Quote:
Originally Posted by VladR
Vampire is already maxing out the board,from what i could infer reading the forums, so it's hard to say exactly how much more performance could be extracted from that FPGA board without knowing how much space does each feature take.
|
The Vampire's FPGA doesn't have the integrated ARM CPU, though.
Quote:
But I would reckon that if we dumped AMMX, then we could have the current 2 integer cores plus a parallel FP unit.
|
Only if you can persuade someone sufficiently talented to work on it full time for months or years. (I would love to be proved wrong, but I very much doubt the Vampire team will make their CPU available for use on MiSTer any time soon!)
Quote:
Perhaps if those were simpler cores, 4 of them would fit ?
|
There's a pretty speedy MIPS-compatible CPU that would fit somewhere between 25 and 50 times over in the DE10-nano. You could probably get another 8 or 9 TG68 CPUs into the MiSTer Minimig core without having to remove anything.
(I did one squeeze a design with 26 instances of a ZPU processor onto the lowly DE1 dev-board - about 10% of the size of the DE10-nano's FPGA!)
Quote:
Of course, that would be a non-existing Amiga HW, as there never ever was an Amiga quad-core and most people would revolt at such an "abomination" ...
|
The OS would revolt too!
A far more interesting use of the extra FPGA resources would be the addition of special-purpose hardware for specific projects. Imagine being able to compile a core with you own custom hardware expansion to do your coordinate transforms without the CPU having to do the number crunching. A bit like how SNES cartridges eventually came to have extra hardware within them.
Quote:
I can't do anything with those, that's for sure. The last thing I need to do is start playing with programming cores (as much as I would love to) - that would absolutely guarantee I never ever get to finishing my games...
But because FPGA space is the single most precious resource (from what I can infer indirectly), I figured it's worth asking if I shouldn't just splurge for a slightly more expensive board.
|
It's a tricky question to answer - FPGA space is one of those things where if you've got enough, then more is just a waste, but if you haven't got *quite* enough, boy do you feel it!
Quote:
I might as well not bother with messing with an FPGA board in the first place, and just do all testing directly under WinUAE.
|
If you're never intending your software to run on '030-level machines then I'd tend to agree - once you get into '040/'060 realms and beyond there's no longer any expectation of accurately matching a particular machine's performance.