Same GPU, Same Impossible Job: One Engine Crashed, the Other Slowed Down 8%
Measuring stock HuggingFace against vLLM on one 8 GB laptop GPU — including two results that contradicted what I expected
Sep 5, 20269 min read1
Command Palette
Search for a command to run...
Latest articles
5.1× From Overlapping, 11.6× From Processes, 0× From Fewer Bytes
Four ways to read 10,000 files out of object storage, and which one did nothing
Aug 1, 202614 min read
I Replaced a Flink Cluster With One Python Process
Every dependency is a liability you pay whether or not you collect its benefit.
Jun 8, 20269 min read7The GIL Fingerprint: Why My Kafka Producer Stalled at 14k events/sec
One CPython fact explained three mysteries: an unmovable plateau, five failed fixes, and the thread count that finally worked.
May 16, 20269 min read9