To be fair, the arewefastyet benchmarks are intentionally very broad, and stress test many different parts of a JS engine. It may be the case that TurboFan has a very fast sweet spot for numeric code (which would explain why they'd want to use it for asm.js), but is slower for the more general cases.

I'd be interested to see how TurboFan performs on individual test cases; for example, Octane has several asm-specific tests, and based on their planning I wonder whether TurboFan is faster than Crankshaft for those individual tests even if it's slower for the rest of them.

You can see breakdowns for various subtests on awfy. Of possible interest are http://arewefastyet.com/#machine=28&view=breakdown&suite=asm... and http://arewefastyet.com/#machine=28&view=breakdown&suite=asm... as well as http://arewefastyet.com/#machine=28&view=breakdown&suite=oct...

Huh super interesting. It looks like currently TurboFan is competitive w/Crankshaft on the asm.js benchmarks, and much, much slower on everything else. I wonder if the V8 team expects TurboFan to someday beat Crankshaft on the asm benchmarks, given current perf differences.