Benchmark results for reading from in-memory cache
| Parent ↗ |
Summary
Used both AKTIVE and NETPBM as backends.
The results are minimally different between backends.
For AKTIVE the cache is definitely faster than the reader itself, by 2 to 5 times.
For NETPBM it is not as clear-cut. For most configurations the reader is actually somewhat faster, up to 2 times. The cache overtakes the reader only for the higher thread counts (>= 11).
Plots
AKTIVE vs NETPBM
Cache vs file, AKTIVE
Cache vs file, NETPBM
