1474 [From ] Dan Mccoy [REC'D] [MSG 1474 OF 1483] [To ] Jeff Robertson [Reply To 1473] [Date ] Sat 12 Dec 92 15:13 [Subject ] Re: SCSI II: Commodore's implementation. Along the same lines, here's some text grabbed from Portal that SteveX posted in regards to what he heard that the Toronto WOC. It really does appear this new guy at the head of CBM's engineering department is moving Commodore ahead in development! Only hope I. Gould allows this all to continue! Here goes... -- Beginning of text -- I said I'd write a bit about Lewis Eggebrecht's talk at WOC, so here it is. Lew started out by introducing himself. Lew is the VP of Engineering at Commodore, and he started there to fix up their DOS machines, and then jumped over to the Amiga. He spent many years in the PClone market, including being the leader of the original IBM PC design team. He also worked for Franklin making Apple II clone parts. He then talked about Commodore, going through their engineering department. He listed many engineering groups, VLSI being the largest of them at 23 people, and then Software Engineering. If you include CATS in with Software Engineering, then that the software group is larger than VLSI. Lew stressed the changes he was working on in design methodologies within Commodore Engineering. The way he described this made a lot of sense, and it sounds like he is working at getting Commodore in a position to keep churning out state-of-the-art technology. He said he wanted to be one of the first with accelerators taking advantage of new CPU's, instead of one of the last. He then moved on to talk about the new chips Commodore is working on. Much of this has already been discussed by the various magazines covering the WOC/Pasadena show, so I will only mention things here that I didn't remember hearing back then. There are two new chipsets in the works. A "low end" chipset, and a "high end" chipset - though Lew actually said the high end chipset is farther along in development right now. The low end chipset is basically AGA cleaned up. It doubles the bus bandwidth again, giving us 8x the original chipset bandwidth (and double the bandwidth of AA), and making some new modes available to take advantage of the extra bandwidth. The pixel clock goes up to 57mhz, so we get to do basically double the horizontal or vertical size of the old modes, or we can scandouble the new AA 72hz modes. In other words - finally, a usable non-flickering 800x600 mode. Also mentioned for the low end chips was 16 and 24 bit true colour video, 8 bit chunky pixel video modes (Yeah!), double the blitter speed, a 4 byte serial FIFO (for both input and output), 8 meg chip ram address space, and faster floppy support (so the chips can handle high density floppies without having to slow down the drive, and probably do 2.88 meg or more floppies). That's the low end chipset. All this will be packed onto 2 chips (instead of the current 3), and a lot of the extra glue chips will be combined as well. This chipset will only work in 32 bit machines - no more 68000 Amigas from the looks of things. The high end chips are where things get interesting. The low end chips contain a total of about 110k transistors, but the high end crams 750k transistors into 4 chips. (The 68000 processor, by comparison, has 68000 transistors in it, while the 68040 has somewhere around a million). The high end chips will use either 32 or 64 bit VRAM, and give you from 12x to 20x the bandwidth of ECS. The pixel clock goes up to 114 MHz, where we can get modes like 1024x1024 with a 72hz refresh rate. This is in 24 bit. Lew started listing off some amazing things here, and I'm just going to mention them here - he didn't give many details. A direct high-speed serial interface is going into the chips, for use with CD-ROM drives. I don't quite understand this myself... Video capture will also be possible. 16 bit audio at up to 100KHz. 8 audio channels. A 32 bit "bit blitter" (as opposed to the "word blitter" we have now). Hardware video decompression. A Processor independant bus (so they are "Risc Ready" (his words)). This is something he mentioned that surprised a lot of people: A modular video chip design that will let you add more chips to increase the capabilities of the chips. You will be able to add more chips and get more display bandwidth - his example was increasing from 1024x1024 to 2048x2048. You can also add more blitters this way. There is a new "Demand DMA" system for this chipset that schedules DMA rather than using a fixed allocation. This whole chipset is asynchronous, letting you separate the pixel clock from the CPU clock from (one other clock he mentioned that I forgot). Generally amazing stuff, and it's probably early to be depending on any of it, but it does sound like they have some amazing stuff in the works. Next he talked about some of their future plans. Among them is moving the entire product line over to the AGA chips, "and an AGA CDTV would be obvious". He didn't give many details. He did say they would design future systems with modular upgrade paths - but he made it clear that none of the future designs would be compatible at the chip level with previous systems. He did say a DSP board for the 4000 would be available in the spring, and that it would be a replacement CPU module. You will have the DSP on the CPU module, and then you will have "Personality Modules" which will be a hardware/software combo that will let you use it for specific applications - for example, some phone jacks and software would let you use it as a V.32 modem. You can also do very good voice synthesis, speech recognition, and other things with the DSP that the 68040 can't handle very well. SCSI-II for the 4000 should be available mid-January. He said on the floor after the talk that a 4000T (tower) would be available in mid-February but I didn't hear anything more of that. Lastly, I asked whether future chipsets would do scandoubling like AGA does it (which is, by doubling the bandwidth required and fetching the entire scanline twice) and he said that they would not. He also said they will have a product announcement soon for 4000 owners involving this. He didn't give any details - but it certainly sounds good! So there you have it. The future looks bright. If you want to post this somewhere, you have my permission, but leave my signature attached. --------------------------------------------------------------------------- Steve Tibbett:Portal=SteveX:Usenet=SteveX@cup.portal.com:BBS=613-731-3419 --------------------------------------------------------------------------- -- End of text -- All disclaimers apply. (*) Dan (*) (E,R,O,C,A,P,N,L,F,<,=,!,I,J,.,-,U,^,M,G,H,?,RET,X,W,Q) Area: [3] [Amiga WorkBench] [1474/1334-1483] [43 min] =>