>So long as decent audio is on the board and ethernet PCI cards aren't
>incredibly expensive, go ahead, kill the ISA. If person X wants to add card
>Y, where is he going to put it? By everything I mean the basics that
>everyone needs. Serial, parallel, video, SCSI. There is going to be
>a lot of discussion about _what_ video and if SCSI is needed (there were
>the last time). There will be lots of shouts for Ethernet on the board (I
Well video is handled by the CPU so the only things to argue about here
are the amount of VRAM and choice of RAMDAC. As for SCSI, Drew Eckhart
is in favour of a rather fast NCR PCI SCSI controller for which he has
written a driver. I see no reason for not going along with this unless
the NCR chip is particularly expensive.
>need it for one) or on a cheap enough PCI card. Audio tastes vary wildly too.
>Some only want 8kHz bleepers, others (me included) want 44.1kHz multi channel
>boards. I think we need to define the minimum configuration board that
>*everyone* (80% or so) wants to see. And the rest will have to go on cards.
Here we go again, design by committee. Wouldn't It Be Nice IF...
>Otherwise there will be continual arguments "drop <xx>, I don't want it and
>it makes it too expensive".
True, but it's an all or nothing situation. If you want all the features and
performance of a workstation then you're very unlikely to end up with a
price tag of less than $1000. Something has to go. Remember that at one stage
we were talking about designing a RISC based accelerator to plug into a
PC ISA slot - no features, just CPU and memory.
>Thats one hell of an assumption pal. Look at what I wrote and read it this
>time. I certainly wouldn't want an uncached board with the processor
>continually waiting for memory. But some will balk at the cost of cache chips
>(I'm assuming a decent sized/fast cache. ~0.25MB to 1MB) with board. They'll
>want to buy the motherboard without cache and bung in the chips as they can
>afford to. Amazing concept huh? Thinking about what someone else may want ;-)
OK, let's go with your 80% straw poll. Anyone not wanting second level cache?
I thought not. Cache memories are reasonably cheap so penny pinching in this
area is not a good idea. Also, designing the board to allow the cache (or
any other feature for that matter) to be optional will probably require a
modest amount of extra hardware. Allowing such options will require
that parts of the board are socketed - increasing production cost and reducing
"I spend in her hand, and spew in her lap;" --- J. Wilmot