VIA 2002 Chipsets Plans

[NFS4]

VIA KT333 - 200/266MHz FSB, V-Link 266MB/s, DDR 266/333, AGP 4X, Sample Q1 '02, Production Q2 '02
VIA KT333A - 200/266MHz FSB, V-Link 533MB/s, DDR 266/333, AGP 4X/8X, Sample Q2 '02, Production Q3 '02
VIA KM333 - 200/266MHz FSB, V-Link 533MB/s, DDR 266/333, AGP 4X/8X, Integrated Zeotrope GFX, Sample Q2 '02, Production Q3 '02
VIA K8HTB - K8 HyperTransport Bus, V-Link 533MB/s, DDR 266/333, AGP 4X/8X, Sample Q2 '02, Production Q3 '02

http://www.vr-zone.com/#1922

 

[Insane3D]

Thanks Brandon...good info. I'm curious about the KT333...the KT266A come with the correct clock generator chip for 166mhz FSB operation (333mhz), but supossedly the chipset doesn't support it. Makes me wonder if it really is capable, but Via doesn't want it used, otherwise there would be no real reason for the KT333 since a KT266A @ 166mhz FSB would be essentially the same thing... Also good to see they are improving upon the already very nice V-Link technology. With a bandwith of 533mb/s, it should perform very similar to any boards using Hypertransport, like the Nforce boards. I would imagine their wouldn't be much difference between the 533mb/s of V-Link vs. the 566mb/s of Hypertransport, although Hypertransport is supposed to evolve some and increase the peak bandwith from the current 566mb/s. Makes me wonder why no one is bothering with the Multi I/O integrated chip design of Sis...it is still the leader by far with 1.2gb/s of bandwith, plus you get the added benefits of having the NB and SB on one chip. I would think things like less traces, cleaner layout, and cheaper production would make it a very attractive technology.

 

[AGodspeed]

VIA K8HTB - K8 HyperTransport Bus, V-Link 533MB/s, DDR 266/333, AGP 4X/8X, Sample Q2 '02, Production Q3 '02

Yummy <drools>

Anyway, it looks like we'll know what Hammer can do by next summer. I wonder if we'll see Hammer (ClawHammer) before next October.

 

[Adul]

hypertransport bus sounds very interesting.

 

[Demonicon]

Woohoo HTB, thanks for the info NFS4.

 

[kuk]

Why is VIA already predicting a KT333A?

I mean, the KT133A was a revision of the KT133, alowing 133 bus speeds to be used. KT266A was a revision of the poor performing KT266.
The only thing special that can be seen in the KT333A is the 533MB/sec V-link, and I think its only use is to provide enough bandwith to an integrated video solution (KM333 + Zeotrope).

 

[ST4RCUTTER]

Why is VIA already predicting a KT333A?


Cause VIA already knows they're gonna fuxor up the first iteration...

 

[Executor]

LOL


Yeah Via needs to get things right the first time

 

[vailr]

Southbridge

VT8235 - 6 channels of AC'97 audio, ATA133, HSP V.90, 6 USB 2.0 ports,
Via Mac with MIII interface, DSL support and advanced power management,
Sample Q1 '02, Production Q2 '02

 

[LXi]

I dont think they're predicting the KT333. I think they saw the revision as a necessity, why, first of all V-Link will be upgraded to 533MB/s. The other major feature is AGP8x, I would assume that the 8x standard isnt available quite yet by the time KT333 comes out, so they're planning on a revision incorporating the new V-Link and AGP8x. The actual performance differences, we dont know, but looking at their history,the "A" will most likely have a huge gain, but Im actually betting that it's not the case this time. nVidia does refreshments for every generation of their video cards, nobody seems to be complaining about that.

 

[flexy]

<< hypertransport bus sounds very interesting. >>



yeah..sounds like something from star trek

 

[AGodspeed]

yeah..sounds like something from star trek

I just hope Cisco uses HT. Nothing can compete with HT currently (in terms of scaling), and so making it a standard certainly wouldn't be a bad thing.

 

[Athlon4all]

Nothing unexpected really, only thing missing is KM266, maybe they're afraid to release another Int. chipset using the out dated Savage4 3D/Savage2000 2D core with nForce and SiS 750 in the picture. Oh well, KT333 looks nice but only if AMD releases a 166fsb CPU(I really think that I will be VERY suprised if AMD doesn't do that for Throughbred, to compete with Northwood). I really think that VIA may be ready for a comeback in the P4 arena with P4X333 because, unless Intel's Dual Channel DDR chipset is gonna be using QBM, P4X333 should be much, much cheaper than this Dual Channel DDR Chipset, and still have the same bandy. I dunno, could Intella be using QBM in this Dual Channel DDR Chipset (btw what's it's codename?). I am though very, very excited about K8HTB(Who isn't though?), I am though curious as to how is designing a chipset for Hammer different b/c of the integrated north Bridge and Memory controller, you know?

<< Cause VIA already knows they're gonna fuxor up the first iteration >>

LOL, yea I bet!

 

[Insane3D]

"Oh well, KT333 looks nice but only if AMD releases a 166fsb CPU"

Not really. The limiting factor with the FSB of current Tbirds/XP CPU's is the chipset/dividers and memory. If the board is designed to run @ 166mhz FSB, with all buses in spec, and the proper memory is used it should work fine on the Tbirds and XP's. Only problem will be with the XP's needing to be unlocked to reach that speed. The Athlon/XP's EV6 bus should scale just fine up to 200mhz FSB with proper chipset and memory support. All AMD would need to do is change the multiplier and sell them as 333mhz FSB chips. It's sort of the same thing with the 200mhz Tbirds/Durons and the 266mhz pieces. The Duron's and 200mhz FSB Tbirds will run fine @ 266mhz, but the multiplier needs to be changed. The only physical difference is the multiplier...

 

[AGodspeed]

I am though curious as to how is designing a chipset for Hammer different b/c of the integrated north Bridge and Memory controller, you know?

I am also curious. I have seen roadmaps with chipsets from ALi, VIA, SiS, etc. displaying what seem to look like a northbridge. Of course, Hammer is supposed to have an onboard memory controller in the form of an integrated northbridge. I don't think it's possible to have TWO northbridges. But anyway, that's too far off to be thinking about anyway.

 

[AbRASiON]

Not really. The limiting factor with the FSB of current Tbirds/XP CPU's is the chipset/dividers and memory. If the board is designed to run @ 166mhz FSB, with all buses in spec, and the proper memory is used it should work fine on the Tbirds and XP's. Only problem will be with the XP's needing to be unlocked to reach that speed. The Athlon/XP's EV6 bus should scale just fine up to 200mhz FSB with proper chipset and memory support. All AMD would need to do is change the multiplier and sell them as 333mhz FSB chips. It's sort of the same thing with the 200mhz Tbirds/Durons and the 266mhz pieces. The Duron's and 200mhz FSB Tbirds will run fine @ 266mhz, but the multiplier needs to be changed. The only physical difference is the multiplier...

So what you are saying is, another divider on AGP / PCI / etc and the EV6 protocol will do it straight off the bat?
So we could run an UNLOCKED Duron @ say 5x166mhz you think? (for example?) all it needs is the right board traces and multiplier on the chipset.

That's very cool, very cool indeed - I didn't know that, I thought EV6 would mean we are stuck @ 133mhz.
and 333mhz does not interest me with 400mhz being rumoured from SIS very very soon.

 

[Insane3D]

I had an Epox 8K7A that I got a 1.4ghz Tbird running up to a 195mhz FSB. The PCI speed was so high at that speed that it didn't do it for long, and the memory was pushed too far. I just bought a regular Duron 850 a couple weeks ago that would run a 165mhz FSB on my old Iwill KK266-R. Getting to 200mhz FSB might be a bit trickier as that is the max the EV6 was meant to do, but I would imagine with capable hrdware, it would not be impossible.