|
Home > Archive > Starsiege Tribes > July 2005 > PCI-E (was: Miracle's wish list)
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
PCI-E (was: Miracle's wish list)
|
|
| ScratchMonkey 2005-07-23, 8:33 pm |
| Randy Graham <ragmanx@spamex.com> wrote in
news:n0e2e11f29s4p8n9an9cnifpvlrqcbin9e@4ax.com:
quote:
> Yeah, but PCI-E really has nothing to do with PCI, except the
> ridiculous use of the same first 3 letters identifying it.
> Significantly different slots, electrically and physically.
It still moves data to and from I/O space, doesn't it? I don't really care
how that happens, whether it's USB, Firewire, PCI-E, or some other
connection system, as long as bandwidth and latency are reasonable for the
application.
So how does PCI-E differ from good ol' PCI?
I know that PCI in its basic form is an auto-configuring multi-master 32-
bit parallel bus with a 33 MHz clock, giving it 132 Mbytes of bandwidth.
There's a per-slot select line to select the configuration registers and
determine the resource needs of that slot, used only at boot time to
allocate address space and interrupts. Interrupts are shared.
| |
| Randy Graham 2005-07-24, 8:54 pm |
| On Sat, 23 Jul 2005 09:34:03 -0500, I heard the following crap spew
forth from ScratchMonkey <ScratchMonkey.blacklist@sewingwitch.com>:
quote:
>Randy Graham <ragmanx@spamex.com> wrote in
>news:n0e2e11f29s4p8n9an9cnifpvlrqcbin9e@4ax.com:
>
>
>It still moves data to and from I/O space, doesn't it? I don't really care
>how that happens, whether it's USB, Firewire, PCI-E, or some other
>connection system, as long as bandwidth and latency are reasonable for the
>application.
>
>So how does PCI-E differ from good ol' PCI?
>
>I know that PCI in its basic form is an auto-configuring multi-master 32-
>bit parallel bus with a 33 MHz clock, giving it 132 Mbytes of bandwidth.
>There's a per-slot select line to select the configuration registers and
>determine the resource needs of that slot, used only at boot time to
>allocate address space and interrupts. Interrupts are shared.
I had to read up on it to know the difference. The one thing that
really sticks out from my reading is that PCI-E is very high speed
(relative to old PCI) and serial instead of parallel. Also, the
connector is like 3/4 inches long (somewhere around 2 cm, for those
that do that kind of thinking), compared to, what, 3 inches, for PCI.
Also, the PCI-E bus can be split so, say, one card gets 16 of 20
"lanes" for throughput (not really lanes, since it's serial, but
effectively 80% of available PCI-E bandwidth), while another might
require 2X, and a third and fourth card can each work at 1X speeds.
Or, you could have 2 8X boards, and a 4X. There are multiple ways to
allocate the total available throughput.
In the past, I had a friend tell me that PCI actually required
electrical reflection (forgot the actual name, but that's apparently
the effect) to function properly, so early PCI builders that tried to
get some sort of terminating characteristics built in to the boards
for cleaner signalling actually didn't work with other PCI boards. No
idea if that's true, but thought I'd pass along that little bit o'
info.
ragManX
http://www.gamepatches.info/ - Are you up to date?
| |
| ScratchMonkey 2005-07-25, 9:10 pm |
| Randy Graham <ragmanx@spamex.com> wrote in
news:o6n7e19maichg0sffm0ogkme3mcqidn3ei@4ax.com:
quote:
> In the past, I had a friend tell me that PCI actually required
> electrical reflection (forgot the actual name, but that's apparently
> the effect) to function properly
That's it. Think of waves in a pond hitting the hard walls and creating
reflections. In most circuitry the reflections look like signal changes and
so are undesirable. Anywhere signal takes a sharp corner, including
connectors, causes some reflections. The real waves are reinforced or
canceled by the reflections, degrading the signal quality.
quote:
> so early PCI builders that tried to get some sort of terminating
> characteristics built in to the boards for cleaner signalling actually
> didn't work with other PCI boards.
A terminator is like a piece of cloth hung in the pool to soak up
reflections, or like the curtains in movie theaters to damp sound
reflections on the walls. An ideal terminator exactly soaks up all the
energy.
| |
| Smeghead 2005-07-25, 9:10 pm |
| On Mon, 25 Jul 2005 12:21:28 -0500, ScratchMonkey
<ScratchMonkey.blacklist@sewingwitch.com> wrote:
quote:
>Randy Graham <ragmanx@spamex.com> wrote in
>news:o6n7e19maichg0sffm0ogkme3mcqidn3ei@4ax.com:
>
>
>That's it. Think of waves in a pond hitting the hard walls and creating
>reflections. In most circuitry the reflections look like signal changes and
>so are undesirable. Anywhere signal takes a sharp corner, including
>connectors, causes some reflections. The real waves are reinforced or
>canceled by the reflections, degrading the signal quality.
>
>
>A terminator is like a piece of cloth hung in the pool to soak up
>reflections, or like the curtains in movie theaters to damp sound
>reflections on the walls. An ideal terminator exactly soaks up all the
>energy.
Now, O Guru, tell us of fiber optic attenuation and the methods to
reduce it.
--
--==< S m e g h e a d >==--
"Yes, I like my coffee hot and strong...
Like I like my women: Hot and strong,
with a spoon in them" -- E.Izzard
| |
| ScratchMonkey 2005-07-25, 9:10 pm |
| "Smeghead" <tribesfan@hotmail.com> wrote in
news:hnaae19l6sfts2l7vhpo83enc15ab9jvu3@4ax.com:
quote:
> Now, O Guru, tell us of fiber optic attenuation and the methods to
> reduce it.
I must plead that I'm in the dark on that one.
| |
| Smeghead 2005-07-25, 9:10 pm |
| On Mon, 25 Jul 2005 13:58:09 -0500, ScratchMonkey
<ScratchMonkey.blacklist@sewingwitch.com> wrote:
quote:
>"Smeghead" <tribesfan@hotmail.com> wrote in
>news:hnaae19l6sfts2l7vhpo83enc15ab9jvu3@4ax.com:
>
>
>I must plead that I'm in the dark on that one.
HA! Fiber Optics... Attenuation... DARK! Good one, mate!
I salute you, Scratch!
--
--==< S m e g h e a d >==--
|
| |
|
|