simple question but could the resolution be added to the end of the prescale?
x1 (256 x 192)
x2 (512 x 384)
x3 (768 x 576)
x4 (1024 x 768)
x5 (1280 x 960)
Also for people that have the computer to handle it could a X6 prescale be added?
Last edited by fintogive (2016-10-31 00:50:17)
There is no technical limit to what dimensions the output framebuffer can be. The 5x resolution you see is simply a limitation of the Windows port. The Cocoa port can get up to 16x.
Practically speaking, if you're using anything less than an Intel i7 Skylake 6700K CPU, your performance will become poor beyond 6x. Even a Skylake 6700K CPU will start experiencing poor performance beyond 8x.
well i have a fast computer that why i was wanting an x6. i know it has slow down issues. but why is the windows port slow?
with high res is can be.
Most of the performance issues with the Windows port has to do with the following:
1. Using the CPU to perform video tasks that should be done on the GPU (scaling, rotation, filters, etc.)
2. Extraneous framebuffer copying.
3. Video output is not DMA transferred from main RAM to the GPU.
The DMA transfer issue is a major bottleneck in the Windows port that is greatly exacerbated by high resolution rendering. For the record, DeSmuME X432R has a Direct3D display mode that does support DMA transfers on virtually every GPU out there. In fact, this is one of the big performance advantages that X432R has over our Windows port.
Last edited by rogerman (2016-11-02 23:23:59)
I don't see how the opengl display method could do anything any faster unless the 2d compositing ended up being done on the gpu. I've got no clue how shoddy directdraw is on modern systems, except probably 'very'
hmmm... well why not port the direct 3d mode code from x432r to the windows port? or make your own direct 3d mode?
and zeromus for me opengl is a bit faster on my computer but not by much.
i doubt rogerman's explanations. i think the video outputs are DMAed in either case. I don't think direct3d is magic.