home *** CD-ROM | disk | FTP | other *** search
- Path: comma.rhein.de!serpens!not-for-mail
- From: mlelstv@serpens.rhein.de (Michael van Elst)
- Newsgroups: comp.sys.amiga.programmer
- Subject: Re: CHIP RAM speed test resul
- Date: 28 Mar 1996 11:18:59 +0100
- Organization: dis-
- Message-ID: <4jdp2j$t1a@serpens.rhein.de>
- References: <4j6jv0$1im@serpens.rhein.de> <5827.6659T112T770@mbox.vol.it>
- NNTP-Posting-Host: serpens.rhein.de
-
- bizzetti@mbox.vol.it (Fabio Bizzetti) writes:
-
- >It would be sufficient to modify the interface CPU <->chipram<-> Alice.
-
- You say as if this were easy.
-
- >>Yes. That's why "dual ported RAM" is nonsense.
-
- >Come on.. I didn't mean to simply change the chips!!!!
- >There must be a circuit to exploit the new chips' nature, and getting a fast
- >SIMM and multiplexing data is perfectly fine, allowing practical dual access,
- >but (perhaps) could be more complicated at the end than real dual access ram
- >(ofcourse with the circuit to use it).
-
- I am just talking about that. Real "dual ported RAMs" just exist in tiny
- sizes, they are used for communication between several CPUs or as
- register banks. Then there are chips that are just multiplexed RAMs with
- the multiplexer inside. Still small and still very expensive and
- effectively the same as standard RAM on a multiplexed bus. And then
- there is the variety of VRAM that is just usuable for special applications.
- It still is multiplexed but one side uses a very wide bus so that
- multiplexing overhead is neglible if you can use the ~1024bit word.
-
- So, except for the true dual ported RAMs there is no real advantage for
- our application. What you really want is just a RAM controller that is
- as fast as todays RAM chips.
-
- >Maybe the "ram controller" that you wish (to multiplex data) would be more
- >expensive than 2Mb of real dual access ram,
-
- The RAM controller is called Alice. You would need a version that runs
- faster, at least on the memory bus. And no, 2kbyte of real dual ported
- RAM is more expensive than the current Alice. These chips are made of
- fast static RAM. Think about building main memory of cache RAMs and then
- double the price.
-
- >Everything is good to improve performances, and AGA has already a very good
- >22Mb/sec bandwidth to RGB, giving another 22mb/sec CPU-chipram would be *nice*.
-
- Oh, a state-of-the-art video controller that reads 200MByte/s and leaves
- 60MByte/s for the CPU is even nicer.
-
- >> We already have "dual port RAM", just the multiplexing is done outside the
- >> chip.
-
- >This is the problem: it's performed by the 7Mhz (bus clock) of Alice.
-
- Yes, that's why you want a much faster graphics controller.
-
- >AGA can be improved a lot though,
-
- Hah. Could you afford to improve AGA ? The fact that something is
- theoretically possible doesn't mean that it is feasible.
-
- >> They need geometry engines,
-
- >A skilled assembly programmer can avoid multiplications/divisions and all
- >these things that normal programmers learn from standard books.
-
- You mean those books that show how to write efficient rendering routines
- too ?
-
- >It would mean to blow the PC away.
-
- It just means that you know more about c0d1ng than about hardware.
-
- >They cost a lot, the PCI GfxBoard isn't flexible (it gains all from Pentium's
- >MIPS, that are quite more than Walkers' ones)
-
- The PCI board is much more flexible _because_ it gains all from the
- Pentium's speed. With a custom chipset like AGA you are more or less
- stuck with 5-10 year old technology.
-
-
- --
- Michael van Elst
-
- Internet: mlelstv@serpens.rhein.de
- "A potential Snark may lurk in every tree."
-