Sign in with
Sign up | Sign in
Your question

Xeon mainboards support SLI on Intel chipset?

Last response: in Graphics & Displays
Share
November 12, 2004 3:46:11 AM

I'm becoming quite confused on SLI. If it's nVidia's solution, then why is it supposedly found on an Intel chipset Xeon mainboard? I know yall talked about Tumwater using it, but how is it implemented? Did nVidia license it or something?

<A HREF="http://www.tomshardware.com/motherboard/20041111/xeon-0..." target="_new">http://www.tomshardware.com/motherboard/20041111/xeon-0...;/A>

On a side note, I wonder why they didn't try SLI out? :eek: 

--
<font color=blue>Ede</font color=blue>
November 12, 2004 3:50:21 AM

Well, I always thought that SLI will be coming to both platforms. Y wud Nvidia keep SLI away from intels ? I mean, like more than 60% market is intel ? isnt it ?

:tongue: <A HREF="http://www.geocities.com/priyajeet/fing.jpg" target="_new"><i><font color=red>Very funny, Scotty.</font color=red><font color=blue> Now beam down my clothes.</font color=blue></i></A> :tongue:
November 12, 2004 4:08:55 AM

Perhaps this will help clear it up...

Dual PCIx16(phy) slots are not Nvidia's. They are possible configuration options for the PCIx standard.
Nvidia only owns the SLI concept of joining two cards to operate together via the top cable. (I don't know how far their patent goes. E.g Can other manufactures create a version with a top connector? Don't know.)

But, the bottom end (the two PCIx16 slots) are unrelated to Nvidia. Anyone can implement that.

So any Manufacturer can create a dual PCIx16 slot MB if they have the resources, which enables a user to put in two cards whether they are Nvidia or other makes.



Geoff.
Related resources
a b U Graphics card
a b Î Nvidia
November 12, 2004 5:06:42 AM

Well in addition to that, the Tumwater does not have hardware assisted Load Balancing (no nV bridge) which is a major facet of nV's SLI-centric/approved mobo designs. So while you can still run nV cards in SLI mode on the Tumwater, the best video performance would likely come from the nForce4 mobos with the load balancing bridge, or other that also support it (not sure whether VIA will get it also [doubt ATI will bother]).


- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! - <font color=green>RED </font color=green> <font color=red> GREEN</font color=red> GA to SK :evil: 
November 12, 2004 11:25:21 AM

Whats the deal with alienwares dual card technology, I heard about theirs months before nvidias, but it seems like nvidias is closer to production and a better solution, is alienware going to junk their project or is it still in the works?

Asus p4c800 Deluxe,1 Gig Mushkin PC3200 Dual Channel Level II V2,Pentium 4 3.0 512k 800fsb HT, Thermaltake Xaser III, Thermaltake Spark 7+, Sound Blaster Audigy2 ZS Platinum Pro, eVGA GeForce 6800 GT
a b U Graphics card
a b Î Nvidia
November 12, 2004 4:59:06 PM

Well I wouldn't be so quick to say nV's solution is better. Unless Alienware's ALX is truely undoable. Right now neither technology has made much of an apperance.

The specs for ALX even mention more than just a dual card configuration, so if possible 3-4 cards may vault ALX ahead of SLI from sheer brute strength. Also ALX's flexibility with card options would be an added bonus. All these questions will not be answered of course until these two companies get off their asses and actually get some realistic technology descriptions and full benchmarking & tweaking investigations underway by independant sources. Even [H]'s look into SLI rellied too much on the PR provided figures.

ALX looks like it's a better method on paper, but I don't think Alienware has the power to drive development in the industry and will have to wait for the products to come to them, unlike nV who can somewhat drive the industry in the directions they need, and that may be it's biggest advanatage. Just like Beta, ALX may be a better technology/method, but without adoption or support, it may simply be another Truform or Glide.

EDIT: Also, I think Drivers will play a VERY big role in the success of these technologies. All the demos and such sofar have been well controlled, and we've seen nothing outside of that control. Using this method we should all be using XGI Volari DUOs since they were able to demonstrate MASSIVE fill-rates on their cards in the respected 3Dmk01 and 03 benchmarks. But what happened? Exactly. I'm still waiting to see real-world tests before decidiing ALX's fate. Both have shown stunning controlled demos, but nothing free-form yet.


- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! - <font color=green>RED </font color=green> <font color=red> GREEN</font color=red> GA to SK :evil: 
<P ID="edit"><FONT SIZE=-1><EM>Edited by TheGreatGrapeApe on 11/12/04 12:02 PM.</EM></FONT></P>
a b U Graphics card
a b Î Nvidia
November 12, 2004 8:27:37 PM

There are two issues with the bridge. In most motherboard the bridge does double duty, it splits the 16x PCIe lanes into 2 8x lanes/sockets or 1 16x lane/socket, and also therein is the load balancing hardware. The tumwater doesn't have similar architecture to most of the other mobos, and it actually has 3 bundles of 8 lanes and they can be assigned as the user pleases. I'm not sure if two of the 8 lanes can be combined to allow 16 lanes for one PCIe graphics slot or not, or whether it is always a max of 8lanes to each PCIe 16xGraphics slot. Also they are 16x slots, but that's really the form factor now, and not so much the actual lanes.

It's been a while since I read the Tumwater stuff, but I think because it can independantly assign the lanes, it works well with basic SLI, but definitely does not give the boost of load balancing. This may also explain why the results were somewhat dissapointing in the intial tests which were topped by the OC'd X800XTPE, and subsequently trounced by a true nF4 SLi setup. That's just a guess on my part, but it does make alot of sense since it gets the boot of the load sharing whereas with the Tumwater the load was probably shared 50/50 resulting in one card waiting for the other to catch up. Also it could simply be improved drivers too of course. Who knows 'til we have actual gear hitting the streets.


- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! - <font color=green>RED </font color=green> <font color=red> GREEN</font color=red> GA to SK :evil: 
November 12, 2004 8:34:42 PM

I think he's referring to the chipset's ability to perform the load balancing, not the physical PCB for connecting the cards.

As far as I know, that PCB is required.

"Who is General Failure, and why is he reading my drive?"
P4 3.0C HT, Intel D865GBF, 1GB Crucial PC3200 DDR, 2x WD 36GB Raptor 10kRPM, BBA Radeon 9800PRO, SB Audigy, Hauppage WinTV
November 13, 2004 3:24:24 AM

Quote:
Perhaps this will help clear it up...

Dual PCIx16(phy) slots are not Nvidia's. They are possible configuration options for the PCIx standard.
Nvidia only owns the SLI concept of joining two cards to operate together via the top cable. (I don't know how far their patent goes. E.g Can other manufactures create a version with a top connector? Don't know.)

But, the bottom end (the two PCIx16 slots) are unrelated to Nvidia. Anyone can implement that.

So any Manufacturer can create a dual PCIx16 slot MB if they have the resources, which enables a user to put in two cards whether they are Nvidia or other makes.

I do know about the physical specs of the boards, I was just wondering how the SLI technology (as THG stated in that article) got in there when it's nVidia's stuff?

That said, based on what you explained, wouldn't nVidia's technology be nothing much really, if anyone now can make a PCIe board with two 16x slots and have fun with dual card configs? In other words, are you saying that creating two 16x slots like those can allow dual card configs without requiring a special connector or PCB?

--
<font color=blue>Ede</font color=blue>
November 13, 2004 3:25:15 AM

Quote:
You dorks need to take this conversation the the motherboards forum:mad: 

Make me. [/sexy provocative voice]

--
<font color=blue>Ede</font color=blue>
November 15, 2004 12:54:30 AM

--Quote----------------------------------------------------
In other words, are you saying that creating two 16x slots like those can allow dual card configs without requiring a special connector or PCB?
-----------------------------------------------------------

If when you say dual card you mean SLI, no I don't mean that. Having two PCIx16 slots will support whatever is designed to use them whether separately or as a pair. And as a pair does not have to be SLI.


Geoff.
November 16, 2004 1:44:02 AM

Wasn't refering to only SLI but any dual card technology. Because it sure seemed that way with the mainboard and the way THG talked about it. (even refering to it as SLI)

--
<font color=blue>Ede</font color=blue>
!