PCI-e 1X In A PCI-e 16X

[dawman]

I am not quite sure how the bandwidth I currently use from my PCI-e 1X (2.56GB) has any effect on the XITE-1s performance, but I am a little too nervous to switch anything on my rig as I can't afford to take any chances until a new rebuild and weeks off.
I remember reading here that before the X58 chipset and Gigabyte UD series fixed their PCI-e 1X in the BIOS upgrade that guys were using a PCI-e 16X as a way to get around the Gigabyte bug, to no avail, but they still seemed to have the XITE-1 working.

My question is has anyone here used the 16X, and notice the extra Bandwidth as being helpful...?

[jksuperstar]

you shouldn't see any extra bandwidth. 16x means there's 16 lanes available. the xite can only use 1 lane, so the other 15 would sit idle. however, if that slot was talking directly to the processor and not through the chipset, you might gain some really solid low latency out of it

[Bud Weiser]

you shouldn't see any extra bandwidth. 16x means there's 16 lanes available. the xite can only use 1 lane, so the other 15 would sit idle. however, if that slot was talking directly to the processor and not through the chipset, you might gain some really solid low latency out of it

Assume, there are 2 PCIe X16 slots on the mobo and in the 1st of these, there´s the graphics card.
What´s up w/ IRQ sharing if the XITE-1 PCIe card will be inserted into the 2nd PCIe X16 slot ?
On my motherboard, both X16 slots share,- so I doubt it´s a good idea to use 1 of these for audio at all,- but I can be wrong w/ this assumption.

Or,- will the graphics card use different lanes than the audio PCIe card ?
Does a (1) PCIe X16 graphics card use all 16 lanes ?

Example:

PCIeX16 #1 (blue) = full speed,- insert graphics card here
PCIe X16 #2 (black) = AUTO, x4 and x2 Mode in BIOS

AUTO adjusts speed according to what´s being plugged in
x4 de-activates PCIe X1 slot
x2 always runs @PCI Express x2 speed

Now I wonder if that´s usable for the XITE-1 PCIe card too.

Bud

[dawman]

I'd be interested in hearing those details too.
I will be using the Z77 chipset that supposedly, like the RAM, has very short trace lines due to their integration into the CPU.
So I guess I won't see until I actually make the move.
So far I just upgraded with 4 new drives that are the Vertex 4s that have 90k IOps during random reads, which is badically what our sampled sounds used along with their targetted buffers in RAM.
Huge upgrade for the price, performance has more than doubled, but then again I use nothing but new developments where NCW lossless audio compression from Kontakt is used.
It's a leap above the pack like Giga ( Aria ) PLAY and VSL...But then again those are strictly for indoor use or pretend players in automated shows.

There's a really well programmed group here in town that has 3 chick singers, smoking Drums and Bass and the gals sing really well, but when they go to their Controllers and pretend to be playing parts, whiners like me notice the long forearm gloves and other gawdy bracelets and jewelry and know that performing with such bulk is impossible, but they look great, and the Hollywood Strings from PLAY is stunning.

The most I could ever wear would be a Mr. T starter kit.
Us live guys go through cycles of getting burnt out and needing changes to stay active and alert.
In the 80s I loved the automations and off and on I enjoy it, but after 6 years of working my ass off covering all of these parts live, I am ready to fully automate something, get a life sized cut out of myself and go sit at the bar and collect some cash....

[dawman]

you shouldn't see any extra bandwidth. 16x means there's 16 lanes available. the xite can only use 1 lane, so the other 15 would sit idle. however, if that slot was talking directly to the processor and not through the chipset, you might gain some really solid low latency out of it

That's what I am thinking too. The controller ia an on chip PCI-e 3.0 and the 16X sitting next to the processor, very short trace lines.
Same with RAMs proximity and the memory controller getting better latency when the DIMM's run in parallel, not serial and directly next to the CPU.

[dawman]

Here's my thoughts on this.
Below is the Chipset trace lines. If this could be a better latency option using the PCI 3.0 spec I am willing to take it.
Worse case scenario, I avoid a riser card and modify the case backplate so I can set the PCI-e 1X as before on the X58 rig.

I don't quite understand the SM Bus. Seems slow and must go to the chipset, then CPU, and also gets it signal mixed with the graphics.
I am sure since the GPU can be tweaked that this will still work, but seperating the 2 data streams and going direct to the CPU seems much better, so using a riser card and the PCI 3.0 trace lines might just work.

What Say You....

TRACE LINES.JPG