Most of the results (despite how the graphs are distorted) are actually really close.
Sunspider differences were actually big with Edge 108ms, Chrome 190ms, Firefox 254ms
Octane had Edge winning with 33489, Chrome second with 31839, Firefox last with 30307. That means Edge was about 10% faster than Firefox, with Chrome splitting the difference. Not huge.
Kraken had Chrome at 938ms, Edge 1160ms, Firefox 1224ms, so around 25% slower for Firefox - enough to be noticeable, depending on what you're doing.
Jetstream had Edge winning with 219, Chrome with 184, Firefox trailing badly at 154, so again a fairly substantial gap. Looking a little at the details, all had around the same throughput and whatever's being measured on latency was the driver for the differences.
For the Oort WebGL graphics, Firefox was best with 10000, Chrome second at 9940, Edge third at 9920. Those are not differences that excite me.
Peacekeeper (no longer maintained) had Firefox first at 4655, Chrome second at 4325 and Edge trailing badly at 3091 - not quite as lopsided as the Sunspider results, but quite the reversal.
For WebXPRT (HTML5+JS), Edge won with 448, Firefox at 402, Chrome at 396. That's 10% faster for Edge, but margin of error for Chrome and Firefox.
And finally for the HTML5 test Chrome had 499, Firefox 462, Edge 460 - again around a 10% difference between slowest and fastest.