Low Profile Multi-Display Graphics on the Cheap

[shabodah]

They would work at the same level as an x1 PCI-Express video card which are rare but expensive.

For me the beauty of it would be you could have 2-3 busses operating at much higher rates and be seperated from each other.

You're missing the point of PCIe. It is not a "bus" in the general sense any more than SATA is. There is nothing inherently making PCIeX1 cards more expensive than PCI 2.3 or PCI-X cards. Even if (1) PCI 2.3 video card gets the same bandwidth as one PCIeX1 card, things change as soon as you add another PCI card. You seem to understand the point of having separate "buses," but don't seem to understand that that is part of the point of PCIe.

 

[hcforde]

YES, I do understand. Let me give you a real life situation. I had a P4 with a 4 port PCI video card. I added it to a Quadro 900XGL AGP card. The bandwidth remained the same, but each time I added a monitor the speed of a change of monitor content dropped.

I changed to PCI-E very fast, added Pci-X still very fast PCI-X not quite as fast as

I was saturating the PCI bus

I am also currently running SCSI on the PCI-X bus but plan to go the the WS Quartet (or equiv)after I find a descently priced PCI-E SCSI RAID card for the exact same reason you stated but don't seem to think I understand. The 16 lanes of each PCI-E slot has great potential but the early adopter cards are very expensive. Now, I have seen 2 port PCI-E video cards for $30.00 NEW. However 4 port video PCI-E cards are VERY expensive--more than the flagship Nvidia or ATI cards, The Matrox offering is $799.00

When used for SLI the x16 PCI-E slot gets reduced to x8 that sounds like BUS behavior to me. Just because the manufacturers don't want to call it that I am looking at its behavior. HOWEVER, It is maturing very quickly and some are able to keep the x16 bandwidth if cards are used in both slots and I can only see this getting better. And for all intents and purposes the bandwidth is not saturated at x8 speeds. In the mean time there is a lot of corporate technology that can be had for great prices in the PCI-X format, but general consumers need not apply. That said, 66Mhz PCI slots may be too little, too late for most, but every little bit helps in the transition to all PCI-E slot motherboards. And I for one am looking forward to it for the reasons you mentioned.

 

[krazyIvan]

Its interesting that with the increased interest of MCPC's in cases that are A/V in form-factor, more companies don’t come out with half-height, passively cooled mid-upper range video cards.

 

[shabodah]

When used for SLI the x16 PCI-E slot gets reduced to x8 that sounds like BUS behavior to me.

Good point there, but as you said, the higher end SLI chipsets do not do this. And how the two PCIe x16 slots are set has no effect on any PCIe X1 slots, whatsoever. So, even a regular micro-atx board with one PCIeX16, one PCIeX1 and two PCI slots can run a lot of monitors (depending on what you need to do) with a regular graphics card and a card like this:

http://www.newegg.com/Product/Product.asp?Item=N82E16814161017

You could even throw in a pair of these:

http://www.newegg.com/Product/Product.asp?Item=N82E16814129062

 

[TheGreatGrapeApe]

The real issue is who and what is driving the market? Niche products can be very profitable and sometimes I have niche needs. Matrox has been very profitable in their niche.

Actually they haven't been VERY profitable in their niche, just stable profitability. Very profitable is the low-end and OEM stuff which they have little of. Their profit or margin per unit is good, but their overall ROI is low.

Everyone wants to be the big dog in the big market but eventually that market gets wittled down.

And eventually the Big dogs goes after the little dog's niche and will hemorrahge money to do it. Expect Matrox to lose alot of business in the display market in the future because ATi/nV/SiS/Intel are all going after the traditional Matrox market, and their smaller niches will not provide enough revenue to survive. Matrox is moving to capture/editing-solutions, more than just display cards. nV is even going after Matrox and Chyron's pro market.

Matrox saw this a few years ago when it got blasted on delivering the Parhelia with too little too late.

The Parhellia was better than the R8500, and competative with the GF4ti (with more features than both).

The have changed their business plan and done very well in Video and 2D imaging.

For now, but they are simply staying afloat, and if the big booys target those markets, buy buy Matrox. The X1300 on PCI , AGP and PCIe is a wake up call IMO, providing more video option than the P and G series, and the M is being attacked by these products and the FireMV ones.

If the general PCI video card manufacturers would raise the level of their PCI cards to work at that level their could be an increase in PCI video cards.

Why bother? PCI 2.3 requires motherboard redesign so why not simply go with PCIe, most new workstations have at least 1 PCIe slot nowadays, and most have the old AGP PRO 4-8X 50/110.

They would work at the same level as an x1 PCI-Express video card which are rare but expensive.

They are no longer expensive, and not that rare;

http://www.newegg.com/Product/Product.asp?Item=N82E16814161017

FireMV is expensive @ $400, but on X1300 @ $100 a pop, you get dual link DVI unlike the base Matrox P series PCIe 1X.

But for basic 2D and 3D which is what Matrox does, this X1300 is cheaper and better than most;
http://www.newegg.com/Product/Product.asp?Item=N82E16814102001

The only area Matrox still owns is PCI-X, and then great multi-mon software.

Next barirer for Matrox, not one of their card will be compliant with Vista's AeroGlass, thus reducing their market ever more.

For me the beauty of it would be you could have 2-3 busses operating at much higher rates and be seperated from each other. Those 3 busses being
PCI-E/16 lanes@250MB/s per lane,
PCI-X/133Mhz * 64bit,
PCI/ 66Mhz * 32 bit

Why bother separating them since if you have PCIe and PCI/PCI-X likely the PCI/PCI-X are running off 1 PCIe lane anyways. Persoanlly I'd prefer 2 1X PCIe + 1-2 PEG 8-16X for such a system.

You woud think that the new ASUS workstation boards would have done this but the P5WDG2-ws-Pro has PCI-X but the 2 PCI slots are 33Mhz

The ASUS P5W64-WS Pro has 4 PCI-E slots and 2 PCI slots running at 33Mhz.

Reinforcing the end of PCI other than standard PCI. PCI2.3 makes little sense, just like PCI-X 533, no point anymore. And alot of 2.3 like on the Tyan boards are also showing as 32/33.

Machines like this are not unheard of in financial circles and the scientific arena. some of these machines have 20 monitors.

In which case you'd be better off with something more along the lines of a Tyan board with many PEG slots and many PCIe or PCI/PCI-x slots.

Niche, YEAH, But a profitable niche

Profitable in indiviudal terms, but not profitable when seen overall for the business, and it's owning a dying niche for now pfotiability is evaporating fast. They will have a 'maintain' strategy for people who already have MAtrox, solutions, but that only works for so long.

Matrox needs new hardware to continue to be a viable entity, and I say that as someone who loves Matrox.

 

[hcforde]

YEAH, but what is the jist of my argument and what am I looking for as an end product? Everything is done incrementally and from what I see we will get to all PCI-Express boards faster than any other change of slots the industry has seen in the past. BUT IN THE MEANTIME. The change will still be incremental. You did not address that the more PCI cards you have the slower the whole thing gets so why would I want traditional PCI. I DON'T. For me the best setup is as explained for what I do.

Also extra PCI-Express lanes are a redesign because it is usually another chip that has to be included to get them. As time goes on chip resources will be allocated to putting them in because there will be more cards to use them.

It is more than just the technology of being able to do it. Each board they make is to fit a market segment "a niche". The question is asked what niche would like what on their boards, and how can we fit our technology to them so they will buy it. It is called marketing. Finding out what the customer wants and giving it to them so that they will buy the probuct in an amount that will give them a profit.

There are some people who do not understand it. A few months ago Gigabyte came out with the QUAD Royal a 4 PCI-Express slot board. All the gamers were only looking at their application of the board. Some reviews were even asking the question, Is there a market for it?

Well DUH!!!

The European and Asian market got the ASUS P5W64 4 PCI-E slots long before it hit the US market and if they are calling it the Wall Street Quartet they have their market.

 

[shabodah]

Also extra PCI-Express lanes are a redesign because it is usually another chip that has to be included to get them. As time goes on chip resources will be allocated to putting them in because there will be more cards to use them.

Technology will change, sure, but I don't really see what you're trying to get at here. Sure a northbridge/southbridge chip could eventually need to have more and more PCIe lanes, but we could be seeing many different revisions of PCIe by then, much like Sata 6g is in the works now.

 

[TheGreatGrapeApe]

You did not address that the more PCI cards you have the slower the whole thing gets so why would I want traditional PCI. I DON'T. For me the best setup is as explained for what I do.

No, you example involved multiple PCI, PCI-X and PCIe solutions.

ALL PCIe would be far better.
As for 'incremental' why? PCI is dead, PCI-X is also dead. PCIe is already the new standard, and already we are seeing al PCIe board, and if the server market didn't already have a ton of PCI-X Giabit Ethernet and SCSI RAID controllers we'd see them gone too.

Also extra PCI-Express lanes are a redesign because it is usually another chip that has to be included to get them.

What are you talking about? The lanes are either there or not, if you want more PCI or PCI-X then you still need the PCIe for that too. So whether it's a single chip solution or dual like nV's PRO chipset makes little difference be it 40 or 48 lanes, you still have to allocate them properly.

A few months ago Gigabyte came out with the QUAD Royal a 4 PCI-Express slot board.

Which they leaked about a year and a half ago, and which was discussed here in many ways.

All the gamers were only looking at their application of the board.

Why not? Should they have been discussing the possible 24 drive raid controller options or optical networking controller possibilities?

Some reviews were even asking the question, Is there a market for it?

Well DUH!!!

Well that's a mature thought, I'm sold. :roll:
The question is not whether there is any market at all for a product, but whether there is enough of a market for it to become more of widespread product. You could sell a handful of $50,000 graphics systems if they were fasty enough, there is a market for it, but is there enough of a market to make the cost of production and the R&D focus worth it?

The European and Asian market got the ASUS P5W64 4 PCI-E slots long before it hit the US market and if they are calling it the Wall Street Quartet they have their market.

Which is still not really a good guided solution, true Wall Street system builders have been using 3rd party inegrators like ColorGraphic for years to get dedicated systems, and they don't care about the cost, matrox G200-MMS or G450-MMS would be fine for them. The market for the Asus board if anything is the wannabe broker. It should be called the 'Day Trader Tetrad' because that's their market, day traders and gamers.

Heck when it comes to multi-mon, the flight sim gamers tend to be the ones who break ground for everyone else, including the financial & statistical users.

The reality is that the 2D impact is very low on resources, and the biggest slowdown for 2D implementation is the system resources required to provide fresh data for the monitors. The burden of the PCIe lanes and the cards themselves is minimal compared to gaming.
The only reason that an array of G450s , FireMVs or Quadros will be limited versus a bunch of PCIe 1X X1300 is the move to 3D desktops; stick to Win2K, Unix/Linux, and apps like Xinerama and it won't matter. But then again, it also means no need for the 'WSQ' either.

 

[kilroy0097]

Just a heads up. There is a NVidia GeForce 8600GT Low Profile card out there now. I can't find anyone in the US that sells it but it does exist.

http://www.galaxytech.com/Product_Details.asp?id=135

Anyone know where I could perhaps grab this card through online order?