First nigger

Anyone seen this yet on the Pinnacle site? Seems related to what we're discussing here:
http://www.pinnaclesys.com/support/display.asp?FileID=633&ProductID...

to #57
"If on the other hand all you do is play games and occasionally surf the web, then it really isn't a problem. Unless you are simply un-happy that the problem exist at all. "

yes.!!! i m unhappy about the bug exists.
and it has existed for a long time.

since pci is not a brand new technology, and other manufacturers can do it right, there is no reason via can't do it right.

so if via says it supports pci 2.1 and ata100. i expect it to work within specifications, at the least.

While I really think VIA needs to fix this problem, I find it funny how everyone is claiming that suddenly the VIA hardware they have is shit.

Unless you are doing heavy duty sound work or plan to install a ATA133 RAID 0 array in your machine, this bug doesn't seem to make a damn bit of difference for what most users do.

My machine still represents the best price/performance ratio that I can find for what I do with it. The performance characteristics of my machine didn't suddenly change when this happened. That is what I am concerned with, not some potential limitation that might occur with the hardware that is a year or two out in the pipe-line.

I'm giving VIA a grace period in which to handle this problem. If they correct it in a new revision of their boards, fine. If they create a BIOS patch for a work-around, fine. If they ignore the problem they go on the black list.

Until that happens, VIA is still a valid option for anyone who isn't doing a specific application or configuration that stresses their PCI bus.

i always hated VIA in the past, but now i thought they were top notch - until i was made aware of that crappy PCI bus.

VIA isnīt cheap anymore. two years ago, intel mobos were quite more expensive than others, so everybody should have known what he does when buying via. nowadays, e.g. the ABIT BD-7 RAID and the KR-7 RAID cost about the same. so i expect the same quality of the chipsets.

hell, i told my father to buy athlon & KT266A. now i warned him not to do so, because i can see where it leads with RAID, NIC, dolby 5.1...

with prices being equal, there is no excuse for this. just go the via site and see how they tell you they make the fastest and the best chipsets.

of course, if you buy a chrysler, you cannot expect BMW quality. but if you decide not to go for the BMW but for a mercedes for the same money, you can expect the same quality

Ryu, redpriest:
Thanks. Seems Intel & AMD are the ones to place more trust in given the public policy they have on their chipsets. Will definitely give them more weight in future purchases after this VIA thing.

i know ryu kinda covered this in the beginning of the thread, but still, i would be nice to see some hard numbers using the same test methodology...
Ngoc-Huong Nguyen

question about that article. in the test setup is states they tested an amd 760 mobo, but it doesn't show up on the burst chart listing all the chipsets. what gives? does this mean that the amd 760/761 are or are not effected? inquiring minds want to know. :)
Ngoc-Huong Nguyen

[q]A few small comments, my little fan to the flames...[/q]

Danger, danger! Danger Will Robinson!

[q]Wasn't the Rambus giveaway to push that memory architecture, and so that Intel could reasonably compete with AMD?[/q]

I'd never heard that theory. With the choice made not fix the MTH and with the i815 still being tapered out, VC820 + RDRAM was the only viable alternative.

I'm not sure any company is so generous as to create a defective product so they can give the consumer a better one free.

[q]If you buy a product knowing most of it's quirks and features, do you have a right to be upset when something new and undiscovered comes out? [/q]

I don't see why not. Given that three of the five core logic manufacturers don't document their errata it's hard to walk into a motherboard purchase and no what kind of quirks you'll be receiving. Features are usally touted if performance related and clarified if they have conditions to operation. Like the i815 and nForce bank limits for example.

[q]An example: If you bought a low-end car (yeah, you said they were different industries, but bear with me) knowing it was a low-end car, would you be uspet if you found out six months later that there was a problem with the brakes not performing as well as the high-end car's braking systems? What if the difference in brakes between the cars was equalized (i.e. PCI spec) and the cheaper car still didn't stop as well?

This may be different (and completely impossible to follow) but the point I'm making is this: even if both vehicles had the same brakes, but the expensive one stopped better, would you get upset? And isn't the issue of stopping your car more important to people than PCI bus performance?[/q]

I think your basically trying to mirror TwoFer opinion on price affecting QA, which of course [ihoul/i] result in a better product.

There are two fundamental problems with that concept though.

1. Reality doesn't tend to always fit.
2. Computers are in a constant state of increasing in power and depreciating in price.

The profit margains in computers will always be tight and their will be an ever increasing demand or desire for a certain level of proper operation/stability while the products value is in freefall. It's also important to note that there is a increase in complexity with this cycle.

So, in short, yes, the users are entitled to complain as the economics of PCs are not the same as other industries.

BTW, about point 1. i was on crack and it completely agrees with you...so i apologize. :) but i still think your analogy needs work.
Ngoc-Huong Nguyen

well, i'll rebutt. the brakes example doesn't work. a computer standard is different from brakes. lets assumes we are looking at disc brakes. the general principle of how disc brakes work is the same between the cars, but the actual cost of the brakes can very depending on how much money is spent in construction. i'd say your comparison is more accurate if you consider the whole motherboard and the whole brake system, and then say this:
given on two seperate brake systems( motherboards ) i use the same material - in this instance say what the rotor is made out of - the material with behave the same. HOWEVER, the pci bus is comparable to using the matrial, not the actual implementaton of the motherboard/brake system as a whole. Ryu is right, the pci bus should work! he's not lambasting via for crappy socket holders, raid implementations, or anything else. what he is saying is that via did NOT use the same material ( going back to the brake analogy ) as the others - hence we got a crappy mobo.
it would be like buying two rotors - both claiming to use the same rotor material, but one suffers significant erosion and brake fade. thats what is happening. in your example, yes i would be pissed.
1. brake system is vital to car operation and more importantly can cause risk of death and dismemberment if they don't work properly. if you want to have one good thing running one your car, make sure its the brakes. not the engine,the brakes.
2. as i tried to say earlier( hope i made sense :) ) the performance of the brakes, even given the same rotor material, can very dramatically. how large were the calipers? was is a single caliper system or multiple? were the rotors slotted? did the car have rear disc or rear drum? how large is the bake master cylinder? all these and more factor into this.

i wrote the last part at the risk of souding like an jerk who is just trying to pick apart an argument, but if you have worked on cars you'll recognize these issues. all i am saying is that you can't treat the pci bus like a brake system.
Ngoc-Huong Nguyen

Ah, I try to post intelligently, but the fire has already died...

A few small comments, my little fan to the flames...

Wasn't the Rambus giveaway to push that memory architecture, and so that Intel could reasonably compete with AMD?

If you buy a product knowing [i]most[/i] of it's quirks and features, do you have a right to be upset when something new and undiscovered comes out?

An example: If you bought a low-end car (yeah, you said they were different industries, but bear with me) knowing it was a low-end car, would you be uspet if you found out six months later that there was a problem with the brakes not performing as well as the high-end car's braking systems? What if the difference in brakes between the cars was equalized (i.e. PCI spec) and the cheaper car still didn't stop as well?

This may be different (and completely impossible to follow) but the point I'm making is this: even if both vehicles had the same brakes, but the expensive one stopped better, would you get upset? And isn't the issue of stopping your car more important to people than PCI bus performance?

no ag 78! thats why i sign my name at the bottom. i was too lazy to sign up ( had to go to a party, but was really interested in this thread ), but i signed my name. i stand by my comments.
Ngoc-Huong Nguyen

[q]I don't mean you've won -- I just mean that if you don't stop soon, you'll go blind. ;)[/q]

Go to hell is more likely. ;)

AG #78: There isn't a line drawn in the sand. There are quite a few people who disagree with my opinion.

Then again I'm not a moderate.

Nonetheless good luck and I understand.

[q]Finally. Yeesh.[/q] I don't mean you've won -- I just mean that if you don't stop soon, you'll go blind. ;)

hey ryu, just wanted to mention, in post 67, i hope i wasn't the person. i just was saying in the game benchmarks and other non-harddisk intensive stuff, the pci problem does not appear to be evident in the vast majority of the reviews done on via chipsets. however, i readily admit the validity of the stuff that these guys did, otherwise why would i get huffed about it? just trying to set the record straight. i don't want to make it seem like i am ok with what via did. hell, i'm still trying to troubleshoot this piece o crap asus k7v using the 133 chipset. i just wanna toss the whole mobo...
in case my earlier statements seem vague i'll make it simple:
i agree with ryu
Ngoc-Huong Nguyen

[q]Have fun by yourself...[/q]

Finally. Yeesh.

[q]These replies are getting rather long. . .[/q] Then let me shorten this one.

You've not given a substantive answer yet to my major point, which was first made long ago in the QA post; instead, you've pointed to a handful of special cases and insisted that's how reality works -- or at least, [iught[/i] to work. And you persist in avoiding discussion of other commodity markets which are in fact reasonable models of this one.

I'm tired of playing; this isn't enjoyable, and I know full well I'm not going to change your mind (and to think [io/i] called [i]m/i] stubborn!). But I think I've presented a reasonable argument about what might be going on in this case, and it's here for the more reasonable to ponder.

Have fun by yourself...

AMD discloses bugs in their chipsets and processors.

redpriest

These replies are getting rather long. . . .

[q]Do you honestly think that all warranties should be equal? Or that everyone should bear the cost of a warranty whether or not they want it?[/q]

If the board manufacturers screws up and their product ends up defective, then the board manufacturer should eat the cost. If the chipset produced by VIA is screwed up, then it's VIA's job to compensate the manufacturers. They should not have to take a hit to their bottom line for VIA's mistake. Yeah, Intel is huge, but part of the reason they are as big as they are is because they do the above.

In the case of a part which is under performing I highly doubt a recall will occur. Hell, I'll be amazed if we even see a patch; I'll explain that more later. If the part has data corruption, that's a slightly different story. The 686B sits extra gray because only one sound card was demanding enough to exacerbate the issue. Leaving the tough question of patching it or replacing it.

[q]While that may be true for some products, it most emphatically is not always true of commodity hardware -- which is precisely what we're discussing. As I said, I suspect you're not paying close attention to the market realities.[/q]

Well, I guess that might explain why I didn't mind paying $200 for a motherboard. I believe your saying that VIA is a bottom rung provider. To repeat what I said earlier, "It's like they think that their target audience is still MVP3 users." And god knows that the MVP3 was targeted at the cheap end.

I still don't much agree with allowing VIA more leeway because of their target audience. Intel and AMD provide products near the same price range and their quality appears substantially higher. The Epox 8K7A was obtainable for $99. I'm quite aware of market realities. I just know junk versus quality when I see it.

[q]Yet Herman suggested it may have to do with optimization for competing services using the PCI bus... my point is that you've closed the door prematurely.[/q]

Maybe. I still find it somewhat fascinating that Intel can move 1024 packets and still find that ability to properly arbitrate the bus, yet VIA only has 24 packets going out and causes PCI cards to suffer and performance to lag. Their professional sound card is a great example.

You will also notice that George's tweak disables many items that have priority over the PCI bus and even still he can't get it much higher than 32 packets.

If it turns out to be something simple to fix I'll be more than happy to share it with the world and announce how amazed I am at the way VIA did something cleverly. Unfortunately I don't think that's going to happen.

A. They never responded to our e-mail about the bug.
B. Their website is devoid of info.
C. They are locking threads about it.

http://forums.viaarena.com/messageview.cfm?catid=6&threadid=5402

Hmm. It seems they removed the lock. Just so I don't look crazy, I'll show another indepent confirmation that they had locked the subject.

http://arstechnica.infopop.net/OpenTopic/page?q=Y&a=tpc&s=50009562&...

[q][i notice the via admins locked a thread discussing the PCI performance on www.viaarena.com[/q][/i]

C. They were locking threads about it.

Nonetheless, it doesn't bolster my confidence much. It shouldn't be all that hard to have responded back to us with, "We're working as hard as we can on a fix."

No, we just got dead air.

[q]Yet there are plenty of cars which are either (and sometimes both); does that make them worthless trash? Obviously not to the people who want, buy, and enjoy them...[/q]

I'm not doing the car thing.

[q]But you're still (or again) ignoring the realities of a commodity market. Nvidia got their advantage from raising the bar without significantly raising the price -- that, in a field which was ripe for those advances. But the basic mobo chipset market was fairly stagnant for a while, with many companies competing for a slice -- not the sort of situation which Nvidia faced, at all.

Forcing your desired quality level on the entire market would only raise the prices for everyone, including those who don't need and don't want the higher quality levels you demand. Seriously: are you prepared to allow me to demand everyone use pure-SCSI architecture simply becuase I like it, can afford it, and enjoy using it? IIRC you gave me hell simply because I dared to use and defend the stuff in the first place! :) <---note smiley![/q]

S3, ATi, 3Dfx/3dfx, NVIDIA, PowerVR, and that other company whose name eludes me. I think Creative even released a 3DLabs product at one point during it all.

I'd say that NVIDIA was competing in cut throat times and this was before they decided to give the industry a new sort of pace to work at. In other words, it was stagnent early on for them as well. They were the ones to give life to the product cycle that exist today.

Moving forward in the disucssion you'll remember that I quoted that the Epox 8K7A was $99. How exactly is that raising the price for everyone?

And it's not too bad a chipset. Seems some of the early steppings had some AGP problems. Not sure if those steppings were released to market or not. As one bug requires FastWrites to be disabled and I don't think I have seen a 760 board that doesn't provide FastWrites as a option in the BIOS.

http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_...

Of course mating it with a 686B creates unique problems that wouldn't normally exist with a 766 south bridge instead.

http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_...

From the 760MPX guide:

http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_...

[q]48 AGP Transactions are Non-functional

Products Affected. B0

Description. AGP cycles are non-functional in this silicon revision due to a silicon development tools issue.

Potential Effect in System. The system may hang when attempting to use AGP cards.

Suggested Workaround. AGP cards can still be used in PCI mode (including PCI mastering) but actual AGP cycles do not. To use AGP cards, do not load the AGP miniport driver in the system. This will force the AGP card to run only PCI cycles.

Resolution Status. This will be fixed in a future silicon revision.[/q]

AGP must be a serious bitch to implement. I swear, it seems all the companies go and *&^*&6 it up. Of course this B0 stepping may not be on the shipping 760 MPX boards.

So much for extra dollars equalling extra quality, eh? :)

Well, any ways, moving on.

[q]. Seriously: are you prepared to allow me to demand everyone use pure-SCSI architecture simply becuase I like it, can afford it, and enjoy using it?[/q]

If everyone used it, it's price would reach a mainstream level. Of course the fact I can buy a IDE drive (1000BB-SE) that places within 80+% of the X15 36GB drive for half the price and three times the storage, I'm not sure why people bother with SCSI except for servers and work that involves next to zero locality.

For non-servers the price to performance ratio on IDE is crushing SCSI.

http://www.tech-report.com/onearticle.x/3220

So much for extra dollars equalling superior performance, eh? :)

[q]My point is you're insisting it's a "bug" while it may be nothing more than incomplete use (for reasons other than oversight) of the allocated bus throughput; but if you look at other commodity PC hardware -- take modems, or NIC's -- you'll find that's common practice. And people are aware that you generally get what you pay for, which is what keeps both ends of the spectrum in business.[/q]

Maybe. Then again you might pay twice the price and only get an extra 20% or less of performance against something half the price too.

That or twice the price for a motherboard that can't do AGP.

[q]This still applies, BTW, whether it was oversight or not... there's no law that every ounce of performance must be squeezed from all hardware. Hell, in my own business I sometimes specify commercial-grade materials for parachutes instead of the more-expensive MIL-SPEC alternatives -- sure, there is often a difference in ultimate strength or other properties, but cost is generally important and if the commercial-grade stuff is adequate (which was my point in the QA discussion) why throw the money away? Should Matrox recall all their older graphics cards because faster memory has come out? Hell, no -- the design implemention worked at the time, no matter whether it used the entire theoretical bandwidth or not -- and regardless of later changes in the market standards.[/q]

Umm, failing to implement an ancient industry standard that's been at 133MB/s (for what close to a decade now?) is nowhere near the same as saying that NVIDIA needs to take my GF3 back and give me a GF3 Ti500.

And while in your business it might be okay to shave performance down a little bit, I thought the point of the PC was to progress forward in progress? Not sit around wasting cycles. Better, faster, harder, stonger as my fave Daft Punk song would say.

[q]And for a PC-related example, why don't you condemn Nvidia for not building Matrox-quality 2D into their products? That's one which doesn't rely on an automobile analogy![/q]

Because you're one of those invididuals who believes NVIDIA is to blame and I'm one of those individuals who think the 3rd party manufacturers cut corners. So I don't see much need to blame NVIDIA.

[q]If that is all that's bothering you, why don't you condemn Nvidia and Microsoft for their lying and misdirective behavior? They've both been caught often enough at it...[/q]

If I could think of an example, I might would. I don't really keep up with MS scandals; it's generally boring. It's like watching extreme left and right go at in politics. As for NVIDIA the worst I can think of is Derek Perez not keeping his word; and often at that. Unfortunately I can't get Derek fired anymore than I can get VIA to hire some competent engineers. Let's also not forget that Derek's company doesn't push a shit product. While VIA, well VIA needs to hire some competent engineers. Furthermore I don't see much point in firing Derek for being only a liar. Where as VIA is a liar and needs to hire some competent engineers.

[q]No, it was because none of the dozens of tech website reviews of the product revealed the problem -- and many thought (up to today, anyway) that the VIA chipset mobos were the best you could buy (go check ol' [H]ard Kyle, f'rinstance). Even here on TR there was absolutely no indication that the VIA PCI bus was somehow limiting performance in the benchmarks.[/q]

Don't forget that the hub architecture allows the IDE controllers to work without using the PCI bus. For the purpose of seeing if this bug has a real world impact, pretty much all the web reviews are wrongly geared.

And with older chipsets like the KT133A, the TBird and mostly CPU intensive benchmarks likely carried them across the goal line.

I detect a pregnant chad and I want a recount.

[q]IIRC I pointed out long ago that RAID 0 is one of the few places where the chipset shows its limited throughput -- and IDE RAID's acceptance is newer than the VIA PCI implementation, as is IDE133 and the increased STR of todays larger hard drives. We've finally reached a point where the problem showed up, and it's time for VIA to address it.

I'm not denying the existence of the limitation; I just think that calling for them to "die a horrible death" is a bit over the top...[/q]

Actually I said, "or at least suffer a nasty recall to whip their ass into shape."

Not that them going out of business would break my heart. I'll settle on the above though if it will get them to think twice before they ship a glaring bug.

[q]Here's another example of deliberate use of lower-cost, lower-performance hardware: the Intel SDR i845 mobo. And one of apparent under-utilization of possible performance (for reasons still unknown): the Intel DDR i845 mobo (Damage even speculates on it in his review).

But I doubt you have a major problem with either of those... is this a consistency issue? And even if it is, is it worldshaking? I think "yes" to the former and "no" to the latter. [/q]

The i845 is a mainstream targeted product. Despite it's lower performance I still expect a quality implementation. The Intel i845's lower performance is manufacturer level issue. Chances are though that the Intel board does not allow for manipulation of memory setting and it being an Intel board it queried the SPD and took on the lowest possible speed settings. Many OEM targeted boards do this, I do think it's quite lame, but seeing as how OEM's cut corner and use the slowest RAM they can find or overclock. Just like my Apple story that I can't find in the archives at the moment, I suppose the feature serves a purpose to protect Joe Average.

Of course Joe Average can always buy better RAM. It's sort of hard for him to buy a better PCI bus without changing mobos.

Here's another example of deliberate use of lower-cost, lower-performance hardware: the Intel SDR i845 mobo. And one of apparent under-utilization of possible performance (for reasons still unknown): the Intel DDR i845 mobo (Damage even speculates on it in his review).

But I doubt you have a major problem with either of those... is this a consistency issue? And even if it is, is it worldshaking? I think "yes" to the former and "no" to the latter.

[q]VIA has their own product division now I believe.[/q] AFAIK they're simply reselling badged versions manufactured by others. [q]I don't see where that somehow alleviates them from honoring their product.[/q] Do you honestly think that all warranties should be equal? Or that everyone should bear the cost of a warranty whether or not they want it? [q]One can build a product of equal or better quality for less.[/q] While that may be true for some products, it most emphatically is [iot[/i] always true of commodity hardware -- which is precisely what we're discussing. As I said, I suspect you're not paying close attention to the market realities. [q]Optimization? 24 packets of data compared to 1024 packets hardly seems like an optimization to me.[/q] Yet Herman suggested it may have to do with optimization for competing services using the PCI bus... my point is that you've closed the door prematurely. [q]As for quality level, I have a hard time justifying buying cars that aren't practical or fuel efficient.[/q] Yet there are plenty of cars which are either (and sometimes both); does that make them worthless trash? Obviously not to the people who want, buy, and enjoy them... [q]So, yeah, I wonder why every company doesn't aim to meet high standards and then work to differentiate or change that great product so that it will fit a lower price segment. That's just smart business...

Hell bare minimum my quality level would make computing in general more enjoyable for everyone; even if they only utilized a tenth of it.[/q] But you're still (or again) ignoring the realities of a commodity market. Nvidia got their advantage from raising the bar without significantly raising the price -- that, in a field which was ripe for those advances. But the basic mobo chipset market was fairly stagnant for a while, with many companies competing for a slice -- not the sort of situation which Nvidia faced, at all.

Forcing your desired quality level on the entire market would only raise the prices for everyone, including those who don't need and don't want the higher quality levels you demand. Seriously: are you prepared to allow me to demand everyone use pure-SCSI architecture simply becuase I like it, can afford it, and enjoy using it? IIRC you gave me hell simply because I dared to use and defend the stuff in the first place! :) <---note smiley![q]That's somewhat different than a product that has bugs, is forced into deprecation by a new model, and thus remains with those bugs forever because it was deprecated by a respin.[/q] My point is you're insisting it's a "bug" while it may be nothing more than incomplete use (for reasons other than oversight) of the allocated bus throughput; but if you look at other commodity PC hardware -- take modems, or NIC's -- you'll find that's common practice. And people are aware that you generally get what you pay for, which is what keeps both ends of the spectrum in business.

This still applies, BTW, whether it was oversight or not... there's no law that every ounce of performance must be squeezed from all hardware. Hell, in my own business I sometimes specify commercial-grade materials for parachutes instead of the more-expensive MIL-SPEC alternatives -- sure, there is often a difference in ultimate strength or other properties, but cost is generally important and if the commercial-grade stuff is adequate (which was my point in the QA discussion) why throw the money away? Should Matrox recall all their older graphics cards because faster memory has come out? Hell, no -- the design implemention worked at the time, no matter whether it used the entire theoretical bandwidth or not -- and regardless of later changes in the market standards.

And for a PC-related example, why don't you condemn Nvidia for not building Matrox-quality 2D into their products? That's one which doesn't rely on an automobile analogy![q]As for lying, they have never been very truthful.[/q] If that is all that's bothering you, why don't you condemn Nvidia and Microsoft for their lying and misdirective behavior? They've both been caught often enough at it... [q]And? You act no differently. [/q] When I complain about a product (PowerLeap's stuff is a great example) I clearly state that it's my opinion, based on my experiences; I don't recall insisting that they "need to die a horrible death" or anything so draconian! I just say I won't buy it, and I don't recommend it to anyone.

If you left your statement as it was in #57 I don't think we'd have much room for argument: [i]"If on the other hand all you do is play games and occasionally surf the web, then it really isn't a problem. Unless you are simply un-happy that the problem exist at all."[/i]

As I said earlier, I think it bothers you that I don't join you in your blanket condemnation of VIA (and others, for that matter). If it's absolute consistency you demand, let me point out that two of my favorite quotes are from Emerson: [i]"A foolish consistency is the hobgoblin of little minds"[/i] and Whitman: [i]"Do I contradict myself? Very well then I contradict myself, (I am large, I contain multitudes)."[/i] [q]What, because you found one person quoting that they can't see the performance impact in a series of non-disk limited/PCI bus intensive benchmarks?[/q] No, it was because none of the dozens of tech website reviews of the product revealed the problem -- and many thought (up to today, anyway) that the VIA chipset mobos were the best you could buy (go check ol' [H]ard Kyle, f'rinstance). Even here on TR there was absolutely no indication that the VIA PCI bus was somehow limiting performance in the benchmarks.[q]Why don't you send them over to the Storage Review forum and have them dig through all the threads about why their RAID-0 scores are so damn funky.[/q] IIRC I pointed out long ago that RAID 0 is one of the few places where the chipset shows its limited throughput -- and IDE RAID's acceptance is newer than the VIA PCI implementation, as is IDE133 and the increased STR of todays larger hard drives. We've finally reached a point where the problem showed up, and it's time for VIA to address it.

I'm not denying the existence of the limitation; I just think that calling for them to "die a horrible death" is a bit over the top...

<nod> Yes, that pretty much is it, Ryu. Neither the i815 or the i845 appealed to me based on total price vs. features and performance. I really don't do anything that taxes my PCI bus hard enough to get this problem and my configuration is pretty darn good for general computer use, computer illustration, and gaming.

BTW, I wasn't trying to say that Intel might have put the bug in on purpose, but rather found it late in the QA process and decided to send it to OEMs anyhow.

In any case, I know the example was VERY thin, but I also think ascribing only two possible paths for how VIA's issue came to be is also very thin.

[q]I'll ask again:
Are Intel the only company to publically disclose bugs with their chipsets. Thanks...[/q]

To my knowledge, yes. The only company I am unsure about is AMD.

[q]Intel could very well have been putting the board out so their OEMs could ship SOMETHING, knowing full well they would have to pull them later on. Better that than the risk that AMD might be able to use the delay to get into the OEM market more deeply...[/q]

That's a rather deep conspiracy. Especially seeing that it cost Intel a replacement of each motherboard produced by every manufacturer (Asus, MSI, etc) and a stick of then super expensive RDRAM per replaced board.

I have a hard time buying that. Especially seeing that not all MTH users experienced the bug. If they really did it on purpose I think they would have developed a less costly contingency, like a stepping fix of the MTH (saved them the cost of a RDRAM) or a driver(saved them the cost of replacing any of it). They did neither.

[q]All I know is that Intel's design decisions on the P4, their attempt for force RAMBUS on everyone, and their failure to create another BX-quality chipset has led me to be very happy with my poor, flawed AMD/VIA solutions.[/q]

Well, the i815 and i845 are solid chipsets. I suppose the i820 and i840 RDRAM version wasn't bad. The i850 is up in the air, I guess it depends on if any of your PCI cards do multiple memory address reads or if 90MB/s is good enough in that single scenario.

I understand your point though. RDRAM was too much for no perceivable benefit in quality or performance. The P4 is too much for performance slower than a cheaper and of equal quality competitor. The same level of quality engineering at a price we had become accustomed to had disappeared.

I'll ask again:
Are Intel the only company to publically disclose bugs with their chipsets. Thanks...

[q]But Intel builds motherboards based on their chipsets, and VIA doesn't -- therefore Intel is in a somewhat different position, aren't they?[/q]

VIA has their own product division now I believe. At one time they didn't, but I don't see where that somehow alleviates them from honoring their product.

[q]Besides, it's well-known that warranties vary with the quality of the product -- that's a big reason for the much-longer SCSI drive warranties; and VIA isn't building BMW's like Intel often does...[/q]

One can build a product of equal or better quality for less. Dollars invested does not always eqauate to better quality.

[q]But that hardly exhausts the list of possibilities; for instance, it might in fact be an optimization for something other than raw throughput. You're making an unwarranted assumption and basing a lot on it.[/q]

Optimization? 24 packets of data compared to 1024 packets hardly seems like an optimization to me. Furthermore if that was a way to cheat in benchmarks, I'm sure all the core logic companies would be doing it. Intel used their busmaster drivers to cheat on some benchmarks years ago, so it seems obvious none of them are above it.

[q]Can I assume that you don't drive anything but BMW-class cars and better? Oh, that's right -- you drive a Corolla... ;) Obviously you've made a choice about what's an acceptable quality level for you.[/q]

Yeah, and I bet my Corolla will keep on driving until the wheels fall off. Money does not always guarantee quality. As for quality level, I have a hard time justifying buying cars that aren't practical or fuel efficient. Bringing cars into this is a bad idea as that industry and its offerings don't parallel well.

[q]But you keep ignoring the points about market position and quality levels overall; shall I assume you understand the point, but simply don't care? Or is it that anyone who chooses a quality level lower than you is deserving of nothing but scorn? And of course, ditto to the product itself?[/q]

I am a believer in that you aim high and you let your product trickle down. Market position doesn't have to play into that. ATi is the market leader in graphics cards and their hardware is damn spiff and their drivers suck ass. NVIDIA was the lowest underdog around and they did the above and now they are the #2 company and doing their damndest to best the #1 company.

They didn't have to push mediocre, low priced, or a insanely expensive priced a product to get there either; with perhaps a exception of the G2U, which is a item for seperate debate.

So, yeah, I wonder why every company doesn't aim to meet high standards and then work to differentiate or change that great product so that it will fit a lower price segment. That's just smart business.

If you made a kick ass parachute that would revolutionize the industry and discovered you could scale it down, mass produce it, and it would still have the same quality as your expensive high end chute, but only target a different audience. You'd probably make money hand over fist.

Hell bare minimum my quality level would make computing in general more enjoyable for everyone; even if they only utilized a tenth of it.

[q]Aside from the very significant fact that you're putting words in their mouth and calling them liars for it... Are you willing to condemn Nvidia for releasing a series of improved drivers, not to mention upgraded chips?[/q]

Different market, different demands, different operation, and different expectations. A video card becomes deprecated, not buggy. And if you own a NVIDIA one you can keep applying application level driver fixes until the card's plain inadequate to run.

That's somewhat different than a product that has bugs, is forced into deprecation by a new model, and thus remains with those bugs forever because it was deprecated by a respin.

The only parallel is that the new model brings about more performance, which is an appreciable gain. A graphics chip can't have the same impact as a failing in your core logic. It's like the food chain, core logic is higher than graphics. I can't even think of a graphics card that has been recalled in my years.

As for lying, they have never been very truthful. Remember that warm rest bug in the KT133A Northbridge? Remember how they said AMD was going to make an annoucement about it? I wonder what ever happened to that. I guess they managed to blow enough smoke.

[q]Actually, it's not fair at all: it appears that you're condemning people and products because they fail to meet your own, individual, personal standards-of-the-moment.[/q]

And? You act no differently. Powerleap, older SiS chipsets, NVIDIA 2D, that old Intel Neptune board you hate so much, Microsoft, etc. And it's the primary reason we collide. I think X product is poo, you think I'm being too hard. You think X product is poo, I think you're being too hard. That or we accuse each other of not having the facts, which is just a result of having vastly different experiences.

I see nothing wrong with condeming a product that does not meet my standards. Some people would even call that reviewing a product.

[q]I can't think of a better proof of my earlier statements; you're just ignoring the realities of the manufacturing and marketing worlds, Ryu.[/q]

What, because you found one person quoting that they can't see the performance impact in a series of non-disk limited/PCI bus intensive benchmarks?

Why don't you send them over to the Storage Review forum and have them dig through all the threads about why their RAID-0 scores are so damn funky.

Ryu, I really don't want to get in the middle of this, but I found it interesting that you say that:

[q]If VIA didn't know, then they didn't test. If VIA did know, then they didn't care. Either way, the fact that it carried over from chipset to chipset for years shows they aren't much concerned about anything else than saving some cash. It's like they think that their target audience is still MVP3 users.[/q]

But laud Intel for recalling the badly defective MTH board, a bug that was found by their volume OEMs, not Intel's QA, IIRC. With another, less common manufacturer, that board might have gotten to market.

If Intel didn't know, then they didn't test. If Intel did know, they didn't care.

Intel could very well have been putting the board out so their OEMs could ship SOMETHING, knowing full well they would have to pull them later on. Better that than the risk that AMD might be able to use the delay to get into the OEM market more deeply...

Let's see what VIA's response is to this problem, before we compare it to how Intel responded.

All I know is that Intel's design decisions on the P4, their attempt for force RAMBUS on everyone, and their failure to create another BX-quality chipset has led me to be very happy with my poor, flawed AMD/VIA solutions.

That being said, how VIA response to a problem of this magnitude WILL have an impact on my purchase decisions. If they make a good effort to make it right, then I will remain interested in their products for my home/hobby use.

If they fail to follow through then I will take my changes with NVIDIA and SiS, unless Intel can get out of their doldrums and put out a processor that beats AMD in price/performance AND puts out a solid, DDR-based chipset.