[NTLUG:Discuss] Re: pci-e and linux ( and radeon -vs- nvidia ) -- PCIe x16 v. x8 v x4

[Bryan J. Smith]

Justin M. Forbes wrote:
> Actually, NV45 is bridged, and having troubles.
> http://www.theinquirer.net/?article=20321

Chris Cox wrote:
> Hmmm.. so maybe that link was to the NV45 after all.

Yeah, now everything makes more sense.

AFAIAC, they should just not worry about the full PCIe x16 DTR, and just
aim for PCIe x8 DTR.  It's really the same, the DTR is of little
difference when the processing is all on-board.  Even Tom's Hardware
showed that anything PCIe x4 or greater is virtually insignificant
(within 1-2% of each other) in "real world" (non-synthetic) benchmarks: 

http://www20.graphics.tomshardware.com/graphic/20041122/index.html

DTR only matters when you have something like the 6200 "Turbo Cache" and
are using main memory.  Even cards in SLI modes only use x8.  In fact,
I'm thinking of getting a SLI mainboard not for dual-video, but to run a
PCIe x4/8 I/O card in one of the PCIe x8 slots (and the video in the
other, at PCIe x8).

> Sigh... hard to tell...
> I guess Nvidia just needs to be forthcoming with info
> instead of playing cloak and dagger.

Well, do most consumers really want to know this stuff?  I lot of people
I know don't know the differences between how AGP and PCIe work at all,
but claim they do.

I don't blame nVidia for not sharing their manufacturing issues and
internal designations.


-- 
Bryan J. Smith                                    b.j.smith at ieee.org 
-------------------------------------------------------------------- 
Subtotal Cost of Ownership (SCO) for Windows being less than Linux
Total Cost of Ownership (TCO) assumes experts for the former, costly
retraining for the latter, omitted "software assurance" costs in 
compatible desktop OS/apps for the former, no free/legacy reuse for
latter, and no basic security, patch or downtime comparison at all.