[Nouveau] "enable ctxprog xfer only when we need it to save power" introduces big performance regression

[Martin Peres]

Le 17/01/2012 21:55, Lucas Stach a écrit :
> Isn't it possible that the performance regression seen with xfer 
> disabled by default is caused by slow memory clock speed? Martin, you 
> saw only a 1% performance drop on your 8600 which is running with full 
> speed by default. Marcins nv92 is likely running at a much lower clock 
> speed. It would be nice to know if the perf regression is still as big 
> when running with full speed. We could stop caring about this when the 
> perf regression is not observable in the highest perf level.
Well, actually, the performance hit should be higher when running at a 
highest memory clock since we context switch more often (there is no 
logic here, just a personal experience).

Anyway, I tried again last week end at every possible speed and I 
haven't been able to spot any performance hit *at all*.