home *** CD-ROM | disk | FTP | other *** search
- Path: FreeNet.Carleton.CA!de351
- From: de351@FreeNet.Carleton.CA (K. C. Lee)
- Newsgroups: comp.sys.amiga.misc
- Subject: Re: Msg from Carl Sassenrath (VISCorp)
- Date: 20 Apr 1996 22:18:14 GMT
- Organization: The National Capital FreeNet
- Sender: de351@freenet3.carleton.ca (K. C. Lee)
- Message-ID: <4lbnr6$jmv@freenet-news.carleton.ca>
- References: <1996Apr18.203528.12163@scala.scala.com> <4lan52INNchm@maz4.sma.ch>
- Reply-To: de351@FreeNet.Carleton.CA (K. C. Lee)
- NNTP-Posting-Host: freenet3.carleton.ca
-
- Valerio Ortelli (orv@sma.ch) writes:
- > In article 12163@scala.scala.com, dave.haynie@scala.com (Dave Haynie) writes:
- >> The whole AA chip set is
- >> about 200,000 transistors (three chips), AAA was just under a million
- >> (four chips). An aggressive graphics and sound chip of today will get
- >> several million transistors on a single part. They take 2-4 years to
- >
- > More is better ?
-
- More is not necessarily better. On the other hand, it tell you how
- complex a design is. When you have 64-bit blitter (vs 16-bit), ALU, better
- DRAM/VRAM interface, more functionalities, then it would be likely to use
- more transistor.
-
- Going by your argument, a 6502 is better than a 133MHz Pentium ?
-
- > So, why RISC ?
- > And, what mean RISC if a RISC has more transistors ?
-
- CISC & RISC was used to classify the CPU instruction sets. RISC takes the
- approach that doing less stuff per instruction to reduce the overall
- complexity. By having a lower complexity, they can run the chip faster.
- These days it is a bit difficult to define the two as CISC chips uses a
- lot of the RISC tricks.
-
- If a RISC chip has more transistors, it might suggest that more of the
- transistors are used in a way to run more instructions in the same cycle
- or to reduce the chance of doing silly things that slows down the chip.
- >
- > But ... if I want "just these few transistors in that way to do that thing
- > in that contest" am I wrong ?
-
- Yes ! The original OCS was so much optimized that it was difficult to
- improve on it without a lot of work.
-
- > Dont say AGA is out of date, you may do something similar up-to-date
- >(if you can).
-
- It will take 2-3 years before you get a new generation of chips. By then
- your new chipset would only last about 6-9 months before new ones are
- needed to beat the other guys. They don't have a full team (say 30-40
- analog/digital/VLSI engineers) and spend $15 millions dollars a year for R&D
- alone for nothing.
-
- > Maybe is more convenient (if it is) not to touch your architecture and OS
- > and continue to go your way.
-
- You mean a slow death ? What's wrong about setting up an OS standard like
- RTG that allow the hardware to continue to develope while having the same
- OS software interface ? You simply raise the architecture to the system
- software level. You don't complain that Motorola changes the microcode,
- do you ? It is simply pounding on the coders' head that they should
- programmed to OS's perfered ways of doing things.
-
- > Just be careful, if your eyes dont see the missing of the copper or some other
- > AMIGA-only feature dont thing that this feature is not present. It will maybe come
-
- What about seeing things that you don't have on the Amiga before ? This
- has already happened on high ended 3rd party Amiga graphics cards that
- uses SVGA accelerator chips. An external copper can be easily added to anay
- SVGA accelerator chips if you know enough about hardware and FPGA.
-
- > orv
-
- K. C. Lee
-