The worst part of the KT400...

[Cotita]

The name.

Had it been kt333A, plus, ultra, super duper or whatever there would be as many problems as now.

Instead of branding it as "unofficially" DDR400 compatible it should have said it supports 333mhz tbreds. That along with 8xagp and ATA133 makes a better impession than a faulty DDR400 mobo.

At least at launch KT400 mobos with DDR400 memory are no faster, or even slower than the KT266A with "el cheapo" PC2100 memory

[TheOneKEA]

I don't think VIA intends to let that lapse. The only reason it HASN'T been fixed yet IMO is because 85-90% of VIA's talent, people, and attention is focused on Barton, Hammer, and Opteron. PSN, I think VIA intends to fix DDR400 for the KT400 in the same fashion they added the 133MHz FSB to the KT133: release an A version; KT400A.

[CainSyris]

Of course, everyone forgets the original name for Kt400:

KT333A.

;)

It was VIA who tried to shake off that whole A syndrome with this launch, but this card did not get to that point. I remember that every time I look at a review for this card. I think that you should consider what comes out after as the real KT400. ;)

Just like the real KT266 was the KT266A and this is the real KT333, the KT333A. ;)

Or something like that. ;)

[TheOneKEA]

Not quite. KT133A and KT266A chipsets only added proper FSB speeds, IIRC. The KT400, despite the whole DDR400 fiasco, is different enough to warrant a full new name. Hence, KT400.

I have a theory: VIA is not bothering to fix its DDR400 controller for two reasons: A) they want to put DDR2 into the KT400 or B) they have something else to release which DOES fix the DDR400 problems, or otherwise.

Comments? This is just a theory.....

[StealthHawk]

does anyone have access to benchmark data that shows a KT400 using DDR333 against a KT333 using DDR33? i've only seen a KT400 using CL2.5 DDR400, and quite predictably, the KT333 defeated it.

edit: ok, i found a review and the KT400 with DDR333 is just as fast as the KT333

[Switch]

Originally posted by TheOneKEA
Not quite. KT133A and KT266A chipsets only added proper FSB speeds, IIRC. The KT400, despite the whole DDR400 fiasco, is different enough to warrant a full new name. Hence, KT400.

I have a theory: VIA is not bothering to fix its DDR400 controller for two reasons: A) they want to put DDR2 into the KT400 or B) they have something else to release which DOES fix the DDR400 problems, or otherwise.

Comments? This is just a theory.....

The KT266A didn't have any FSB changes. Just a tweaked memory controller. Fixed a USB problem when using a FSB of over 143 too.

[TheOneKEA]

Oh - I thought it did. Whoops.

[Switch]

Originally posted by TheOneKEA
Oh - I thought it did. Whoops.

Over all though, the KT266 was a very unimpressive chipset. The KT400 resembles it in many ways. I remember the days when the KT266 had just been released, in many applications it didn't beat the mature KT133A. Looks a lot like today's KT400 not being able to surpass the KT333. In the end though the KT266 did end up beating the KT133A in all applications. Of course, shortly after tha the KT266A was released and it was the fastest thing available. Lets just hope that VIA will do the same with KT400, either fix it or very shortly replace it. Being Via, they're probably going to put most of their efforts into replacing it.

The only reason why I'd even concider buying a KT400 based board is if it had that 1/6 PCI divider, allowing me to run an Athlon XP at 400MHz FSB.

[TheOneKEA]

Oyah - serious wonderfullness factor there. Yee-hah :D :D :D :D

200x15=3000

Try that and see what happens.

[StealthHawk]

more importantly, AMD's Irongate and Alladin's chipset both outperformed VIA's original KT266 by a good margin.

[TheOneKEA]

I've never seen an AMD chipset in operation; how's its performance and its feature list compare to nVidia and VIA?

[SnakeEyes]

I had an Asus A7M266 hybrid chipset motherboard (AMD / VIA), and I can actually say that for everything except memory bandwidth, it was on par (or better, in the case of PCI bandwidth) than my KT333 motherboard is right now. But I'm using an XP2100+ overclocked in the new motherboard, which is where my biggest issue with the A7M was- Asus decided to pull both the onboard dipswitch and the bios capability for adjusting the fsb, otherwise I'd probably still be using that motherboard instead of this MSI, despite the memory bandwidth improvement. The PCI issues with the Via chipset don't seem to be causing instability at all, but it does severely cripple my IBM 2x60GB 7200RPM RAID0 array's performance, especially compared to the same drives/controller working on the Asus. It's actually a noticable difference in performance too. :(

I'll most likely end up going with the nForce2 for my next upgrade, probably paired with a Barton and 512MB PC2700CAS2 ram (unless I can afford some brand name 512MB CAS2 PC3200..) and either an R9700Pro or the nV30. (All this speculation / comparison is making me impatient for next spring, when I'll be able to do my upgrade, even though I've already got a fairly kickbutt system compared to lots of people :p )

[StealthHawk]

Originally posted by TheOneKEA
I've never seen an AMD chipset in operation; how's its performance and its feature list compare to nVidia and VIA?

seeing as how the chipsets were basically discontinued early last year, the wre last updated over 1.5 years ago. as such, they lag in the feature department. i honestly don't know if they even supported ATA100, but anyway, they are far behind in terms of features now and wouldn't be worth buying.

the performance was less than Alladin's mobo(assuming that was the one that beat VIA's KT266, it might have been a SiS chipset). The KT266 was also slower. The KT266A however, was not. The KT333 is also slightly faster than the KT266A. nforce performs at the level of a KT266A or slightly below last time i checked. newer drivers or BIOS revisions may have increased performance.

[SnakeEyes]

I don't recall which IDE spec was supported with the AMD 76x chipset StealthHawk, but you might be right. For me it was a moot point though, since the A7M didn't have onboard RAID, and I'd installed my own Promise PCI IDE RAID controller for my hard drives instead. I can tell you for certain that that card is only DMA66 capable, but it still outperforms the onboard RAID on my KT3 Ultra, as well as the onboard DMA133 IDE controller built into the KT333 chipset on this motherboard. All of these problems seem to be related to an ongoing issue with Via and the PCI bus, btw, which is why my next motherboard won't be using a Via.

As far as other features, or lack thereof, the AMD 760 was fine. It only supported PC2100 memory, but allowed for decent overclocks using that memory (I ran mine stably at 150MHz fsb). The bandwidth at the time was higher than the competitors on the market (Via, Alladin), a fact which caused AMD to extend the life of the product, even though it was only meant to exist long enough for AMD's primary supporters to start volume production of DDR-capable chipsets (the forementioned AMD / Alladin). It took the KT266A before AMD finally killed the 76x chipset (er, not talking about the MP version, which lasted even longer).

BTW, If I was really feeling masochistic, I might even have decided to swap the A7M back into my case for a bit (along with the latest BIOS upgrade) keeping the rest of my current hardware, just to get a feel for the relative performance compared to my KT333(CE). I never had the same video card, processor, and faster memory on that motherboard (I upgraded everything at one time), so that I really can't say for certain this newer motherboard performs better, just that it has more features.. ;)

Er.. on the KT400, won't touch that. If Via ever fixes the PCI issue properly so that tons of experimentation and luck aren't needed to get all your hardware working at least close to the performance levels that it should be, I may reconsider. The fact that I seldom see >40k in drive benchmarks with my 2xIBM 60's on this motherboard (using either the PCI or the onboard RAID) in RAID0 mode with the KT333, while I consistently had close to 60k with the 'old' AMD 760 hybrid chipset tells me something.. :(