Chipset for ClawHammer

[Prof.Wizard]

I was wondering what chipset would you use when you'll buy your first ClawHammer system:
The performing, reliable, but expensive AMD 8000?
The versatile, cheap, and feature-full K8HTA?
Or something else?!

Please elaborate your reasons too.

PS. I don't know if the above chipsets will be the analogs of AMD 750 and VIA KT133/A, but let's say they are...

[Tea]

First system, and the other three or four that will follow within the next month or two: AMD chipset.

Then I'll switch to the VIA K8HTA/A and ship hundreds of them :)

Rule of thumb: VIA's early implementations need care. Better to spend the extra. But they get them right after about three months, at which point they are excellent.

SiS and Nvidia ... no thanks. Last time I bought SiS Athlon boards they were utter crap. (Probably the board maker though - I daresay that an Epox or an FIC could get it right. Those ECS things were utter pox. I've sent them both back for RMA: they are unsalable.)

[Prof.Wizard]

One thing I've noticed about the VIA solution is they will be using their own V-link and not AMD's HyperTransport.
Dunno which is better, but the second received a lot of hype...

[Prof.Wizard]

One thing I've noticed about the VIA solution is they will be using their own V-link and not AMD's HyperTransport.
Don't know how true is this exactly...

From Anandtech:
V-Link or HyperTransport?
While visiting DFI we noticed that they had a Hammer board design with an AMD AGP 8X controller but with an ALi South Bridge. When we asked them why they chose the odd combination we realized something we hadn't thought of before - motherboard manufacturers won't be able to use VIA South Bridges with AMD AGP controllers for Hammer. The reason being that AMD's chipset uses HyperTransport to connect all of the chips while VIA uses their own V-Link architecture which isn't compatible with HyperTransport.

but then again VIA says:
All K8 series chipsets feature HyperTransport™ technology, AGP-8X, Double-Speed V-Link and USB 2.0 delivering exceptional system performance and stability. The K8 series will also include integrated 2D/3D graphics chipsets enabling OEMs and System Integrators to create solutions at every price point.

What's happening?! :-?

[Buck]

Where is the option for NONE? I'd prefer to stay away from Clawhammer for a while, and let everyone else (particularly OEMs) find the bugs in all of the chipsets.

[Prof.Wizard]

LOL. Maybe true...

[Pradeep]

All AMD for me. Very happy with my TigerMP and want to continue the experience. Will there be SMP versions?

[Prof.Wizard]

So, no one trusts a 1st-try chipset from VIA, eh?

[Rocco Siffredi]

Nope... though it is said that they have found what the cause of the problems with Creative cards and various others is.

[Prof.Wizard]

http://www.viaarena.com/htmlimages/computexk8hta.jpg

[LiamC]

No VIA for me (well at least not in the first pass).

VIA does use HT and V-Link in it's chipset implementation.

In traditional (VIA) chipsets, the north bridge (NB) connects to the south bridge (SB) via (ha ha) V-Link. Other makers use their own variant (SiS) or the PCI bus (AMD 760, ALi MAGiK1).

With teh Clawhammer, things change somewhat because there is no traditional NB - but there is an AGP device.

VIA is using HT to talk from the CPU to the AGP device and then V-Link to their SB. This would have saved them development costs.

In all other chipsets (AMD, ALi, SiS, NVIDIA), it's HT all the way. This means that you could potentially use whichever AGP Tunnel (NB) and whichever SB gave you the features and price points that you desired. VIA seem to have cut themselves out of this market.

Note: I don't guarantee that it possible with all combinations. Some motherboard makers found out about design quirks when they wanted to develop MP Athlon chipsets with VIA SB's.

AS for the VIA PCI issue, I have been somewhat active on the Tech-Report BB about this issue.

1. Bus Parking is an Intel extension to the PCI 2.1 spec.
2. The core issue is whether support for Bus Park is required. From discussions with other posters more familiar with the PCI 2.1 specs than I, it seems that it is optional.
3. If point 2 is true, then card manufacturers should check for support before implementing. This points the foul finger of fate at card manufacturers.
4. If point 2 is not true, then VIA cannot claim (full) PCI 2.1 compliance.

This discussion should get interesting now this has come to light...

[Tea]

Indeed, LiamC. Keep us posted please.

[SYROB]

VIA, yea right; and deal with Snap, Crackle and Pop other than in my cereal...

I have had nothing but trouble with VIA... (Super 7, TBird)

Sure , works for you guys...

SYROB

[flagreen]

How about a "My head hurts" option to vote for?

[Platform]

Even as much as I have lambasted Via over the past several years, I still think they will EVENTUALLY be as good as Intel at producing decent solid-performing chipsets.

Maybe in about 5 or 6 years, chipset functions as we know them today, will probably be pulled inside the processor package in order to attain higher integration, performance, and lowering of system cost. The recent Intel stab at this (high integration) didn't fare too well at all. When 0.05 or 0.04 or whatever micron manufacturing finally becomes a reality, the push to attain higher integration will likely come back into the plans of AMD and Intel and whoever else is left in the general purpose microprocessor business.

[James]

Of course Sun has had the PCI chipset built into the CPU for quite a while now.

[flagreen]

It seems like VIA has been at it for quite sometime now and still hasn't got there stability or performance wise. You would think they have already been building them long enough to achieved a little better in each department. Though I think Intel's patents on AGP have hindered them. Maybe there are other patents Intel holds that are holding via up?

[J-Frog]

It seems like VIA has been at it for quite sometime now and still hasn't got there stability or performance wise. You would think they have already been building them long enough to achieved a little better in each department. Though I think Intel's patents on AGP have hindered them. Maybe there are other patents Intel holds that are holding via up?

I've never had a problem with my KT266A chipset :D . I won't be replacing it anytime soon, so I'll wait and see if any bugs are found in the new chipsets before upgrading.

[Prof.Wizard]

I've never had a problem with my KT266A chipset :D . I won't be replacing it anytime soon, so I'll wait and see if any bugs are found in the new chipsets before upgrading.
EXACTLY same situation here. I think I'll go with my ASUS A7V266-E deep into 2003... :wink: (Christmas 2002 :right: Athlon XP 2500+)

[Fushigi]

I've never had a problem with my KT266A chipset :D . I won't be replacing it anytime soon, so I'll wait and see if any bugs are found in the new chipsets before upgrading.I've never had stability problems with my KX-133 based Asus K7V Athlon 800 system running W98 1st ed. It ran 24x7 running SETI in addition to being my main system for some time before I gave it to my fiancee, who also doesn't have any problems with it. It's been used with Tekram SCSI, onboard IDE, Asus V6800 GeForce DDR graphics, Intel & Kingston NICs, USB, from 256 to 284 to now 512MB PC133, etc.

Everything seems to work just fine for me & I'm on the original Athlon VIA chipset. I've updated the BIOS & the 4-in-1 drivers a couple of times, but it's been more to stay current than to fix anything specific. I haven't updated any drivers or firmware in over a year.

I have no particular prejudice against VIA based systems. The next time I buy a new system board, I will consider them along with AMD and any other competitors except ALi, who made the crappy chipset on my A7A266. The A7A is bad enough that it has caused me to also avoid Asus in the future as I am now convinced that they do not care about the quality of their products.

- Fushigi

[Tea]

Fushigi's summary expresses my experience of and feelings about VIA, ASUS and ALI perfectly. Making a moderately sub-standard product is one thing. Charging 50% more than the going rate is another thing. Doing both at the same time is unforgivable. ASUS still make better boards than ECS, in my opinion. But then, so does my grandmother.

[Prof.Wizard]

Doing both at the same time is unforgivable. ASUS still make better boards than ECS, in my opinion. But then, so does my grandmother.
Cool. Tell her to make me one together with those delicious cookies... :)

[Tea]

Ahh. Tis true: she is a great cook, real old-fashioned country style stuff. I didn't know you were a Z-80 man, Prof. :)

[Prof.Wizard]

I didn't know you were a Z-80 man, Prof. :)
A what?!

[Tea]

http://www.redhill.net.au/c-1.html#z80

[Bartender]

I'm surprised that your brain hasn't exploded yet Tea, hanging upside down for so long.

[Tea]

Quite right, Bartender. Perhaps I should take a rest.