[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
PCI market trend info, png's take
- To: Mailing List Recipients <pci-sig-request@znyx.com>
- Subject: PCI market trend info, png's take
- From: png@woof.net (Peter N. Glaskowsky)
- Date: Tue, 12 Nov 1996 18:00:58 -0800
- Resent-Date: Tue, 12 Nov 1996 18:00:58 -0800
- Resent-From: pci-sig-request@znyx.com
- Resent-Message-Id: <"uCWKA.0.nY1.BlIYo"@dart>
- Resent-Sender: pci-sig-request@znyx.com
> Can anyone point me to some information on what
> the PCI market trend looks like a year from now.
> Are most of the PCI components expected to be
> 32-bit/33MHz, 64-bit/33MHz or 64-bit/66MHz? A
> mix bag or what? I realize this information may
> or may not be generally available, so any thoughts
> or ideas as to where the PCI market is heading
> will be appreciated.
I'll take a quick stab at this, although I could assemble more specific
prognostications given more time. I think that 5V, 32-bit, 33-MHz PCI will
continue to be the most common version for conventional expansion cards for
at least three more years. There's just very little reason to make an
Ethernet card or a hard disk adapter run any faster than that, or on a
lower voltage. PCs generally have only a limited ability to manage
high-bandwidth devices anyway, with one conspicuous exception:
Graphics chips and cards for PCs will migrate very quickly-- wide
availability by this time next year-- to 3.3V, 32-bit, 66-MHz PCI and/or
AGP. 2D and 3D are already using up all of the PCI bus; AGP provides a
relief valve, giving graphics cards somewhere else to go.
Notebooks will be the other market for 3.3V PCI, both in motherboard
devices and in CardBus devices, but only in the 32-bit, 33-MHz variety.
64-bit, 33-MHz PCI may only be of interest to servers and workstations.
64-bit 66-MHz PCI will wait a few years to give AGP a mid-life kicker, but
today's 66-MHz variety of PCI won't be able to replace 33-MHz PCI for
expansion cards due to electrical loading and therefore slot-count
limitations.
I think we'll need another extension to PCI to solve that problem-- a new
bus, probably very similar to today's PCI, that supports 66- to 133-MHz
operation with multiple devices, at least four slots. Such a bus could just
be a new electrical interface, probably based on the same stub
series-terminated logic (SSTL) interface that next-generation SDRAMs will
use. It could also have new and better protocols. I'm just guessing here,
though.
. png
Peter N. Glaskowsky
Senior Analyst
MicroDesign Resources
Publishers of the Microprocessor Report and Technology Roadmap
Hosts of the Microprocessor Forum and PC Tech Forum
408-328-3934
Œ {