[OT] Future HW support for Linux?

Roger Larsson (roger.larsson@skelleftea.mail.telia.com)
Thu, 30 Dec 1999 00:45:41 +0100


Hi all,

_INTRODUCTION_

For several years the PC HW and Windows has walked hand
in hand.

Improvements in the HW has been aimed to make Windows work faster.
Most notably 3D boards that has Direct 3D more or less in HW.

At the same time other HW stuff has moved very slowly.

As an example:
My computer, a PentiumPro (soon to be retired)
have
an USB header on the MB.
ISA
PCI
EIDE/33
does not have
AGP
I2O

Why has AGP moved faster than USB? Why has new computers been delivered
with AGP graphics but IDE Floppy and PS/2 keyboard + mice?
- Performance reasons (for games, but Voodoo3 does a nice job on PCI)
- Support in Windows? (Win95 DirectX upgrades, NT?)
- Existing alternatives (PS/2 ports are there...)
- Price (for interface circuit)

_BRIDGE_

With a open source OS a HW vendor can make any HW improvement as long
as they like to add their drivers to the kernel source and it give
big enough improvements to be worthwhile.

Note: that you get the same OS for the same price.
Not game/workstation/server with different price tags.

_QUESTION_

So, lets ignore Windows for a while.

What HW improvements could give big enough improvements (performance
and/or usability) to make a difference?

Some examples (performance):
- Page zeroing
- improved RT clock (request next event time,
fast read (memory mapped?))
- faster context switch (swapable execution contexts)
- faster kernel entry/leave
- dedicated distributed processors (like I2O, but suppose you put
something like an ARM in your chip set to run low level device driver
code, and more...)
- . . .

(usability)
- Power management
- external periphal connect solutions (I hate this kind of spaghetti!)
- . . .

-- 

The Internet interprets Windows as damage, and routes around it.

Roger Larsson Skellefteċ Sweden

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/