VIA KT880 Chipset

[DR]

nForce2 Ultra 400 has been around since July 2003, some seven months. Plain nForce2 has been around for over a year. While KT880 may boast better base features than nForce2 in some respects (SATA), it's still deficient and reliant on other chips for others (FireWire).

With NVIDIA ruling the 32-bit AMD enthusiast market with iron fists, VIA are going to find it hard to break into the market with KT880, given that performance is barely more than KT600, if at all, and it still lags behind NVIDIA's effort in that respect.

http://www.hexus.net/content/reviews...lld19JRD03MjQ=

[Agent]

A quickfire initial win for the nForce2 Ultra 400 board.

The AN7 takes the win again, leaving KT880 trailing. The gap isn't huge, but it's there. LAME next, a completely CPU bound application

Have the graph bars swapped half way through the tests ? Some of them don't seem to add up to whats being said.

The little blocks at the top of the graph make it impossible to distinguish which colour is to which bar as they are too small. I'm assuming the VIA is the top one ?

[DR]

the colours look wrong.... Rys will fix when he ges back to his house

[Allen]

They are the right way round... Blue is the Abit and green is the KT.

First graph smaller is better, second graph bigger is better. Blue wins both times.

[Agent]

Ahh, thats why !

Haven't woken up yet !

Sorry for the false alert !

[Allen]

Hehehe.

I should be a proof reader!

[slicemaster101]

I agree with your conclusions more or less but I think that this race was just to close to call. I think you are rating it like this is going to be the performance of all VIA KT880 based boards. This is a VIA reference design and is basically a preview sample. I think once third party board makers get a hold of KT880 and mature BIOS for it, it will probably beat the nFORCE2 ULTRA400. Personally I think most of the wins you gave to either nFORCE2 or to KT880 were just to close to call and with in the margin of error; I would have considered them a draw. At this point in time I would have to say they are equally fast platforms. VIA simply has the edge with a superior south bridge however nVIDIA sure is a proven overclocker.

Signed,
Slice

P.S. on the graphs next time use two very different colors like red and blue or just make the graph key larger.

[Rys]

It's too late to market, performance is derivative and no better than KT600. If KT880 was ever going to be faster than nForce2 Ultra 400, we would have seen it already in KT600.

No dice, it's just a cheap chipset for those that haven't got nForce2 already, KT880 won't ever be a margin of error quicker than nForce2 and performance won't sell it in the traditional sense.

Its strengths lie elsewhere.

Rys

[Zak33]

...but I think that this race was just to close to call........................most of the wins you gave to either nFORCE2 or to KT880 were just to close to call and with in the margin of error..

You know what? In most cases you'd be right, on most web sites, or in most magazines..... but both of our Hardware boyos are really immensely thorough.

Which means that Ryz would have done those tests exhaustively to be sure.

His reviews are very well respected

Anything the man says is a weighty statement. So if he's sure, then so am I.

[slicemaster101]

I am not saying that HEXUS articles are not of high quality, because they are very high quality. I have been reading them for some time and enjoy the non-bias objective coverage of products. However I still cannot agree with you on the fact that you think this is a fully developed product. This board that was tested is a sample and we all know that performance will increase on finished after market board designs. I agree that this product is rather late to market and is unlikely to pick up some real steam; however, this chipset based on your preliminary testing shows that it is honestly going to give nForce2 a run for its money. You can interrupt the raw data however you wish but the way It comes across to me is that the platforms are basically running neck in neck at this point with the sample reference board. Only time will tell how well this chipset will do as a final product and based on your test results I personally would give KT880 some major thought if I were in the market for a new socket A platform. I do think the platform needs some time to mature as far as BIOS and other tweaks go just like any chipset and I am pretty sure it will perform better with time. Whether it will perform better then nForce2 remains to be seen but performance of the KT880 can only get better from here and I am sure it will at least match nForce2 in all aspects of performance.

Signed,
Slice

P.S. I respect all other opinions on this subject. I am just simply stating and justifying my own.

[Rys]

Hi Slice,

I'd be agreeing with you if it weren't for the fact that VIA reference boards are traditionally as fast as it gets with their core logic. All too often we see performance to praise from a reference board, only too see slightly slower performance on retail examples from VIA's board partners.

It's happened with the last 4 or 5 chipset releases that I can remember, sadly.

I don't expect retail boards to be any faster and it's a fair assumption that we're CPU limited anyway and there's not much performance, if any, left to find in the chipset on Socket A.

As soon as a retail example arrives, I'll be happy to run the numbers and if it's even faster, I'll be the first to climb down from my high horse.

It's a nice chipset (I love VT8237) but I'm very sceptical about future performance increases on retail hardware.

Rys

[slicemaster101]

Fair enough Rys,

And like you said it will be interesting to see how retail boards do.
I do however strongly agree with you in regards to the VT8237 Southbridge, it totally kicks A$$!

Signed,
Slice

[justlnluck]

So now it is June, and still no sign of this chipset on the market? I saw a preview of MSI's offering of the KT880 back in April on www.tomshardware.com but MSI's website still does not have any information about this mainboard. I thought this chipset was supposed to be out at least a month ago. What has happened?

Justin

[Rys]

Your guess is as good as ours. My personal thoughts are that board partners are reluctant to put time, money and effort into producing boards that have little chance of standing out from the crowd, especially compared to any nForce2 or KT600 designs they already have.

Rys

[hartz]

I think, if Rys is correct that the new Via memory controller is indeed a copy of the PT880 one and if this does indeed mean that device IO can not use the spare bandwidth provided by dual channel DDR, then this will impact on AGP performance, as seen in the Graphics tests.

On the other hand, if Rys' suspicion is correct that there may be some "bugginess" in the AGP bus parts of the chipset, then this may get fixed in release boards and may still be improved apon. But AGP bus technology is not new, while, at least to Via, dual channel DDR is new, so I suspect the earlier rather than the latter to be the case.

I am a newbie to this hardware site, and I can only say "Hexus, where have you been all my life!" ... So much better than Tom's h/w :-)