@fwoot

The QPI links do make it easier** to make multi socket motherboards but that is mainly targeted at workstations and servers rather than consumer desktops. Consumers hardly know what to do with their current quad core processors and would receive little or not benefit from having 16 theoretical threads. Its likely we will see a hybrid enthusiast/workstation board that supports 2 sockets much like the previous "Skulltrail" boards but it will be priced out of most peoples range and I wouldn't be surprised if they locked it down to only working with "extreme edition" (or whatever they end up calling them) Nehalem cpu's.

**easier in terms of chipset's at least. I imagine routing the traces for two sockets each with 1366 pins is a complete nightmare and costs a fortune. If nothing else that's going to ensure that multi socket boards cost a little bit extra.

@AuDioFreaK39

From what I understand the memory support is mostly just a voltage limitation. At the moment there are no DDR3 modules I know of that can sustain anywhere near 1900mhz with less than 1.65v but chances are that will change in the coming months. The motherboard makers are probably just covering their backs for when higher DDR3 speeds at less than 1.65v become a reality.

If my PSU wasn't overloaded I would probably spring for some neons. As it is I've hot wired an old PSU to power my gfx card separately... waiting for the moolah to get a more stable setup lol

I'm currently a little bit behind the computing trend with a E6320 C2D Cpu, 2GB of ram and a 7800GTX gfx card I knicked off a mate. Even so I can play most games that interest me.

If these new 4670 ATI cards come out and perform as expected I might get one of those, considerably lower power usage and probably a bit faster. Not sure a CPU/MB/Memory upgrade is worth it until Nehalem matures.

I don't tend to play online games, don't really have the competitive bug... but from what I've seen Xbox Live does seem to be a bit more mature. I'd probably use it more if it was free.

After reading this article I thought I would develop a bit of a counter point. Although its a shame to see Nehalem potentially being held back like this its worth considering a few extra factors before considering this a complete loss.

DDR3 has architectural improvements over DDR2 which means even at the same frequency it should offer similar if not better performance (even given its higher latencies). As DDR3 matures the latencies at 1066 should drop without the need for any increased voltage so there will still be headroom for a bit better performance has the platform ages.

Also, memory bandwidth is a function of both frequency and channels. The memory controller on the high end Nehalems is triple channel so theoretically it should be to give a 50% improvement in bandwidth over Core 2 Duo (Yeah right... not going to happen... but it should be faster).

Finally the onboard memory controller and removal of the FSB should unlock a bit more bandwidth, as it did with AMD's Athlon/Opteron series.

Overall we are going to be looking at memory performance that should be strictly faster than DDR2 at 800mhz, and probably a bit higher than the "unofficial" DDR2 1066mhz, used on the Core 2 Duo platform. The bandwidth will also likely be greater than the AMD Phenom/Athlon processors at least until they adopt DDR3 sometime in 2009.

Although DDR3 1066 sounds slow, when you take into consideration everything it should be more than sufficient. The only people that are going to suffer are the kind of people that like to mount neon lights onto everything in the case and brag about how high their numbers are.

Thoughts?

Its all about choice and revenue...

If an end user could buy an AMD cpu + chipset then run SLI AMD would still get money for the chipset. With the current situation anyone that wants to have SLI is forced to buy an Nvidia chipset on an AMD platform.

Choice is always good as an end user (competition invariably leads to lower prices), and it would be good for AMD because they could increase their share of the chipset market on their own platform. It would however be bad for Nvidia since anything AMD gains Nvidia invariably loses.

Some people are always going to want to use SLI, regardless of whether crossfire is better or not. If AMD could sell chipsets to that market as well they would make more money.

But hey, since Nvidia would probably rather release a FX5200+ than give AMD SLI its a bit of a moot point

I own an Xbox 360, my housemate a PS3 and we also both have PC's capable of moderate gaming (not quite Crysis level  ).

It depends what type of gamer you are and how much you expect from your console. If all you want is to play the normal console games like racers, shoot-em-ups, rhythm games etc then the 360 serves well. The PS3, though potentially technically superior, is considerably harder to program for and so games on both platform run pretty much the same (infact some 360 games actually look better).

However the PS3 is a considerably better multimedia device. Its streaming capabilities work better and the blue ray drive does give it an advantage. It remains to be seen if the new 360 dashboard update will correct this but I wouldn't bet on it. Pretty much every big game is coming out cross platform now so its games catalogue is expanding fairly rapidly. Biggest downside as far as the PS3 is concerned is the price, which can only be justified if you are actually going to make use of those extra multimedia features.

As usual the PC is the best option for the "hardcore" gamer, especially with the "games for windows" initiative on Vista bringing the online experience and peripherals from the 360 where appropriate. Plus on the occasion when I actually have to get some work done its nice to have a PC with good specs so I can do silly things like run 5 virtual machines from memory simultaneously.

Overall I decided on a combo of Xbox 360 core system for "casual" gaming and a midrange PC (built myself, components circa £300) for strategy/fps and work. Doesn't cripple the finances and covers most if not all of the bases.

Just my 0.02p

Well short of an entirely new architecture (which is undoubtedly on the way but not in '08) its pretty much all Nvidia can do right now.

They might be better served taking a leaf out of ATI's book by using clock speed to differentiate products rather than cutting bits of the silicon off. Limiting the number of shaders and memory bus has previously hampered most of their more mainstream products, whereas 50 - 100 Mhz off the clock speed (as is the case with ATI's 4870/4850) would only result in near linear performance loss. Of course that would probably require better yields than they are getting.

Overall this is probably the best Nvidia can do in the short term.

Hello all...

I've been an avid Fudzilla reader for a while now, after previously reading the Inq' (still do, but if Charlie Demerjian posts one more Nvidia rant....).

Keep up the great work guys!