buildcache/doc/benchmarks.md

Benchmarks

LLVM Compilation

For this benchmark, LLVM (2019-10-22, Git commit: 2c4ca6832fa6b306ee6a7010bfb80a3f2596f824) was built from source as follows:

mkdir build && cd build
cmake -G Ninja -DCMAKE_BUILD_TYPE=Release ..
time ninja

The system used for the benchmark was:

• CPU: AMD Ryzen 7 1800x (8-core x86-64, 3.6 GHz).
• Disk: 1TB NVMe (960 EVO)
• RAM: 32GiB DDR4 @ 3000 MHz
• OS: Linux (Ubuntu 18.04)
• Compiler: GCC 7.5.0
• BuildCache: 0.19.0

No cache

Time	Speed
12m17.8s	1.0x

Local cache, no compression

	Time	Speed
Cold cache	12m56.9s	0.95x
Warm cache	1m3.5s	11.6x

Cache size: 196.8 MiB

Local cache, LZ4 compression

	Time	Speed
Cold cache	12m56.3s	0.95x
Warm cache	1m5.8s	11.2x

Cache size: 75.3 MiB

Local cache, ZSTD compression

	Time	Speed
Cold cache	12m56.6s	0.95x
Warm cache	1m2.6s	11.8x

Cache size: 49.7 MiB