Logic gates to select single values on an 8/16-bit bus

[james_s]

I was thinking of some graphical games that ran so fast on a 286 or 386 that they were unplayable. I remember one in particular on a 33MHz 486 you would hit start and Game Over popped up in about 100mS.

[ALonda]

I would think that "I designed and built my own cpu" would rank pretty high for bragging rights, no matter how fast it is.

In any case, since most logic chips are rated for 40-50MHz I imagine a cpu based on them could atleast hit 10MHz or so before turning into a quantum time teleporter machine.

/A

[rstofer]

I would think that "I designed and built my own cpu" would rank pretty high for bragging rights, no matter how fast it is.

In any case, since most logic chips are rated for 40-50MHz I imagine a cpu based on them could atleast hit 10MHz or so before turning into a quantum time teleporter machine.

/A

There is serious cred for building your own CPU, regardless of how it is done.

If, for example, you want a clock rate of 20 MHz (fairly fast, a Z80 only ran 6 MHz, the 8080 only 2 MHz), you have 50 nS to do memory access if you don't want a wait state.  If you give up half that in decoding, there isn't much time left for t_acc of the memory.

It's just a number to look at.  If might not be an issue at all!  I haven't looked at the datasheets so I have no idea how much time is being used in decoding.

When I first started messing with computers in 1970, I was using an IBM 1130.  Somewhere around '74 I built a bipolar emulation using blown-link PROMs to hold microcode and 74181s for the ALU, all wire-wrapped.  Memory wasn't available at that time so I waited until 2003 or so to build it over again with an FPGA.  My version runs all of the IBM software, unchanged.  And it runs at 50 MHz versus 400 kHz of the original.  With just a little effort, I could probably kick it up to 100 MHz, 250 times the speed of the original!  On a chip the size of a postage stamp.