Benchmarking Mercury As The "Fastest Firefox Fork" With AVX, AES, LTO + PGO
With the modern JetStream 2.0 benchmark again I wasn't finding any measurable benefits with Mercury 115 compared to upstream Firefox 115, but did continue to be pleased with the direction the performance is moving with Firefox 118.
With CanvasMark there appears to be a strange -- and severe -- regression with Firefox 118 nightly.
With the WebAssembly benchmarks there wasn't much change in performance across the benchmarked browsers when it came to the raw performance.
But PSPDFKit's WASM benchmark for example was showing lower CPU power consumption and utilization when running with the Firefox 118 alpha state.