bug: GGUF mmap parser leaves tensor metadata empty
- fix: parse GGUF tensor metadata for NUMA placement
- Docs
- not yet supported