1994 by GEnie ========================================================================== This file is brought to you by The Commodore 64/128 RoundTable on GEnie This file may be published or excerpted in User Group newsletters providing credit is given in this manner: "Copyright 1994 by GEnie From the Commodore 64/128 RoundTable File#:#####" This file maybe be distributed, if distributed whole and unaltered, on non-profit BBSs or non-profit networks. For more information on GEnie call by modem: 1-800-638-8369 (8-N-1 300/1200/2400) Enter: HHH Then reply: xtx99018,commrt Then enter: Commodore And enjoy! ========================================================================== Welcome to the CMD News Release night!!! The news release will be followed by a question and answer segment. Doug Cotton from CMD is our featured speaker for tonight. Hi Doug Hi! ;) We know you are have some great news for us C= users tonight. So let us get right to it. Thanks for coming all... and yes, I do. ;) CMD is pleased to announce that we will soon be releasing a new hardware product. This product is initially aimed at the C64 or C128 in 64 mode. The product will be released early next year, with a target of February 1996. The product is actually two products in a single product line. The two models in this new line are named the Super64/10 and Super64/20, and are 10 and 20 MHz accelerators, respectively. Prices are expected to be approximately $150 for the /10 model, and $200 for the /20 model. There are a lot of features and such, but I suspect we can cover those as we get into the Q & A. ;) We will be going into a little 3 minute "whooping time" followed by a listen period 20 Meg!!!! Yeeeeehaw!!!!! That is great news, Doug How did you get around the VIC chip limitations? <128PowerUser> This is great Doug, but not of much use for 128 natve mode users. A20MHz 64? CCOOLL!!! Wow!!!!!!! 20 Mhz! Think of GEOS! Think of...WOW! will we have to buy a whole new c64? or is just a chip? Okay, first question- hi doug, 2 parts same question,comptability with Ramlink,Swiftlink? Absolutely yes on both, Greg. ;) Let me explain that a little... The Super64CPU's (that's what we'll refer to them as) will have a pass-through port, much as RAMLink does now. They will also be RAMLink "aware" devices, in that the RAMLink Kernal will also be in them, so that RAMLink can be attached and controlled by them just as if it were doing things all by itself. Other cartridges can also work, But this will be limited to things like I/O cartridges, as cartridges which try to do memory mapping of the computer and such will likely fail. So SwiftLink, yes, but things like Super Snapshot... sorry, no. I guess you will answer it but why not 128 in 128 mode. Cant acceleration help us out also? Sure it can, but the reason for 64 mode first is simple, we can have it ready sooner. The 128 mode will take some extra stuff as well as extra time to develop. What extra stuff? How long till 128? We haven't made any definate decision to go ahead with a 128 mode model as yet, but it is under consideration, and will likely be approved to be done after we release the 64 mode models. As for the extra stuff, It takes extra logic to deal with the MMU and other unusual functions of the 128. This means more hardware inside, and extra time to figure out just how to make it all work correctly. That adds cost, which we felt should NOT be passed on to 64 users. Hi Doug- how would you rate a 20MHz 64 vs an IBM of what class/MHz? Okay... we're using the W65C816S processor, which is rated at 10 MIPS when running at 20 MHz. A 386SX processor running at 25 MHz comes out around 8 MIPS, so a little better than a 25 MHz 386 in raw processing power. Now, that the raw numbers. but there are some other things to consider. There's an enormous amount of bloat in "Intel" machine programs, and they simply aren't written as efficiently as 6502 code. My bet is that our programs will be able to run as fast as most any 486 machine normally sees in applications. Yah, I'm familiar with numbers comparisons, my Amiga's Motorola vs Intel's hogs. Thanks. Lou Schonder, the famous Wednesday night host in the Commodore RT has a question Doug, would one have access to the extended commands of the new CPU, and Can it work on an SX-64? (Can it POSSIBLY fit inside an SX-64?) As for the commands, yes. The 65816 commands are all available in both native and emulation modes (though many have less use in 8- bit vs. 16-bit mode). As for the SX, it should be able to work, but much will depend on the strength of signals and noise on the expansion port of the individual computer n this case. I'll have to check more on that with Mark. ;) Let me verbalize a moment before the next question... If a 128 version is released at a later point, we will offer an upgrade policy. I've been told this is official. There will also be some kind of upgrade policy for the /10 to /20 model. What will happen to the SIMMs we have in our RAMLINKS now when we put a 64 accelerator ahead of it? Can we still use the programs installed in the RL? :) Leave those SIMMs right there! Yes, as I mentioned earlier, RAMLink can be attached to the pass-through port of the Super64CPU's, so it can still be used in exactly the same way as you use it now. Greg Noggle, our famous Sunday night Commodore RT host has a question. I assume there is some extra memory being added in this addon,if so how much and can we access it,and what is the projected cost? Ah. Let me get into some of the specs of the system. The on-board RAM will consist of 64K of Fast Static RAM (25 ns, I believe). This is as fast as they make it, at this point, and it has to be this fast just to keep up with the processor. Now, I know that some of you are saying, "Then why do 66 MHz PC's use 60 ns RAM?" Here's the scoop - The processors used in Intel and other "fast" PC's have methods of pre-fetching data (long before it's actually needed), so they get away with using slower RAM. We don't have this luxury, so we've got to use the fastest RAM available - leading edge stuff. But this means we'll get better overall performance (no wait states on the RAM). The down-side is that this kind of RAM is very expensive, say somewhere around $150 a MB, so don't look for us to put extra FAST RAM in the unit. It would just raise the cost. However... I've been hearing talk about something called a "Rocket-Socket" to be incorporated into the design of the Super64CPU's. A little bird told me that this little jewel would probably let us add as much RAM as we desire, up to 16 MB (or perhaps more if a virtual memory scheme were used). Now, I would expect that DRAM would be used in this case, though, with some wait-states to keep the cost down. Will there be a need for something to speed up our external devices (drives) to maximize the Super64 usage? Best thing I could suggest is that you use the fatest peripherals you can afford. Naturally RAMLink and the HD top that list. And as long as we're talking about peripherals, I should also mention that the Super64CPU's are DMA aware, and are therefore compatible with Commodore REU's (in addition to being RAMLink and GEORAM compatible) price? Again, the projected prices at this time are $149.95 for the /10 and $199.95 for the /20 model. geoHarold, another famous Commodore RT host on Friday nights, has a question. A couple of questions. What about compatibility with devices that go into the modem port, such as the Handyscanner 64, CBM/Aprotek modems and RS-232 adaptors? That will depend entirely on their software Harold, as it is in control of those devices. There's no hardware conflict. I'd expect programs that do "software RS-232" not to work in accelerated modes, though. Have you experienced problems in getting software to work with the Super64? Such as GEOS, word processing, spreadsheets and games? And utilities? Software compatibility will come down to an item by item determination. Games will be nearly useless in accelerated modes though, can you imagine Pac-man at 20 MHz? Bet you couldn't say "game over" before it was. But most productivity software, and especially GEOS, will benefit. Utilities, that covers a broad area, and I'd have to know specifically which you're concerned about to answer that. Is there a separate power source for the Super64? If I understand correctly, there will be an option for a power supply. It won't be necessary for using the unit alone or with a RAMLink. But you'd probably want one if you were using just an REU with the Super64CPU. Chance, one who has helped out in the RT many times, has a question. Are there any features of this new accelerator available for the programmer programming specifically for it? Any new tricks, or does it completely mimic the 6502/10? There are indeed a lot of tricks. The 65816 can mimic the 6502 in emulation mode. But in it's native mode, most of the regiters expand to 16 bits, and the address bus goes from 16 to 24 bits. In both modes, you'll have 24 addressing modes (as opposed to the 13 we have now) and a whole lot of additional opcodes. As for programming info, We'll be offering some developer packages, and we'll probably be extending one or more of the currently available Assembler packages to deal with all the extra's. ;) Snog Pitch, our famous Saturday night host has a question. I have a question with the Ramlink passthru. Will that be straight thru, or an elbow type of connection? Straight through. Also, we expect the Super64CPU module itself to be very small, at least in length, probably no more than 3 to 4 inches, maybe less. Keith Smith, editor of the Tonawanda Commodore User Groups Newsletter has a question. Okay, Doug, just a couple of quickies: First- is there a bypass method/switch? Yes. Let me cover the switches. There are two three-position switches in the current design. One is a speed switch, with SLOW, FAST, and PROGRAM modes. The SLOW and FAST modes FORCE the unit into 1 MHz or 10/20 MHz modes, respectively. The PROGRAM mode defaults to FAST mode, but allows software control, using the same method as employed in the Schnedler Turbo Master CPU (so programs modified for that accelerator will work with our unit). The second switch also has three positions. OFF/NORMAL/JIFFYDOS. OFF disables the unit entirely, and the other two could be viewed as being identical to the JiffyDOS switch we put on 64/128's equipped with JiffyDOS. Second, is CMD basing the probability of a 128 version on the sales of the 64 version? If so, I have zero use for a 64-only version. Actually, we'll base the probability of a 128 version on response received from users about that product. So, those interested in a 128 version should write directly to CMD and voice their opinion on that matter -- don't call, WRITE. ;) And lastly, unrelated to the SuperCPU, and my apologies if this has been asked to death, but has CMD been in touch with ESCOM re: The aquisition/re-production of 64s/128s/65s? ;) As for talking with ESCOM, we have done that, but we have not talked to them about aquiring any rights to produce the 64/128/65. We have no interest in doing that at this particular point in time, but should it be necessary, they're open to discussion of the matter. Maurine, editor of the Muskegaon User groups newsletter My question on switches has been nicely answered, but I would like to know what the Super64CPU looks like physically. It's not clear if it goes inside or out or both, is detachable, etc.? Good question. Right now it looks like a bunch of chips on a board! But the final unit will be in a case, and yes, will be fully external. Mark also tells me that there will be no need to open your computer at all to attach it. Great! One more question, Some people in my users group were wondering, does CMD have any emulator production plans? None. Our market has always been the 64 and 128, and we don't expect that to change any time soon. We can sell an emulator, and it may very well be our last sale to that person. We'd rather sell to those staying in the market. How did you get around the 1mHz limitation of the VIC chip? Didn't, and you can't. Even Commodore had to slow down to 1 MHz for VIC access in the 128. So here's how it works, Just like the 128. We do something called clock-stretching, that allows us to sync with the 1 MHz clock when data has to be copied down into I/O, or screen memory. So will all raster interrupts and other tricks be out of the question? No, not out of the question. But they'll have to be written specifically for the accelerator if that's the intended purpose. Many of those routines base things on timing of a cycle in the computer, and that's out. Can't have things working that way. This was the final question. Direct any other questions to the Bullitin Board in the CMD Category. I want to thank Doug for a very informative evening. Great job, Doug Thanks Tim, and thanks all for being here for the announcement. ;)