DET1@BNANDP51.BITNET (06/15/91)
I saw not long ago a message asking help for installing a double
frequency cpu in the Amiga. I don't remember if it was from a usenet feed
or a bitnet feed, but anyway, I'm interested in making the conversion on
my amiga 500 (a friend of mine is ready to do it on his A2000-A/AT bridgeboard).
We plan to use a 68010-16 and follow the instructions of a schematic found on
an archive from this newsgroup.
My question is (to all of you) : does anyone have the 68000-16 working
in his amiga ?
(to Commodore folks) : does this ugly trick (ugly
because it makes an Amiga look like an atari ST inside) able to cause some
damage in an Amiga (to coprocessors, ram or so) ?
Hope you can help.
Jean-Francois Stenuit
The Lazy Coder
-------******* KISS (keep it simple and stupid) ***************-----------dale@boing.UUCP (Dale Luck) (06/16/91)
In article <7ABF45A76060644B@BNANDP51> DET1@BNANDP51.BITNET writes: > > I saw not long ago a message asking help for installing a double >frequency cpu in the Amiga. >We plan to use a 68010-16 and follow the instructions of a schematic found on >an archive from this newsgroup. > My question is (to all of you) : does anyone have the 68000-16 working >in his amiga ? I have an AdSpeed in an A500 we are evaluating to see if it is a more cost effective accelerator for X-Window type work then any of the other accelerators. It is very inexpensive compared to other accelerators that require extra ram. So far it is quite promising in terms of speed. We bought two for evaluation, one for an A500 and one for an A2000. One of them however would not work in the A500 so we've sent it back and are waiting for a replacement to try out. One of the nice things about this board is it's 16kbs of cache. It caches all fastmem accesses and watches the bus for any external dma to fast memory to invalidate it's internal cache if necessary. I don't believe it cares whether the fetches are data or code so cache control software should not be necessary. I believe it is has a write through cache so all writes are sent to regular memory. It is just reads that may be serviced from the cache memory. It could be a real winner, especially in systems that have C00000 halffast memory since the cached reads would no longer hit the chip bus when accessing that memory. Hope it speeds up serial interrupt response as well. We are trying to get a reliable 19.2Kbaud Compressed slip connection to work at the same time. We'll let you know how it goes. If anyone else has any stories they would like to send me about their Adspeeds, working or not, please email me. -- Dale Luck GfxBase/Boing, Inc. {uunet!cbmvax|pyramid}!amiga!boing!dale
daveh@cbmvax.commodore.com (Dave Haynie) (06/18/91)
In article <7ABF45A76060644B@BNANDP51> DET1@BNANDP51.BITNET writes: >We plan to use a 68010-16 and follow the instructions of a schematic found on >an archive from this newsgroup. The first flaw in your plan: Motorola only makes 68010s up to 12MHz. You can get 16MHz 68000s. > My question is (to all of you) : does anyone have the 68000-16 working >in his amiga ? The basic 14/16MHz hacks published around here in the past don't work. Some may work under some conditions, but in general, they are nothing but trouble. A few people may be working on more robust speedup hacks. If it takes less than a handful of parts, don't expect it to work. > (to Commodore folks) : does this ugly trick (ugly >because it makes an Amiga look like an atari ST inside) These hacks have nothing to do with an Atari ST. They are ugly because, in general, they don't work. There is a right way and a wrong way to build an accelerator device for a 68000 based Amiga. Every released hack I have seen so far except LUCAS is Just Plain Wrong. LUCAS misses here and there, but it's OK if you know what you're doing, don't have lots of peripherals attached, and don't mind mucking around with it a bit. >able to cause some damage in an Amiga (to coprocessors, ram or so) ? In general, you can't damage anything. You may find things don't work, and I wouldn't recommend running a non-functional machine for days on end, since you may very well have devices confused and operating in a state that could, over time, cause damage. -- Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests" {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: hazy BIX: hazy "This is my mistake. Let me make it good." -R.E.M.
cpmwc@marlin.jcu.edu.au (Matthew W Crowd) (06/18/91)
If you want to build a 16MHz 68000 "accelerator" for the amiga it's a
big waste of time.
The speed increase (if any?) is minimal ~10% i think, so why bother?
--
=======================================================================
Matt Crowd cpmwc@marlin.jcu.edu.au | "Operator, give me
| the number for 911"
Resident Computer Science FTP Mutant | - Homer Simpsondale@boing.UUCP (Dale Luck) (06/18/91)
In article cpmwc@marlin.jcu.edu.au (Matthew W Crowd) writes: > >If you want to build a 16MHz 68000 "accelerator" for the amiga it's a >big waste of time. >The speed increase (if any?) is minimal ~10% i think, so why bother? This really depends on what you are doing. If all you do is increase the speed of the 68000 chip itself you will see a large improvement if you use the floating point libraries. After the improved ieee libraries in 1.2 (nearly twice as fast as the previous version), the percentage of time spent by the processor in mul, div, shift instructions compared to the rest of the code increased to nearly 50% I believe in some cases. Since mul, div, shifts are all multiclock instructions which keep the cpu core real busy, doubling the clock speed of the cpu core will double the speed of these instructions. So for floating point intensive tasks, there is much more than a minimal 10% improvement in speed. The 68010 had about a 5% overall improvement in speed even at 7.14mhz because of the faster mul and div instructions on the chip. (faster than the 7.14mhz 68000) that is. >Matt Crowd cpmwc@marlin.jcu.edu.au | "Operator, give me -- Dale Luck GfxBase/Boing, Inc. {uunet!cbmvax|pyramid}!amiga!boing!dale