Why do people badmouth VIA

[MrBumpy]

Here is my answer to the question: Why to people badmouth VIA?

Bad experiences. Of course, my bad experiences stem back from the Super Socket 7 days like many others'. It all started with an MVP3 motherboard and crackling audio. I am a musician and there is no way I could have a computer with this problem, so I tried a motherboard based on the ETEQ chipset... crackle crackle crackle... did a little research and found out that the ETEQ chipset on my mainboard was simply a relabeled VIA MVP3! I did a little more research and decided to seek out a motherboard based on the ALi Aladdin V chipset. After installing it, my sound was perfect. I also noticed that my hard disks benchmarked much faster. Now, the ALi wasn't perfect, and had issues with my TNT2 graphics card with AGP 4X enabled (or was it 2X back then... can't remember), but I was able to achieve stability without having to sacrifice any noticeable speed (the loss in speed when using AGP 1X was undetectable to me).

Now, here's the interesting thing: people always bring up the alleged "Creative PCI bug", but completely overlook the fact that other sound cards would crackle as well in certain configurations. My dad's VIA board (early Athlon Slot A) would crackle using the onboard sound during hard disk transfers and such, and we actually solved the problem by putting in a SB-Live! Another interesting thing is that when a device doesn't receive enough PCI bandwidth, it is a lot more noticeable with a sound card than with any other PCI device. Example: if a network card doesn't receive it's requested PCI bandwidth, not as much data is able to transfer--not really detectable. Same deal with a graphics card--you just get fewer frames per second. However, with a sound card, when the sound stream is interrupted, you can audibly hear the gaps in the data transfer... they show up as pops and clicks (now, this is also compounded by the differences in sound card design--the SB-Live! was simply less forgiving to PCI implementations that couldn't grant it's requested bandwidth). The PCI latency article posted earlier in this thread explains very well why early VIA chipsets exhibited this behavior, and possibly also explains why other PCI cards reportedly performed badly in VIA's implementation.

 

[will889]

Originally posted by: Remedy

There is a reason MS chose SIS for the Xbox360 - stability being the first reason (EG non
If you think SiS chipsets are more stable than VIA in the AMD market, then your experience must be very short to speak of.

Actually my experience is extremely extensive so far as building with VIA goes. 250+ enthusiast builds. More than that in server implementations Intel and AMD. If you mean experience as in the engineering aspect then no i don't have that. SIS has been supremely stable for about 4 years running now. Uninspiring? Yes. Enthusiast chipset? No. As for more stable than VIA i never really said that. I started with VIA though, and built with hundreds of them. The problems that were evident had to do in part with VIA 4 in 1's used with paticular boards, and part in the actual implementation of hardware and the bios with a paticular manufacturer/s and PCI latencies. Nothing i could not work with or fix usually. If you count Nforce, VIA, Intel, ULI i would put my overall (hands on) experience up there with anyone. If you're an engineer like Peter is than you may have more insight than myself about VIA vs. other chipsets.

 

[ElJefe69]

Via k8t800 pro chipset on a 754 board is by far the most stable, cool running, low power yet do anything and everything board I have owned. I havent had a better board before or since. Kv8pro. the 4 in 1's work fine. I have no idea what people gripe about. I have used via for a few years. nforce sux. it just does. its hot running, and has bugs that I never encountered on my via boards which all ran latest technology and truly passive.

 

[Peter]

Originally posted by: MrBumpyAnother interesting thing is that when a device doesn't receive enough PCI bandwidth, it is a lot more noticeable with a sound card than with any other PCI device. Example: if a network card doesn't receive it's requested PCI bandwidth, not as much data is able to transfer--not really detectable. Same deal with a graphics card--you just get fewer frames per second. However, with a sound card, when the sound stream is interrupted, you can audibly hear the gaps in the data transfer... they show up as pops and clicks

This is because sound and TV cards are isochronous (they have to keep up with real world time to do their thing), which makes them unique. All other kinds of cards don't mind when their data are a little late.

(now, this is also compounded by the differences in sound card design--the SB-Live! was simply less forgiving to PCI implementations that couldn't grant it's requested bandwidth). The PCI latency article posted earlier in this thread explains very well why early VIA chipsets exhibited this behavior, and possibly also explains why other PCI cards reportedly performed badly in VIA's implementation.

Again, it wasn't the /chipsets/ that couldn't keep up - it was sloppy BIOS engineering that didn't initialize the chipsets for a proper balance between throughput and latency. Getting that right didn't matter in the early days of PCI when cards were few and slow; it started to become important when IDE speeds went beyond UDMA-33 and at the same time multichannel sound cards appeared.
Other chipset makers had an "advantage" here because their chipsets weren't configurable in that regard to begin with. VIA's were and possibly still are, and you had to put a few brain cells onto it to make it right.
As demonstrated in the linked articles, get that chipset programming right, and even those old VIA chipsets suddenly perform exactly just like everyone else's.

 

[3chordcharlie]

I've never had an issue with VIA, having a couple of KT266a boards, a kt333a board, and even an apollo pro 133 board that's still running, but starting to flake out after 6 years.

None of them has been the fastest solution available, but they've all been quite stable, which is all I really care about.

 

[eastvillager]

A lot of it comes from those of us who switched from Intel to AMD back when VIA chipset(s) were the only choice to support the AMD processors. We were coming from Pentium/Pentium2 processors, usually on Intel chipsets, were things "Just Worked". No tweaking, no driver hassles, relatively few compatibility issues with add-in boards like video and sound cards.

I eventually got fed up with VIA and threw out the baby(AMD) with the bathwater(VIA) and went back to building machines with Intel (P4 processors).

Then nvidia got in the game with nforce and I was cautiously optimistic. I eventually switched back to an fx-55 on nforce 4 ultra, and have been fairly happy with the experience. If intel hadn't been so dismal at the time, I probably would've stayed with them because of the past bad experiences with AMD on othe chipsets, quite honestly.

I've seen creative on via listed in a couple of these posts. With my experience of the two, I'm always going to point the finger at creative, even though the fix will often be a bios or chipset driver fix to deal with "creative issues".

 

[aka1nas]

It isn't that VIA makes/made terrible chipsets, its that in the past they tended to need at least one or two hardware revisions of the chipset to "get it right". I did own several via-based boards back in the K7 days and they were all decent but every single one had more than their fair share of quirks and compatability issues. Back then, VIA was pretty low end still and the quality kind of showed. They seem to have improved greatly since then. The last major flaw I have heard was that the early K8 chipset revisions did not have working PCI and AGP locks and I was pretty sure there were some socket 939 VIA chipsets that were incapable of using dual-core CPUs.

 

[Peter]

Missing PCI and AGP "locks" are not "flaws", they're overclocker's features that don't matter zip to 99.99 percent of the overall PC market. Run the chipset at its specified speeds, and you simply don't need them.

Same for "taking revisions to get chipsets right" - you're confusing features with flaws. Sure, KT133 didn't support 133 MHz FSB - but it was never specified to, and couldn't possibly have been, given the complete lack of such CPUs when it was designed. KT133A added that feature. Sure. KT266A brought a faster RAM controller than the original KT266 had - that's progress, not fixes. Etc. etc. etc.

 

[aka1nas]

Originally posted by: Peter
Missing PCI and AGP "locks" are not "flaws", they're overclocker's features that don't matter zip to 99.99 percent of the overall PC market. Run the chipset at its specified speeds, and you simply don't need them.

Same for "taking revisions to get chipsets right" - you're confusing features with flaws. Sure, KT133 didn't support 133 MHz FSB - but it was never specified to, and couldn't possibly have been, given the complete lack of such CPUs when it was designed. KT133A added that feature. Sure. KT266A brought a faster RAM controller than the original KT266 had - that's progress, not fixes. Etc. etc. etc.

I am pretty sure that the chipset was supposed to be capable of it, but early revisions were flawed and the feature did not work as it was supposed to. Moreover, we are talking about enthusiast-level motherboards marketed to customers in the retail channel, not OEM systems, so the "99.8% of the market" comment doesn't apply.

 

[aidanjm]

Nice little VIA K8M890 socket 939 micro-atx motherboard:

A8V-VM
http://www.asus.com/products4.aspx?l1=3&l2=15&l3=0&model=1003&modelmenu=1

 

[RadiclDreamer]

from my experience via either have really good chipsets, which usually have a + in the name and then they have some REALLY bad ones.

 

[Flipped Gazelle]

Originally posted by: Peter
Missing PCI and AGP "locks" are not "flaws", they're overclocker's features that don't matter zip to 99.99 percent of the overall PC market. Run the chipset at its specified speeds, and you simply don't need them.

Same for "taking revisions to get chipsets right" - you're confusing features with flaws. Sure, KT133 didn't support 133 MHz FSB - but it was never specified to, and couldn't possibly have been, given the complete lack of such CPUs when it was designed. KT133A added that feature. Sure. KT266A brought a faster RAM controller than the original KT266 had - that's progress, not fixes. Etc. etc. etc.

Wow, hard to believe this thread is still alive.

If you go back to the original question - why does via have a bad rep? - the .01 % (according to Peter) that is the enthusiast overclocking segment has a great deal of influence over the reputation of a component. This is one of the reasons companies work so hard on top-end products that sell in small quantities - they help establish a "hot" reputation. So even though Peter is technically correct on his points, the "damage", so to speak, to VIA has been done. It doesn't mean VIA chipsets are crap. There is a *perception* that the chipsets are not geared toward the enthusiast market. Of course, all this means nothing to the Walmart/BestBuy crowd.

aka1nas, the KT133 was originally spec'd to run PC133 RAM, not a 133 FSB for the CPU.

 

[Peter]

Exactly ... just like its predecessor KX133 (for slot-A), KT133 was specified for 100 MHz FSB and 133 MHz RAM.

The point about not gearing product toward the small enthusiast crowd is, it drives unit cost up, and everything else too.

There's a reason why NVidia chipsets are so much more expensive and run so much hotter than everyone else's ... too many bells and whistles.