VIA Releases Free HyperTransport™ Analyzer Utility for AMD64 Based Motherboards

[funky-d-munky]

http://www.hexus.net/content/press_releases/press_release.php?dXJsX3ByZXNzX3JlbGVhc2VfSUQ9Nzcy

The purpose of this utillity is to benchmark which " AMD64 supporting motherboards currently available feature a below-spec HyperTransport implementation"

Does anyone know which of these new boards are the ones that are cheating?
funky

[The PIT]

I'll make one guess all Via based boards.

[Technoid]

ROFL

[funky-d-munky]

Being MURC that smart-ass reply was to be expected. Actually I was counting on it, for such a baited question.
Any intelligent replies?

[Dr Mordrid]

You do have to admit that the "MURC reply" wasn't without some cause for suspicion, right?

Dr. Mordrid

[Wombat]

It actually was an intelligent answer. Via hasn't done a good job of implementing a high-bandwidth protocol since...uhh...can anybody help me out here?

If they get it right this time, it's because AMD gave them a working reference implementation and said "use it."

Also, let's take a look at this...

VIA’s unique Hyper8™ technology incorporated into VIA K8 Series chipsets eliminates signal interference on the HyperTransport bus to ensure maximum signal quality, removing the need to scale down bus speed or data package size and ensuring that motherboards based on VIA K8 Series chipsets feature a full-specification 16-bit/1.6GHz HyperTransport link to AMD64 processors.

Translation: "We finally learned about this shielding thing the other day, " says VIA rep.

Also, there are very few companies where I would trust their utilities to judge their products fairly against those of their competitors. Via, being of low moral fiber, is definitely not on that list.

[funky-d-munky]

This is becoming a bunch a biased crap. With nothing being said based on fact.
What if VIA gets it right with this design, and NForce3 is the KT133 of the new era? Are ya still gonna slam VIA? I am really looking forward to getting one of the A64 set-ups in the new year. And it won't be a piece'o'crap as I will do my homework and get the best. This was one of the first steps in that direction. When I have questions, I ask. And the best replies have usually been here. Thats why I come to the MURC.
Yes I expected some biased no thought replies, but I usually get a few above average intelligence replies as well.
Do I need to find the NAVMURC (Non Anti Via Matrox Users Resource Center) to have an intelligent conversation?

[Kurt]

it's probably a bit early to ask which is the good K8 solution...

[DentyCracker]

ALi is who they are targeting is my opinion

[magician]

I would append just this: "ashes to ashes dust to dust".

[GT98]

Originally posted by funky-d-munky
This is becoming a bunch a biased crap. With nothing being said based on fact.

Well 99% past performance is normally indicative of future performance.

last time I checked Via still has problems with high bandwidth problems on their PCI bus, unlike SiS or Intel. If you want more info on this problem just check out the desktop Video forum.

I had one Via based Motherboard and it was my last if I can help it. I might look at it them again if they get some positive reviews/buzz from fellow users. Theres only one company that I know of that turned shit around that used to be the bain of many people....ATI.

[Dr Mordrid]

Originally posted by GT98
Well 99% past performance is normally indicative of future performance.

Exactly, and if you want to get the rundown on why the bandwidth issue exists just read this article from TecChannel;

http://www.tecchannel.de/hardware/817/index.html

The key qoutes being;

If this is so, why do VIA motherboards only give you 64 to 90 MBytes/s effectively? The answer lies within the many breaks these systems take during bursts.

Normally, a burst should be performed continuously and without any interruption. The above pictures shows a burst like this, recorded using the Promise Ultra133 TX2 and and Intel-845-motherboard. The whole burst here is performed correctly. The length of a burst, however, can differ depending on the data transferred.

With VIA boards, this high speed transfer from the cache of a hard drive is constantly interrupted within a couple of µs. It has then to be re-initiated, as the next picture shows. Therefore, the effective burst rate drops to 64 to 90 MBytes/s at best.

In contrast to chipsets from Intel, SiS and Ali, VIAs products seem to have difficulties with maintaining high transfer rates close to the maximum speed of PCI for a longer time. As with ATA/133 PCI has a theoretical maximum bandwidth of 127,2 MBytes/s. It seems fair to conclude that VIAs implementation of a PCI bus must be the reason for the problems found.

By use of a PCI probe with our logic analyzer we also checked what was happening within the PCI bus during theses transfers. With an Intel 845 based system we found a constant flow of data, whereas with VIA boards these transfers are interrupted all the time.

To take these scores, we made Maxtors DiamondMax D740X work with the Promise Ultra133 TX2 in burst mode.

(image): VIAs PCI bus: This bus of a P4X266A system transfers only 24 packets of data, after the initial address was sent.

After these 24 packets are transferred, a new address gets requested. The PCI bus can only transfer blocks of 96 bytes of data within one burst with this solution.

(image): Intels PCI bus: After the target address is sent, 1024 blocks of data are transferred. Only then the next address gets requested. This makes Intels PCI bus more efficient than VIAs.

With Intels chipsets being able to transfer 1024 packets within one burst, this makes a whole transfer of 4096 bytes. As a consequence, Intels chipsets gain a higher transfer rate overall on their PCI bus than VIAs chipset.

VIA: 96 bytes/burst

Intel (and others): 4096 bytes/burst

'nuff said since VIA's PCI bus handling doesn't appear to have changed since these tests were done on the KT-133. I say "doesn't appear to have changed" because every benchmark I've done on a VIA chipped mainboard from the KT-133 on shows the exact same problem with PCI bandwidth.

This problem is also exhibited in mixed AMD/VIA mainboards where an AMD northbridge is used with a VIA southbridge, which makes sense since PCI is a southbridge function.

This plus many hardware manufacturers (Matrox included) stating that VIA's PCI bandwidth problem still causes problems with their products should go a long way towards confirming that the problems found by TecChannel persist in current VIA products.

@funky-d-munky: biased my arse. Experience my man....experience, and my experience is not to use anything VIA comes out with before it's been thoroughly tested by someone that knows how....marketing BS be damned.

Unfortunately most review sites don't know how, don't want to take the time or are too comprimsed to properly evaluate PCI busses, which is really unfortunate for users.

Dr. Mordrid

[DentyCracker]

Doc, it IS biased. But justifiably so. Bias is not always a bad thing you know.

[spadnos]

@Mordrid:

If you look at the third to last page (titled "Update: VIAs official patch, Version 1.04"), they mention that the performance is drastically improved with the patch. Looking at the new results, the transfer rates are on par with the fastest of the Intel chipsets.

They do mention that improvements in stability have yet to be proven, although they also mention that they saw a bunch of stability / incompatibility issues go away when they installed the patch. (Oddly, they never mention just what actually gets patched ... )

In my own experience, I don't remember any problems that ended up being the chipset (and unsolvable - I may have had some troubles that got fixed with a driver update or something). I may have experienced less-than-optimal performance, but have unfortunately not had the resources (read: money) to buy several machines and test side by side.

For what it's worth, I'm running an AOpen AK73 Pro-A based on the KT133A, a Soyo Dragon Ultra Platinum based on the KT333, an ABit AT7 Max (also KT133), and a SuperMicro P6DNH based on the Intel 440FX (not a fair comparison to the Athlons ). My previous system was an AOpen AX-59 Pro (MVP3 chipset), and I'm sure I've had other Via based systems.

Maybe I've been lucky.

- Steve

[Wombat]

Originally posted by funky-d-munky
[B]This is becoming a bunch a biased crap. With nothing being said based on fact.

Fact: MVP2 sucked, MVP3 sucked until the CE revision. KT133 was average, KT133A/686B combination was awful.

Fact: VIA <B>lied</B> about the 686B having bugs. Scope traces proved it. VIA ignored George Breese and denied the bug right up until they put a similar fix in their own drivers. (Bastards. I couldn't burn a CD for months). On that matter:

Fact: VIA 4-in-1s hosed more systems than I care to count.

Fact: Right now in GH you'll see posts about VIA's bugged 1394 controller, which they've "fixed" but not noted revision changes on.


This is a short history. This is more about bugs, and excludes most "working, but subpar" products from VIA. For a more exhaustive listing of VIAs failings, see the last 3 years of the General Hardware forum.