Edward_Braswell@a68k.UUCP (Edward Braswell) (06/04/91)
Some of you may remember the old "14 MHz Hack" that's been floating around for A500's (and presumably A2000's). After MUCH! trying, I finally decided it wouldn't work for the A1000, due to differences in timing, in that the A1000 uses a couple of 74F74's and some other logic to generate it's 7M clock for the 68000, while the A500 has all of that stuff integrated into the "Fat Agnus" chip. But, I did find an accellerator that does work. It's not 14 MHz, it's only 8 MHz (7.14 -> 8 MHz => 12% increase in speed). And, it only cost me $2.50 for a 32 MHz crystal oscillator. In checking with the program RAMSpeed (and using a 68010), it gave me the following benchmarks: Clock: 8.209 MHz Chip Ram: 7.747 MHz Fast Ram: 7.519 MHz (My "Fast" ram board apparently has 1 wait state) F80000 Ram: 7.776 MHz (This is the 256K of fast ram ya get from placing Kickstart in EPROMS) The program SYSINFO gave the following "comparisons:" A 500 Standard: 1.54 B 2000 w/ Fast: 1.17 I'm not sure why a standard A500 is so slow, possibly that is for ALL CHIP ram and no fast ram. With the B 2000 (A 2000), you can see the difference. 17% faster breaks down to 5% from having a 68010, and 12% due to running at 8 MHz. What I did was place the 32 MHz clock onto lines from the RGB out connector for EXternalCLocKIN, EXCLKENable, +5 volts, and ground. I places a switch between ground and both the EXCLKEN and the ground for the crystal itself, as the Enable is "Active Low". With the switch, I can switch from normal to FAST and back "On The Fly" with no problems! Actually, there _IS_ one small problem with this hack: while in the fast mode, the RGB and video outputs are UNREADABLE. The screens get very messed up, as their Syncs are not at the specified NTSC frequencies, and my monitors can't "sync up". OK, yes, this is a major problem, but there are 2 solutions: 1. Modify the monitors so that they CAN sync with the new speed. 2. Use FAST mode blind. Yes, I know, what's the use of an accellerator if you can't see an output? Well, one use is for things you don't need to see, like ray-tracing. If you're doing heavy-duty ray tracing on an unaccellerated 68000 system, then the FAST mode would cut the time by 12% (and getting a 68010 would bring the saving to 17%!). Let's say the picture would normally take 2 hours to draw. You could save 15 minutes! Another good use for it is in LZHing / UN LZHing / Warping / unwarping, etc.. large files. If you don't mind waiting until it's done to see anything, this hack will save you time. Ultimately, what can you expect from a $2.50 accellerator? I think it's a bargain. Oh, if anyone wants specific pin outs / schematic, just ask and I'll upload them.
taab5@isuvax.iastate.edu (Marc Barrett) (06/06/91)
In article <Edward_Braswell.3260@a68k.UUCP>, Edward_Braswell@a68k.UUCP (Edward Braswell) writes: >Some of you may remember the old "14 MHz Hack" that's been floating around >for A500's (and presumably A2000's). After MUCH! trying, I finally >decided it wouldn't work for the A1000, due to differences in timing, in >that the A1000 uses a couple of 74F74's and some other logic to generate >it's 7M clock for the 68000, while the A500 has all of that stuff >integrated into the "Fat Agnus" chip. > >But, I did find an accellerator that does work. It's not 14 MHz, it's >only 8 MHz (7.14 -> 8 MHz => 12% increase in speed). And, it only cost me >$2.50 for a 32 MHz crystal oscillator. [stuff deleted] I tried exactly this a couple of years ago, only I tried it with both 32Mhz and 40Mhz clock oscillators. I found that the monitor I was using at the time (a cheap composite monochrome monitor, to alleviate the interlace flicker) was able to sink up to the higher video frequency when the 32Mhz oscillator was used, but not when the 40Mhz crystal was used. For this reason, I think a multisync monitor should be able to sort everything out and produce a solid display when this hack is done. Now that I have a multisync monitor, I've been wanting to try the hack again, but I now longer have the thing (it was cheaply wire-wrapped, and I threw it out long ago). The next time I get around to it, I will try the hack with the multisync monitor and report my results. BTW, this hack, when used with the 40Mhz clock oscillator, can produce a 40% speedup clear across the board on everything from the blitter to the sound chip to the CPU. This is because, unlike all other acclerator boards or speedup hacks, this speeds up the custom chips as well as the CPU. > Actually, there _IS_ one small problem with this hack: while in the fast >mode, the RGB and video outputs are UNREADABLE. The screens get very >messed up, as their Syncs are not at the specified NTSC frequencies, and >my monitors can't "sync up". OK, yes, this is a major problem, but there >are 2 solutions: > > 1. Modify the monitors so that they CAN sync with the new speed. > 2. Use FAST mode blind. Or 3. Use a multisync monitor. See above. > >Yes, I know, what's the use of an accellerator if you can't see an output? > Well, one use is for things you don't need to see, like ray-tracing. If >you're doing heavy-duty ray tracing on an unaccellerated 68000 system, then >the FAST mode would cut the time by 12% (and getting a 68010 would bring the >saving to 17%!). Let's say the picture would normally take 2 hours to draw. >You could save 15 minutes! > > Another good use for it is in LZHing / UN LZHing / Warping / unwarping, >etc.. large files. If you don't mind waiting until it's done to see >anything, >this hack will save you time. > > Ultimately, what can you expect from a $2.50 accellerator? I think it's a >bargain. > > Oh, if anyone wants specific pin outs / schematic, just ask and I'll upload >them. It's very simple. Just send the output of the clock oscillator to the external clock input pin, and tie the external_clock_enable to the ground pin specifically for the external_clock_enable. ------------------------------------------------------------- / Marc Barrett -MB- | BITNET: XGR39@ISUVAX.BITNET / / ISU COM S Student | Internet: XGR39@CCVAX.IASTATE.EDU / ------------------------------------------------------------ \ The great thing about standards is that / \ there are so many of them to choose from. / -------------------------------------------------------
merlin@presto.UUCP (Jeff W. Hyche) (06/11/91)
In article <1991Jun6.104357.12137@news.iastate.edu> taab5@isuvax.iastate.edu (Marc Barrett) writes: > > The next time I get around to it, I will try the hack with the multisync >monitor and report my results. BTW, this hack, when used with the 40Mhz >clock oscillator, can produce a 40% speedup clear across the board on >everything from the blitter to the sound chip to the CPU. This is because, >unlike all other acclerator boards or speedup hacks, this speeds up the >custom chips as well as the CPU. > What would this due to the custom chips in terms of timing and heat? I wouldn't even think of doing this to my Amiga. -- // Jeff Hyche There can be only one! \\ // Usenet: hychejw@infonode.ingr.com \X/ UUCP: ...!uunet!sci34hub!presto!merlin
jmt@legend.cma.fr (Jean-Marc Tanzi) (06/17/91)
In article <1991Jun6.104357.12137@news.iastate.edu> taab5@isuvax.iastate.edu (Marc Barrett) writes: > > The next time I get around to it, I will try the hack with the multisync >monitor and report my results. BTW, this hack, when used with the 40Mhz >clock oscillator, can produce a 40% speedup clear across the board on >everything from the blitter to the sound chip to the CPU. This is because, >unlike all other acclerator boards or speedup hacks, this speeds up the >custom chips as well as the CPU. > It's hard to believe it will work. The video is not the only part that has to get the proper timing to work. If the floppy interface gets its timing from the master clock (if not, I wonder how), you won't be able to exchange data with any other machine. That's also a real problem. -- ============================================================================= Jean-Marc Tanzi | Millions of years to make oil and forests, jmt@cma.cma.fr | a century to burn them. Mankind versus Nature: 100000 to 1. | Will Mankind find a tougher planetary opponent?
billsey@nesbbx.UUCP (Bill Seymour) (06/19/91)
In article <11816@mirsa.inria.fr>, Jean-Marc Tanzi writes: > It's hard to believe it will work. The video is not the only part that has to > get the proper timing to work. If the floppy interface gets its timing from > the master clock (if not, I wonder how), you won't be able to exchange data > with any other machine. That's also a real problem. I did this hack (32MHz) to my 2000 several years ago... I installed the Oscillator and a switch onto an old VI2000 board in the video socket. It worked fine except for serial, disk and video. :-} The video can be fudged by either the use of a multisync monitor or via a *big* tweak to the vertical hold. Stay away from the serial port and only write to HDs and ram: and you should be OK. Not that I was even able to switch the sucker on the fly, meaning I could do floppy type stuff to get things started, then switch to the faster speed for renders and such, then back again for floppy saves. Not really all that useful, but a fun hack. :-} > ============================================================================= > Jean-Marc Tanzi | Millions of years to make oil and forests, > jmt@cma.cma.fr | a century to burn them. Mankind versus Nature: 100000 to 1. > | Will Mankind find a tougher planetary opponent? -Bill Seymour nesbbx!billsey@agora.uucp or nesbbx!billsey@agora.rain.com ***** American People/Link Amiga Zone Hardware Specialist NES*BILL ***** Bejed, Inc. NES, Inc. NAG BBS NES BBX BBS Home Sometimes (503)281-8153 (503)246-9311 (503)656-7393 (503)640-9337 (503) 640-0842