Messages in this thread | | | Date | Fri, 4 Sep 1998 16:41:33 -0400 (EDT) | From | Alex Buell <> | Subject | Re: AMD K6 233 Cpu 465 BogoMips?? |
| |
On Fri, 4 Sep 1998 kwrohrer@ce.mediaone.net wrote:
> I disagree; while we might wish AMD would do an x86 SMP system and get > some chipset vendor and motherboard maker(s) to support it, it doesn't > make good business sense for AMD. Look at how Intel is struggling now > with Chrome* and other efforts trying to come up with an excuse for > people who can afford a high-end Intel processor to want one.
Hmm... interesting points to ponder. Now that we've got GPL'd sourceware, how about GPL'd motherboards and GPL'd processors? That is, a bunch of hardware engineers and operating system architects could get together and design a cheap and easily manufacturable motherboard and a processor to go with it and stick the GPL on it. I've designed on paper back in the late 1980s a nice processor that's closely modelled on the 680x0 in concept but has a RISC instruction set with less than 256 opcodes (the objective was to let the opcode fit into a byte), with 32 bit operands. The lower 4 bits of the opcode defined the mode of addressing (i.e 0 = 32 bits, 1 = 16 bits, 2 = 8 bits) for each operand it operated upon, thus making opcode decoding extraordinarily efficient (unlike the horrible x86 instruction set!) I then determined that it could be capable of executing one complete opcode with its operand per clock cycle, thus potentially depending upon the clock generation pulses at one megahertz, it could do a million instructions per second, and adding heavy predictive pipeling could well increase that to perhaps 10 instructions per cycle. Given 1GHz to play with, it could perhaps do a billion (or with 10 instructions a cycle - *ten* billion) operations a second. It would be interesting to devise a SMP system that's superior to anything else out there. I didn't design a bus to go with the processor though, but one would suppose one would have to deal with bus arbitration, bus mastering issues et. al, and issues with I/O intensive operations - all best dealt with custom silicon, or better still, dealt with by another processor dedicated expressly for this purpose. We could even have this processor act as a cache controller to the rest of the system for I/O purposes and as well as a data feed to the other processors. This isn't new - the transputer SMP works like this (I think - last time I read about the transputers was in the late 80s). How about a cheap supercomputer on your desktop with 10 processors in tandem with an I/O processor, that's capable of generating 20 fps broadcast quality video in real time? Or even model the world's climate in real time with a 2,000 processor array?
Along these lines, I also designed a A4 notebook sized computer with a slim 3.5" floppy disk in it, along with a wireless pen (infra-red) one could use.
Yes, you can tell I've been smoking pot as long ago as then. :o)
Cheers, Alex. --- /\_/\ Legalise cannabis now! ( o.o ) Grow some cannabis today! > ^ < Peace, Love, Unity and Respect to all.
Check out http://www.tahallah.demon.co.uk
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/faq.html
| |