Closed GoogleCodeExporter closed 9 years ago
I talked to Dave Mandelin about register usage way back in the early days of
the JM port and actually we already have far more registers in play than
methodjit will ever use. It might be something to consider for ion though.
Original comment by classi...@floodgap.com
on 17 May 2012 at 1:17
And what about using r2 instead of r11 as r11 shouldn't normally be used?
Original comment by Tobias.N...@gmail.com
on 17 May 2012 at 1:23
It's a thought, though so far there have been no problems in practice, so I'm
unwilling to change it now without a good reason (but again, we have to start
from scratch in ion, so).
Original comment by classi...@floodgap.com
on 17 May 2012 at 1:26
We're going to use r2 after all, but for the "trampoline remnant" (see
discussion in issue 167).
Original comment by classi...@floodgap.com
on 18 Jul 2012 at 4:10
I attempted to work on this, and the JS works, but there is absolutely no
difference at least on the G5. This could be because of the G5's
superscalarity, but I am thinking this doesn't really pay off.
Original comment by classi...@floodgap.com
on 18 Sep 2012 at 9:12
We're going to look at this for IonMonkey, but I don't think there's much
payoff here after all.
Original comment by classi...@floodgap.com
on 21 Dec 2012 at 4:45
Original issue reported on code.google.com by
Tobias.N...@gmail.com
on 17 May 2012 at 8:59