Chipsets

[gtaylor]

Having still not bought the bits for my new build, I am now looking at the differences between the nForce2 ultra (such as on the Asus A7N8X Deluxe rev.2), and the VIA KT600 (such as on the ASUS A7V600). Now, features wise they are about identical, except that the nvidia chipset has dual-channel memory. But, I remember the benchmarks from the KT400a compared to the nForce2 and the dual-channel made little or no difference.

Have any of you guys seen comparative benchmarks between the KT600 and nForce2? I have had a look around at the sources I know and trust, but have found nothing yet.

[Peter M]

It's still such that NForce2 is ahead by a quite negligible amount ... with the exception of one or two professional 3D rendering scenarios where the dual channel RAM makes for an about ten percent difference. For everyone else, it's not worth taking the extra effort and not-really-expandability of dual channel RAM NForce setups.

But mind, KT600 performs best with one double sided DIMM.

[causticVapor]

But then again, feature-wise, they can't be beat... the MCP-T has a really good audio codec, and lockable AGP/PCI busses are really a treat when overclocking... which helps as the latest nForce2 ultra 400 chipsets have been shown to reach 250MHz on stock VDD....

also, the HT link between the NB/SB helps when chaining a lot of cards, peripherals, etc. to the system...

just a thought...;)

[Peter M]

MCP-T does not have a codec, it has an audio engine and REQUIRES an external codec to be attached to it by the mainboard designer - the choice and implementation quality thereof being entirely up to those guys. NVidia has a "soundstorm" approval program to ensure a minimum feature set and audio quality, but that's all there's to it.

Overclockability is important for about .01 percent of the market; the vast majority of people doesn't toy around with their computers, they get stuff done with them. Performance at stock speeds is the important thing.

I also don't see why NVidia's Hypertransport link should have any advantage over VIA's V-Link - the speed is about the same, headroom for I/O is identical. When you want an advantage from a really fast I/O link, you need to buy SiS chipsets - these have a 1 GByte/s link from north to south.

[Yoshi]

Originally posted by Peter M
Overclockability is important for about .01 percent of the market; the vast majority of people doesn't toy around with their computers, they get stuff done with them. Performance at stock speeds is the important thing.



you have to check you books, overclocking I would say is important for 2-3% not 0.01%, 0.01% of all the people with computers (estimated at 200,000,000) is 200,000, there are 40,000+ people in this forum alone and tons of other forums around with 95% of them tweaking away.

[causticVapor]

Getting stuff done faster is always an important thing. :r

Seriously, Peter, lighten up. OK, I forgot the MCP-T doesn't have a codec,

and there are several reviews that put the nForce2 consistently ahead of the KT600 by several percentage points...

[ComPooTer]

i like nforce2 i heard of people having a lot of problems with VIA and i think thats what made me buy the asus a7n8x deluxe w/ nforce2 SPP-ultra

[DaveLeclerc]

Which chipset would be best for those not ocing? I am biulding for my brother-in-law and want a mobo with flexibilty but not directly for ocing. Only to allow us to fine tune the system.

[ComPooTer]

depends on what features your looking for

great onboard sound, SATA raid, nforce 2 chipset = asus a7n8x deluxe or u can go for the non deluxe version if u dont need all the bells and whistles.

u can also look into msi their boards r pretty nice

[BipolarBill]

None of this matters. I like the nForce2 and that's all you need to know! :p

Oh - SiS makes good hobby boards too.

[Bigjakkstaffa]

Nforce 2...if only because they overclock like nobodys business *hugs 8rda* :cool:

I do actually quite like VIA, but will acknowledge thay have in the past suffred from several glaring bugs and do seem to take their tiem getting up and doing anything about them, also from what ive read the nf2 is marginally faster

--Jakk:t

[Peter M]

I don't want to start discussing this all over :rolleyes: but all the stuff that's been hyped as "VIA chipset bugs" actually were problems caused by 3rd party violations of PCI specs (Can you say Creative?) and/or sloppiness in implementing BIOSes (which VIA's mainboard customers do, not VIA themselves).

The chipsets themselves were and are OK. If you want to see recent and true chipset screwups, you'd have to look at Intel's 820, 840 and 850.

[$1500-P4 gamer]

I had the !850 rdram Intel chipset. Rock solid-fast and not one bug in 2 years I found. The asus I had oclocked like a mofo and STILL is running with newer mobo's bench wise (friends got it now)! One of intel's fastest chips sets ever and you think it was bad?! This thing STILL runs with mobo's two years ahead of it, I can not name one via that can boast so much-can you? Compairing that to Via is like compairing a porche to a Ford festiva! LOL. Via not buggy- why then ALL the usb and printer problems eh? That third party too? Not being sarcastic-Im asking, are the usb ports and printer port run through some other co's chips???? I was lucky to have my via usb work 1/10th the time I wanted it too. And lots of devices just wouldnt work on usb no matter what I did. Intel chipset updates were never needed by me either but with Via its a weekly visit. Thats why they call their website the via arena, cause your gonna be there alot! ROFLMAO.:D :r

[causticVapor]

IMHO, the 850 chipset was one of the best. And the 840 didn't suck as much as the 820.

Also, the VIA Apollo pro 133 and all chipsets above KT266 are really good.

[Bigjakkstaffa]

I loved the Kt133a...great overclocker, but still did have its problems, admittedly none were massive (excpet for a really big problem with the new UT engine), but VIA did seem to take time addressing these issues and then it came in the form of a beta driver

--Jakk:t

[Peter M]

OK ... history bits ...

VIA USB problems in Windows were (and still are) UHCI driver problems. Just look at the huge mess of issues on MS Knowledgebase. Isn't it strange they're working just fine in Linux? To be fair, the Linux UHCI driver also required at least one complete rewrite to work. UHCI programming model is just quite braindead. Intel and VIA use that, everyone else uses OHCI.

Printer port problems. On 686 south bridges, the printer port logic indeed is in the southbridge - but getting the analog circuitry correct is up to the mainboard designers. Carelessness in that area? Do you want fries with that? With all other south bridges and chipsets, it's in a non-VIA chip.

Intel 820 and 840 both were buggy in terms of SDRAM support chips MTH and MRH, respectively. To the amount that Intel had to scrap the chips and recall the boards. 850 has a non-relievable problem in bus master read performance on the hub bus, affecting IDE, SCSI and Gbit Ethernet performance when writing out to these physical media. Not that people usually notice, most benchmarks just bench the opposite direction. (Yes, PCI performance problems were seen on VIA chipsets too - but that was sloppiness in configuring the PCI controller, easilyy relieved by BIOS and/or driver updates. The i850 has a hardware problem here.)

And before anyone pulls that one from its grave, the much hyped VIA IDE data corruption issues on 686B south bridge finally (and when noone was listening anymore) boiled down to be a bug in the Creative SB!Live.

[DaveLeclerc]

I have been enlightend.
Thank You Peter for another dive into the fantastic history into the "Chipset".
So we are all agreed... nForce2.

[$1500-P4 gamer]

Thanks Peter, that was very informative. I had no idea about the usb and printer stuff-makes sense now. Thats why I asked instead of just blaiming Via. I can def agree with ya on the sb live too! I had the first series, the MP3+ one-hiss pop and click were its favorite passtimes when it worked in a system. Alot of mainbaords didnt get along with that one! I got rid of it, Integrated was far better.:t

[causticVapor]

I have an SB Live! on an 815 board, and in some situations (particularily when EAX is enabled) a whine develops and eventually gets to such a high volume that you're forced to rip the headphones off. This happens every now and then - the Audigy in the other system and nForce2 audio don't have this problem.

And oh, good that you brought up that 850 PCI performance problem, Peter. Almost forgot about it. :eek: