steve@wlbr.UUCP (Steve Childress) (09/30/85)
..
..
Ever wonder why Motorola's microprocessors (6800, 6809, 680x0) seem to
never get as much hype and acceptance as Intel's? Psssst! .. here's why.
Intel puts a divide-by-N counter for the main clock on their micro chips.
Not for marketing reasons, you understand. But in hindsight, it's clever.
Motorola doesn't. Hence, the idiots writing for the media belittle the
680x's for being far slower than the 80x's.
Goodness, gracious, a 2MHz 6809 simply *MUST* be inferior to a 6MHz Z80!
HEY MOTOROLA! The path to riches is obvious! Put a divider in the "68020A"
so that the outside clock frequency will be 500MHz! Or make it a hybrid
with a 100GHz clock!
I can imagine INFOWORLD headlines now!
Regards,
Steve Childress
Eaton IMS R&D Group MS 43
31717 La Tienda Drive
Westlake Village, CA 91360
(818) 889-2211 X2148
{trwrb, scgvaxd, ihnp4, voder, vortex} !wlbr!steve
or ...wlbr!wlbreng1!steveclif@intelca.UUCP (Clif Purkiser) (10/02/85)
> .. > .. > Ever wonder why Motorola's microprocessors (6800, 6809, 680x0) seem to > never get as much hype and acceptance as Intel's? Psssst! .. here's why. > > Intel puts a divide-by-N counter for the main clock on their micro chips. > Not for marketing reasons, you understand. But in hindsight, it's clever. > Motorola doesn't. Hence, the idiots writing for the media belittle the > 680x's for being far slower than the 80x's. > > Goodness, gracious, a 2MHz 6809 simply *MUST* be inferior to a 6MHz Z80! > > HEY MOTOROLA! The path to riches is obvious! Put a divider in the "68020A" > so that the outside clock frequency will be 500MHz! Or make it a hybrid > with a 100GHz clock! > > I can imagine INFOWORLD headlines now! > > Regards, > Steve Childress > Eaton IMS R&D Group MS 43 > 31717 La Tienda Drive > Westlake Village, CA 91360 > (818) 889-2211 X2148 > {trwrb, scgvaxd, ihnp4, voder, vortex} !wlbr!steve > or ...wlbr!wlbreng1!steve Maybe Steve's entire article needs an :-) around it. Otherwise (FlameOn) When Intel quotes a microprocessor clock speed it uses the internal divide- by-two clock not the external crystal. Thus, you use a 12 MHZ crystal in your PC-AT with a 6MHZ 80286, unless of course you've upgraded it to use a 18MHz crystal. At the risk of reopening the "Architecture Wars", Intel could quote the crystal speed and have the same speed bus (4 clocks) as an 68010, or an 32032 and a faster clocks speed 25Mhz for a 12.5Mhz 286. I have found that most of uninformed media concentrates on clock speed therefore a 10Mhz 68000 should be faster than a 6Mhz 286 because 10 > 6. Needless to say this line of reasoning is falicious. If say my Vette runs at 4000RPM, and you say that your Mazada RX-7 runs at 5500RPM, I can not say which car runs faster, unless I know what gears the cars are in and what the gearing ratio is. The auto industry has solved this problem by inventing a very handy standard benchmark, Miles Per Hour (MPH). The computer industry has yet to develop such a benchmark, but we kept trying :-). -- Clif Purkiser, Intel, Santa Clara, Ca. HIGH PERFORMANCE MICROPROCESSORS {pur-ee,hplabs,amd,scgvaxd,dual,idi,omsvax}!intelca!clif {standard disclaimer about how these views are mine and may not reflect the views of Intel, my boss , or USNET goes here. }
henry@utzoo.UUCP (Henry Spencer) (10/04/85)
> Intel puts a divide-by-N counter for the main clock on their micro chips. > Not for marketing reasons, you understand. But in hindsight, it's clever. > Motorola doesn't.... > HEY MOTOROLA! The path to riches is obvious! Put a divider in the "68020A" > so that the outside clock frequency will be 500MHz! Or make it a hybrid > with a 100GHz clock! Actually, in a BYTE article on the 6809 some years ago, its designers said something like: We thought about putting a few divider stages on the clock inputs so we could win the highest-clock-speed race, but we decided that system designers wouldn't appreciate having to put a tuned cavity on their boards for the clock oscillator... -- Henry Spencer @ U of Toronto Zoology {allegra,ihnp4,linus,decvax}!utzoo!henry
jimomura@lsuc.UUCP (Jim Omura) (10/06/85)
In article <370@wlbr.UUCP> steve@wlbr.UUCP (Steve Childress) writes: >.. >.. >Ever wonder why Motorola's microprocessors (6800, 6809, 680x0) seem to >never get as much hype and acceptance as Intel's? Psssst! .. here's why. > >Intel puts a divide-by-N counter for the main clock on their micro chips. >Not for marketing reasons, you understand. But in hindsight, it's clever. >Motorola doesn't. Hence, the idiots writing for the media belittle the >680x's for being far slower than the 80x's. > >Goodness, gracious, a 2MHz 6809 simply *MUST* be inferior to a 6MHz Z80! > >HEY MOTOROLA! The path to riches is obvious! Put a divider in the "68020A" >so that the outside clock frequency will be 500MHz! Or make it a hybrid >with a 100GHz clock! > >I can imagine INFOWORLD headlines now! > > Regards, > Steve Childress > Eaton IMS R&D Group MS 43 > 31717 La Tienda Drive > Westlake Village, CA 91360 > (818) 889-2211 X2148 > {trwrb, scgvaxd, ihnp4, voder, vortex} !wlbr!steve > or ...wlbr!wlbreng1!steve Actually the 68000 family isn't all that fast. Check some cycle counts and you'll find that the 68000 and 68008 are cycle hogs. I don't have figures for the 68010, but the 68020 was a substantial improvement. I think they dropped the mean from over 4 cycles to about 3 cycles for most instructions. That plus much higher clock speed makes it a very much faster processor. -- James Omura, Barrister & Solicitor, Toronto ihnp4!utzoo!lsuc!jimomura Byte Information eXchange: jimomura Compuserve: 72205,541 MTS at WU: GKL6
wdm@ecn-pc.UUCP (William D Michael) (10/08/85)
In article <106@intelca.UUCP> clif@intelca.UUCP (Clif Purkiser) writes: >> Ever wonder why Motorola's microprocessors (6800, 6809, 680x0) seem to >> never get as much hype and acceptance as Intel's? Psssst! .. here's why. >> >> Intel puts a divide-by-N counter for the main clock on their micro chips. >> Not for marketing reasons, you understand. But in hindsight, it's clever. >> Motorola doesn't. Hence, the idiots writing for the media belittle the >> 680x's for being far slower than the 80x's. >> >> HEY MOTOROLA! The path to riches is obvious! Put a divider in the "68020A" >> so that the outside clock frequency will be 500MHz! Or make it a hybrid >> with a 100GHz clock! >> >> I can imagine INFOWORLD headlines now! > >Maybe Steve's entire article needs an :-) around it. > >Otherwise (FlameOn) >When Intel quotes a microprocessor clock speed it uses the internal divide- >by-two clock not the external crystal. Thus, you use a 12 MHZ crystal in >your PC-AT with a 6MHZ 80286, unless of course you've upgraded it to use >a 18MHz crystal. I don't know what an official Intel quotation is, but the Intel salesmen in our area usually quote the crystal frequency. >At the risk of reopening the "Architecture Wars", Intel >could quote the crystal speed and have the same speed bus (4 clocks) >as an 68010, or an 32032 and a faster clocks speed 25Mhz for a 12.5Mhz 286. >I have found that most of uninformed media concentrates on clock speed >therefore a 10Mhz 68000 should be faster than a 6Mhz 286 because 10 > 6. And, in our benchmarks, the 10 MHz 68000 WAS faster than the 6MHz 286. >Needless to say this line of reasoning is falicious. If say my Vette >runs at 4000RPM, and you say that your Mazada RX-7 runs at 5500RPM, I >can not say which car runs faster, unless I know what gears the cars >are in and what the gearing ratio is. The auto industry has solved >this problem by inventing a very handy standard benchmark, Miles >Per Hour (MPH). >The computer industry has yet to develop such a benchmark, but we kept >trying :-). >Clif Purkiser, Intel, Santa Clara, Ca. >HIGH PERFORMANCE MICROPROCESSORS Surely you are not talking about the 86 family. bill michael (wdm@pur-ee)
savage@ssc-vax.UUCP (Lowell Savage) (10/09/85)
> ... (FlameOn) > ... If say my Vette > runs at 4000RPM, and you say that your Mazada RX-7 runs at 5500RPM, I > can not say which car runs faster, unless I know what gears the cars > are in and what the gearing ratio is. The auto industry has solved > this problem by inventing a very handy standard benchmark, Miles > Per Hour (MPH). > > The computer industry has yet to develop such a benchmark, but we kept > trying :-). IPS (Instructions Per Second)? FLOPS (FLoating Operations Per Second)? There will be objections to this analogy by saying that IPS (or MIPS) and FLOPS don't necessarily represent the true speed of the processor in a "typical" application--but then neither does MPH represent the true speed of a car on a "typical" trip. So...we have benchmarks, er races, which are fun to watch and participate in!! There's more than one way to be savage Lowell Savage Obviously, the opinions expressed above are those of every sentient being including my left shoe.
rb@ccivax.UUCP (rex ballard) (10/10/85)
> > Intel puts a divide-by-N counter for the main clock on their micro chips. > > Not for marketing reasons, you understand. But in hindsight, it's clever. > > Motorola doesn't.... > > HEY MOTOROLA! The path to riches is obvious! Put a divider in the "68020A" > > so that the outside clock frequency will be 500MHz! Or make it a hybrid > > with a 100GHz clock! > > Actually, in a BYTE article on the 6809 some years ago, its designers said > something like: > > We thought about putting a few divider stages on the clock inputs > so we could win the highest-clock-speed race, but we decided that > system designers wouldn't appreciate having to put a tuned cavity > on their boards for the clock oscillator... > -- > Henry Spencer @ U of Toronto Zoology > {allegra,ihnp4,linus,decvax}!utzoo!henry Take a look at the "bus Cycle" for a 68000/10/20, there are 4-6 "clocks" per cycle. Much of that is spent Asynchronously doing address calculations for the extensive instruction set. Fortunately, the wider registers, ALU, and internal busing still leave the 8086 in the dust!
peter@graffiti.UUCP (Peter da Silva) (10/12/85)
> Actually the 68000 family isn't > all that fast. Check some cycle counts > and you'll find that the 68000 and > 68008 are cycle hogs. I don't have > figures for the 68010, but the 68020 > was a substantial improvement. I > think they dropped the mean from over > 4 cycles to about 3 cycles for most > instructions. That plus much higher > clock speed makes it a very much faster > processor. But the 68000 has a more powerful instruction set, so the average number of cycles per instruction are matched by a larger number of operations per instruction. The FORTH inner interpreter loop takes up 3 instructions on the 68000, as compared to about half a dozen on the 80xxx. Other operations are similarly enhanced. Plus you don't need to waste instructions and opcode fetches playing around with the DS, CS, SS, and ES registers.
steve@wlbr.UUCP (Steve Childress) (10/17/85)
Re the notion that the 68xx's use fewer instructions than 80xxx's to accomplish a given extent of work... About a year ago, I compared the assembly code emitted by an 80186 Xenix (Intel) FORTRAN (they had no large-scale C) to the code produced by a 68K C compiler. The FORTRAN code was about TEN times as lengthy as the C code for the 68000. The ratio of useful code to overhead code (address arithmetic, segment switching, operand fetch/put, etc) was about 8:1 for the 80xxx and about 3:1 for the 68000. And the code size for the 80xxx followed suit so that the 80xxx code was quite a bit larger even though its average instruction length was smaller.