johnmac@fawlty.towers.oz (John MacLean) (09/28/90)
This may seem like a rather uninformed question: In what area (eg: RAM speed, ROM speed, BUS speed, Timing problems) is the actual restriction to just putting a new (faster) 65816 and new (faster) clock into the motherboard of the GS, letting the FPI (Fast Processor Interface) take care of slow downs? In this way you would just get a N Mhz / 1 Mhz machine instead of the 2.8 Mhz / 1 Mhz machine. Where 2.8 > N > 7. I believe that Transwarps come with 80ns RAMs and the average IIGS memory card comes with 100ns - 150ns RAMs. This could be a problem, but the 100ns RAMs could probably be accessed at 2 * 2.8 = 5.6 Mhz (just an uninformed guess). The IIGS motherboard probably cuts down a high rate clock to produce the 2.8 Mhz and the 1 Mhz signal from the same crystal. If this is true (again just a guess) then replacing the crystal is not a good idea; but how much of a hardware hack would be required to fix this? Someone recently mentioned that the speed of the ROMs was a significant factor when the GS was designed. Could the ROMs be easily copied into ROMs that had faster access capacity (available today) (ignoring copyright issues - this is a technical question)? Would there be any problems with a faster (eg: 2 * 2.8 = 5.6 Mhz) GS interfacing with the slow (1 Mhz) side of the system through the FPI? Note that a 2 * 2.8 = 5.6 Mhz GS would probably be as fast as a 7 - 10 Mhz Transwarped GS (because it would be a non-cached system). Would someone with a bit more knowledge of the hardware than myself like to answer / comment on / shoot holes in some of these issues. John MacLean. -- This net: johnmac@fawlty.towers.oz.au Phone: +61 2 427 2999 That net: uunet!fawlty.towers.oz.au!johnmac Fax: +61 2 427 7072 Snail: Tower Technology, Unit D 31-33 Sirius Rd, Home: +61 2 960 1453 Lane Cove, NSW 2066, Australia.
cs4w+@andrew.cmu.edu (Charles William Swiger) (09/28/90)
>In what area (eg: RAM speed, ROM speed, BUS speed, >Timing problems) is the actual restriction to just >putting a new (faster) 65816 and new (faster) clock >into the motherboard of the GS, letting the FPI (Fast >Processor Interface) take care of slow downs? Putting it simply, all of the above. OK...one of the big problems is that in order to speed up the //gs, you have to replace a lot of chips with faster ones in order to have anything work...just putting in a faster processor and crystal just won't cut it. You would have to replace all of the "fast" RAM with faster RAM (both on the motherboard and get a new memory card with faster chips), you would have to replace the current ROM chip with a faster ROM chip (which can be done, faster ROMs do exist), and you would have to replace most of the address decoding and data bus circuitry. A faster CPU won't do much of anything unless it has faster access to the computer's memory. This is not a home do-it-yourself modification of your computer; it requires a complete redesign of the motherboard. (No hardware hacks allowed ;-) One of the basic limitations of the current system is that all of the slots must run at 1 MHz for the cards to work. The video display also depends upon using 1 MHz RAM in order to keep the video timing on track. Changing the display to use faster memory would be a big help in speeding up the super-hi-res graphics modes. A redesign would also allow the addition of higher resolutions, which would be nice. The //gs uses a 16.7 MHz crystal for the basic system timing, I think, and each processor cycle takes 6 ticks of the crystal to complete. There is also an odd timing problem that forces the 65th processor cycle to be delayed. You're exactly right that having the motherboard sped up would be faster than using a transwarp, that, as you said, uses caching to run the first 2 banks at a faster speed than the rest of the system. <start humor> By the way, what's wrong with this picture: "In this way you would just get a N Mhz / 1 Mhz machine instead of the 2.8 Mhz / 1 Mhz machine. Where 2.8 > N > 7." <Hint: check your math> In response: "I've had too many tests this week: my brain is MELTING, and I CAN'T get UP!!!" <end humor> -- Charles William Swiger cs4w+@andrew.cmu.edu
toddpw@tybalt.caltech.edu (Todd P. Whitesel) (10/02/90)
cs4w+@andrew.cmu.edu (Charles William Swiger) writes: >The //gs uses a 16.7 MHz crystal for the basic system timing, I think, >and each processor cycle takes 6 ticks of the crystal to complete. Not quite; the master crystal is 28.63636 Mhz exactly (within tolerance) and each CPU cycle is 10 ticks of that clock. The 28 Mhz is needed for the new video modes. >There is also an odd timing problem that forces the 65th processor cycle >to be delayed. It is not a problem; in fact it was one of Woz's many hacks in the original ][ and it prevented the video from having different colors on even and odd lines of the video. The stretched cycle occurs at the end of each video line and is there to make each line an even multiple of the color reference (3.58 mhz or 280 ns). This delay is only present in the 1 Mhz side of the machine and it does not affect the CPU unless it is accessing the 1 Mhz side when the stretched cycle occurs. >You're exactly right that having the motherboard sped up would be faster >than using a transwarp, that, as you said, uses caching to run the first >2 banks at a faster speed than the rest of the system. Hmm? The transwarp is supposed to accelerate program code in any bank. Todd Whitesel toddpw @ tybalt.caltech.edu