McGrath for me.
Once he came into his own, starting around the West Indies tour of 1995, only in 5 Test series of the 33 Test series that he played in for the rest of his career (excluding one-off Tests), did his series bowling average go above 30. And only in 8 of those 33 series, did his series average go above 25.
And all this with just 1 series against then minnows (Bangladesh of that time).
Astonishing consistency.
That's a bit like a batsman having series averages of 50+ in 25 of the 33 Test series he plays in.
Also, the impact of his bowling to his team was very evident at the time. He would often remove the best batsman of the oppostion.
I felt when he and Shane Warne were playing together, more often than not, his bowling had slightly more impact than Warne's.
I think he was badly missed in Ashes 2005 as well as against India in 1998 and 2003-04.
Sometimes McGrath's impact is hard to capture in pure stats. When Australia toured South Africa in 1996-97 for a Test series, it was then billed as heavyweight championship finals of Test cricket. Quite a few experts & analysts expected South Africa to win and those who favored Australia talked only about Shane Warne.
Right at the start of the series, on the first session of the first day of the first Test match, McGrath pretty much settled the issue.
In an unbeliveable spell of great fast bowling, he removed the top 4 South African batsmen (including the key batsmen Cullinan and Kallis) for pretty much nothing.
Psychological impact of McGrath's first spell had to be seen live to be believed. Stats simply don't capture it. It set the tone for the rest of the series which Aus won easily.
In McGrath's presence, Aus Test win-loss record was 84-20 (W/L ratio of 4.2), but during the same time - meaning while McGrath was still an active Test player but skipped the tour, or was absent from the playing XI - Aus Test win-loss record was 18-8 (W/L ratio dropped down to 2.25).