[Vm-dev] unit code for new faster bitblt
tim at rowledge.org
Sat Jun 29 01:44:33 UTC 2013
On 26-06-2013, at 9:35 PM, tim Rowledge <tim at rowledge.org> wrote:
> On a related issue, I finally made enough sense of cmake to spot a way to get the relevant define, err, defined, without having to alter the sqPlatformSpecific.h, which is a nice improvement. I'm still inclined to think that some manner must exist to only try to compile the related C files if you want them, rather than relying on the library linking to dump unwanted code.
The pile'o'filesfor this is available for any heroic person to try out on an intel machine; I'd be fascinated to hear if it makes any difference when you already have a huge cache and wide memory bus to cover up mild inefficiencies.
A normal ../../cmake/configure should build a perfectly ordinary interpreter vm. ../cmake/configure --enableFastBlt should make the fast blt, and (I hope) should appropriately skip the ARM specific stuff. I'm pretty sure that some extra IF clause in the plugins/BitBltPlugin/config.cmake script would be useful for skipping the BitBltDispatch & BitBltGeneric files when un-needed but damned if I can work out how right now.
tim Rowledge; tim at rowledge.org; http://www.rowledge.org/tim
Strange OpCodes: HCF: Halt and Catch Fire
More information about the Vm-dev