billium@pro-charlotte.cts.com (Bill Blanke) (08/29/88)
I was flipping through a magazine and saw an ad for CMI's processor accelerator. Its supposed to double the clock speed of the Amigas processor from 7.12 to 14.3 MHz. I've got a few questions about it. (1) Does anybody have this on there Amiga? Has anybody seen it? (2) How compatible is it with Amiga software. I remember someone was saying how it was incompatible with dongles. Thats no problem for me though since none of my software is dongled. (3) Does the 2x speed increase affect games. I guess what I'm asking is if Amiga games are timer based (wait 1 second and then do this) or processor cycle based (execute these instructions as fast as you can, then do this-I would have to deal with an arkanoids ball flighing around twice as fast as normal!). If all are timer based (hopefully) does this mean sometimes sluggish games (like Obliterator) would perk up and move more smoothly, but still remain at their current speed? If they are cycle based is there a way to revert back to 7.12 Mhz, preferably through a hardware switch? (4) What kind of speed increase could I expect over the normal 68000 with ray tracing programs, general multi-tasking, Desktop Publishing programs (not moving memory-blitter does that-I mean all the reflowing of text and all the recalculating the program has to do) and Transformer? (5) How much better would the speed increase be with the math coprocessor? Would the coprocessor be available for all programs or just those that are writtin to use it? Would the 1.3 math libraries take care of those that aren't? If the overall speed increase is good I'd buy one. $200 is a great deal considering $$$ for 68020,30....
daveh@cbmvax.UUCP (Dave Haynie) (08/31/88)
in article <3372@crash.cts.com>, billium@pro-charlotte.cts.com (Bill Blanke) says: > I was flipping through a magazine and saw an ad for CMI's processor > accelerator. Its supposed to double the clock speed of the Amigas > processor from 7.12 to 14.3 MHz. I've got a few questions about it. > (2) How compatible is it with Amiga software. I remember someone > was saying how it was incompatible with dongles. Thats no problem > for me though since none of my software is dongled. So far, the X-CAD dongle has proven to be sensitive to CPU speed. The SuperBase dongle isn't. I haven't used any others myself. > (3) Does the 2x speed increase affect games. I guess what I'm asking > is if Amiga games are timer based (wait 1 second and then do this) > or processor cycle based (execute these instructions as fast as > you can, then do this-I would have to deal with an arkanoids ball > flighing around twice as fast as normal!). If all are timer based > (hopefully) does this mean sometimes sluggish games (like Obliterator) > would perk up and move more smoothly, but still remain at their > current speed? There are a few exceptions, but most Amiga games are synced to the video display rate. If you're using the Copper for anything, this probably falls out automatically, for the most part. The other thing to consider is that Commodore-Amiga doesn't support software that's CPU speed dependent. You should be able to run your 68040 at 60MHz and still expect everything to work A-OK. Even Commodore-Amiga has had a speed related bug or two show up, but we fix them. So should any other software vendors. > If they are cycle based is there a way to revert back to 7.12 Mhz, > preferably through a hardware switch? I would expect it to be pretty difficult, and wouldn't recommend it. The easiest thing they could do in this respect, at least in an A2000 CPU slot, would be to let the A2000's 68000 back on under program control. I use fast 68020s almost all the time now, and have run into very few problems with the software I use (admittedly, very few games). -- Dave Haynie "The 32 Bit Guy" Commodore-Amiga "The Crew That Never Rests" {ihnp4|uunet|rutgers}!cbmvax!daveh PLINK: D-DAVE H BIX: hazy "I can't relax, 'cause I'm a Boinger!"
billc@percival.UUCP (William Coldwell) (08/31/88)
In article <3372@crash.cts.com> billium@pro-charlotte.cts.com (Bill Blanke) writes:
:I was flipping through a magazine and saw an ad for CMI's processor
:accelerator. Its supposed to double the clock speed of the Amigas
:processor from 7.12 to 14.3 MHz. I've got a few questions about it.
Hope I can answer them...
:(1) Does anybody have this on there Amiga? Has anybody seen it?
Yes, and I see it every day - whether I want to or not.
:(2) How compatible is it with Amiga software. I remember someone
: was saying how it was incompatible with dongles. Thats no problem
: for me though since none of my software is dongled.
So far, there really isn't anything that it's software incompatable
with, and the problem with dongles is because they (those pesky software
people) are using software loops to get the info from the dongle, and
viola: it's finished twice as fast as it was expecting...
:(3) Does the 2x speed increase affect games. I guess what I'm asking
: is if Amiga games are timer based (wait 1 second and then do this)
: or processor cycle based (execute these instructions as fast as
: you can, then do this-I would have to deal with an arkanoids ball
: flighing around twice as fast as normal!). If all are timer based
: (hopefully) does this mean sometimes sluggish games (like Obliterator)
: would perk up and move more smoothly, but still remain at their
: current speed?
Yes, it does affect most games that are not timer based. Generally making
them more smoother, rather than "faster". F-18 and Jet are the same speed,
just smoother in all of the calculations. Obliterator is nice, but you see
a real difference with Barbarian.
: If they are cycle based is there a way to revert back to 7.12 Mhz,
: preferably through a hardware switch?
Would you settle for a software switch? (That's what I do.) - If you
want to make a toggle, that's fine (unless you work for the FCC, then
in no means are we condoning such criminal activities...) - just make
sure that you rip up your warranty card after doing it ;-). Incidently,
the Amiga runs at 7.16, not 7.12MHz.
:(4) What kind of speed increase could I expect over the normal 68000 with
: ray tracing programs, general multi-tasking, Desktop Publishing
: programs (not moving memory-blitter does that-I mean all the reflowing
: of text and all the recalculating the program has to do) and
: Transformer?
Ok, general rule of thumb: Anything done with the 68000 is twice as
fast, especially math calculation functions.
:(5) How much better would the speed increase be with the math coprocessor?
: Would the coprocessor be available for all programs or just those
: that are writtin to use it? Would the 1.3 math libraries take care of
: those that aren't?
Only programs that use the IEEE libraries for math functions will use the
68881/2, and no, 1.3 won't make programs that don't use the IEEE shtuf
use the 68881 - but the ones that do... unreal. Hey, you turkeys who
are developing applications that use math functions, check to see if
there is a MathIEEE resource, and have your routines use IEEE!!!!!
Otherwise use FFP. I'll be more than happy to post how to check whether
there is a 68881 installed!
:If the overall speed increase is good I'd buy one. $200 is a great deal
:considering $$$ for 68020,30....
The speed is pretty comparable to a 68020 with 16 bit memory, and with
all of the people who have flooded us with orders, Thanks!
Bill
Look, just because I am working for CMI doesn't mean that I speak for
anyone else here, and they don't speak for me - it's hard enough
trying to speak for MYSELF!
--
William J. Coldwell - Amiga Attitude Adjuster {aka Software Developer}
(503) 684-9300 ...tektronix!reed!percival!billc "Blame the hardware."
[I break for 14MHz Processor Accelerators with 68881's.]
Creative Microsystems Inc. / 10110 SW Nimbus Suite B1 / Portland, OR 97223
mp1u+@andrew.cmu.edu (Michael Portuesi) (09/24/88)
> *Excerpts from ext.nn.comp.sys.amiga: 31-Aug-88 Re: CMI 68000 2x William* > *Coldwell@perciva (3966)* > The speed is pretty comparable to a 68020 with 16 bit memory, and with > all of the people who have flooded us with orders, Thanks! Could you post some benchmarks to validate this claim? The CMI board sounds interesting, but I wouldn't consider buying it without having heard hard facts instead of objective opinions, especially from people who stand to benefit from its sale. --M Michael Portuesi / Information Technology Center / Carnegie Mellon University ARPA/UUCP: mp1u+@andrew.cmu.edu BITNET: rainwalker@drycas "my friends say she's a dumb blonde, but they don't know she dyes her hair"
jmdavis@ihlpm.ATT.COM (Davis) (09/26/88)
In article <wXCdIny00Vsf00YG0P@andrew.cmu.edu>, mp1u+@andrew.cmu.edu (Michael Portuesi) writes: > > Could you post some benchmarks to validate this claim? The CMI board sounds > interesting, but I wouldn't consider buying it without having heard hard facts > > Yes, I too am in favor of benchmarks, if they are good I would seriously consider buying this product. So please post them to the net. BTW, I originally tried to reply to one of the CMI folks who posted to this thread, but my mailer bounced it. -- ________________________________________ | Mike Davis | ..!att!ihlpm!jmdavis |_________________________
iphwk%MTSUNIX1.BITNET@cunyvm.cuny.edu (Bill Kinnersley) (09/27/88)
[In "Re: CMI 68000 2x", Davis said:]
: In article <wXCdIny00Vsf00YG0P@andrew.cmu.edu>, mp1u+@andrew.cmu.edu (Michael
: Portuesi) writes:
: > Could you post some benchmarks to validate this claim? The CMI board
: > sounds interesting, but I wouldn't consider buying it without having
: > heard hard facts
: Yes, I too am in favor of benchmarks, if they are good I would seriously
: consider buying this product. So please post them to the net.
:
: | Mike Davis
: | ..!att!ihlpm!jmdavis
:
I have a board which is somewhat similar, an 8MHz 68010/68881 combination
from Netch Comp Prods. I would like to benchmark it too...could someone
suggest a *good* test to run? (e.g. not Savage!)
Here are some qualitative impressions of the board I have:
1) Impressive claims for speedup are hard to verify in practice. True,
you can do things like square roots and cosines much faster with the 68881.
But even programs which appear to be very numerically intensive wind up
being dominated by the speed of the simplest operations: adds, subtracts,
moves, stores, etc.
2) Both the 68000 and 68010 drive the 68881 as a peripheral device.
This plus the need for decoding the floating point instructions in software
make things much slower than a 68020/68881 would be.
3) My experience is that I can expect to do IEEE with the board about as
fast as FFP without the board. Of course the 68881 will only do IEEE.
The net result is that I get double precision for free. So if my programs
needed that much precision I would be better off. They usually don't.
billsey@agora.UUCP (Bill Seymour) (09/30/88)
From article <wXCdIny00Vsf00YG0P@andrew.cmu.edu:, by mp1u+@andrew.cmu.edu (Michael Portuesi): :: *Excerpts from ext.nn.comp.sys.amiga: 31-Aug-88 Re: CMI 68000 2x William* :: *Coldwell@perciva (3966)* : :: The speed is pretty comparable to a 68020 with 16 bit memory, and with :: all of the people who have flooded us with orders, Thanks! : : : Could you post some benchmarks to validate this claim? The CMI board sounds : interesting, but I wouldn't consider buying it without having heard hard facts : instead of objective opinions, especially from people who stand to benefit from : its sale. : Needless to say, Bill is being a little optimistic here. In several cases, the Processor Accelerator gives speedups in the range of 35% while a 68020 board running at 14Mhz gives about 50%. I'm sure that's the type of comparison he ment. My personal experience is that it almost all cases, the 68020 board (I've only tested the Hurricane) runs somewhere between our speed and twice our speed. You get better results for the '020 if it's coupled with an '881 and much better results if you have 32 bit memory. The point is that the 68020 board runs around $800 while ours is only $200. This makes the PA a viable option for people who want more speed without spending so much money. PRetty much all the benchmarks I have are targeted more at showing off math chips than by showing off processor speeds. Larry Gutkowski has written some 68881 supports routines for use with the Absoft FORTRAN compiler and I have results from these. (The benchmarks and results are posted in seperate files on PLink for those who have access, or know someone who has access, to PLink.) There is an article coming out in Amiga World RSN with some benchmark times included. And I've seen some articles published in user group newsletters where benchmark time have been included. Unfortunately I don't have any of them here at home. I would be happy to run benchmarks for people, but I had to send the Hurricane back, so I'll only be able to give results for the PA, 68010 and standard machines... : --M : : Michael Portuesi / Information Technology Center / Carnegie Mellon University : ARPA/UUCP: mp1u+@andrew.cmu.edu BITNET: rainwalker@drycas : : "my friends say she's a dumb blonde, but they don't know she dyes her hair" Bill -- -Bill Seymour ...tektronix!reed!percival!agora!billsey ...tektronix!sequent!blowpig!billsey Creative Microsystems Northwest Amiga Group At Home Sometimes (503) 684-9300 (503) 656-7393 BBS (503) 640-0842
billsey%agora.uucp@UDEL.EDU (10/04/88)
Received: from CUNYVM by CUNYVM.BITNET (Mailer X2.00) with BSMTP id 5494; Sun, 02 Oct 88 01:06:07 EDT Received: from UDEL.EDU by CUNYVM.CUNY.EDU (IBM VM SMTP R1.1) with TCP; Sun, 02 Oct 88 01:06:03 EDT Received: from Louie.UDEL.EDU by Louie.UDEL.EDU id aa02023; 1 Oct 88 13:10 EDT Received: from USENET by Louie.UDEL.EDU id aa01908; 1 Oct 88 13:04 EDT From: Bill Seymour <billsey@agora.uucp> Subject: Re: CMI 68000 2x Message-ID: <1176@agora.UUCP> Date: 30 Sep 88 14:27:48 GMT Organization: Advanced Solutions, Hillsboro, OR To: amiga-relay@UDEL.EDU Sender: amiga-relay-request@UDEL.EDU From article <wXCdIny00Vsf00YG0P@andrew.cmu.edu:, by mp1u+@andrew.cmu.edu (Michael Portuesi): :: *Excerpts from ext.nn.comp.sys.amiga: 31-Aug-88 Re: CMI 68000 2x William* :: *Coldwell@perciva (3966)* : :: The speed is pretty comparable to a 68020 with 16 bit memory, and with :: all of the people who have flooded us with orders, Thanks! : : : Could you post some benchmarks to validate this claim? The CMI board sounds : interesting, but I wouldn't consider buying it without having heard hard facts : instead of objective opinions, especially from people who stand to benefit from : its sale. : Needless to say, Bill is being a little optimistic here. In several cases, the Processor Accelerator gives speedups in the range of 35% while a 68020 board running at 14Mhz gives about 50%. I'm sure that's the type of comparison he ment. My personal experience is that it almost all cases, the 68020 board (I've only tested the Hurricane) runs somewhere between our speed and twice our speed. You get better results for the '020 if it's coupled with an '881 and much better results if you have 32 bit memory. The point is that the 68020 board runs around $800 while ours is only $200. This makes the PA a viable option for people who want more speed without spending so much money. PRetty much all the benchmarks I have are targeted more at showing off math chips than by showing off processor speeds. Larry Gutkowski has written some 68881 supports routines for use with the Absoft FORTRAN compiler and I have results from these. (The benchmarks and results are posted in seperate files on PLink for those who have access, or know someone who has access, to PLink.) There is an article coming out in Amiga World RSN with some benchmark times included. And I've seen some articles published in user group newsletters where benchmark time have been included. Unfortunately I don't have any of them here at home. I would be happy to run benchmarks for people, but I had to send the Hurricane back, so I'll only be able to give results for the PA, 68010 and standard machines... : --M : : Michael Portuesi / Information Technology Center / Carnegie Mellon University : ARPA/UUCP: mp1u+@andrew.cmu.edu BITNET: rainwalker@drycas : : "my friends say she's a dumb blonde, but they don't know she dyes her hair" Bill -- -Bill Seymour ...tektronix!reed!percival!agora!billsey ...tektronix!sequent!blowpig!billsey Creative Microsystems Northwest Amiga Group At Home Sometimes (503) 684-9300 (503) 656-7393 BBS (503) 640-0842