SiS Xabre pipeline shenanigans?
- Thread starter Joe DeFuria
- Start date
At the risk of getting more deeply enmeshed in one of those Russ/Joe things...
Some people took the position that details of internal architecture were irrelevant in deciding on the merits of a video card. I pointed out that you might be interested in details of internal architecture for the same reason you would be interested in the result of synthetic benchmarks: to predict the performance in games that haven't been tested with the card, or even released yet.
The internal architecture and the market uptake of the card will also determine to some extent how future games will be developed.
Surely if you take enough interest in evaluating video cards to read Beyond3d, you are interested in evaluating performance at that level, rather than just saying "it gets enough FPS in UT2003/Quake3/Serious Sam."
So I was merely pointing out that a professed lack of interest in the internal architecture of video cards appeared to be inconsistent with an apparent interest in the many other ways, besides specific game benchmarks, that video cards can demonstrate their performance potential. Beyond pointing out that apparent inconsistency, I certainly didn't mean to be derogatory.
Some people took the position that details of internal architecture were irrelevant in deciding on the merits of a video card. I pointed out that you might be interested in details of internal architecture for the same reason you would be interested in the result of synthetic benchmarks: to predict the performance in games that haven't been tested with the card, or even released yet.
The internal architecture and the market uptake of the card will also determine to some extent how future games will be developed.
Surely if you take enough interest in evaluating video cards to read Beyond3d, you are interested in evaluating performance at that level, rather than just saying "it gets enough FPS in UT2003/Quake3/Serious Sam."
So I was merely pointing out that a professed lack of interest in the internal architecture of video cards appeared to be inconsistent with an apparent interest in the many other ways, besides specific game benchmarks, that video cards can demonstrate their performance potential. Beyond pointing out that apparent inconsistency, I certainly didn't mean to be derogatory.
Joe DeFuria said:
No, it doesn't. What is written on the box will not influence in any way, shape, form, or fashion, the outcome of whatever measurement tool you want to use. The performance is a measurable truth that does not rely on the specification sheet. If what is written on the box is incorrect or misleading, you may have trouble reconciling what you expect based on the specs and what you measure, but it will not affect how the card runs through whatever tests you find important.
You're more than welcome to look at everything and be upset about whatever. I simply don't find the published specs to be that important, and certainly not important enough for me to get in a tizzy over, especially when there are more accurate ways to measure its "performance".
The particular item you're complaining about (the pipeline configuration) has been obviously "exposed" to those of us in the know through benchmarks, and completely irrelevant to the rest of the layman world( No layman at BestBuy makes a decision based on 8x1, 4x2, 2x4 because they have no earthly clue what it means or how it might affect them).
As for using that pipeline configuration as a judge of future performance irrespective of current performance, you're a better seer than I if you feel it gives you some insight on which card will perform better in the future above and beyond the tests on current measurement tools.
Joe DeFuria
Legend
RussSchultz said:
Well, I think that answers my question. In which case, there's not much sense in carrying on much further.
But I'll ask it again plainly to be sure.
Do you simply ignore on-box specs, and recommend that everyone should as well? (Honest question.) Because that's the only way I can see you having a consistent argument.
I think your answer to that is "yes, we should all simply ignore on-box specs." Correct?
Obviously. But then, you're making a further leap that "measurement tools" tell you everything you need to know. (Again, back to everyone should ignore box specs completely.)
And what if:
1) You can't find / run all the tests you feel are important / they don't exist
or
2) So what happens if can't "reconcile" the results due to inaccurate labeling. Just throw your hands in the air? Assume that future drivers can may improve performance? Is it not a "good thing to know" whether or not future drivers have some potential for significant changes in performance?
Russ:
I had about 3 different responses written, but I ended up deleting them because I think it was really just a rehash of the same arguements being made. I would like to address a couple of things though.
There are a lot of different people who buy videocards. The smallest minority are people like us who actively engage in discussions about them, and who have a pretty good handle on what's happening in the industry. Another are those who have read a review or two about videocards from places like Tom's or Anandtech, and take the time to look at benchmarks to compare cards. The largest majority by far, are those who go out to buy a new videocard, and look at the specifications listed on the back, and look at the box art and product names to make their purchasing decisions.
I believe lieing (or simply being misleading) about the specs hurts all three groups, but in different ways. Those of us who are technically inclined and of a curious nature end up wasting time figuring out why the performance of a card doesn't live up to it's theoretical speed. We also are made to feel naive and rather stupied for having believed the PR of that company in the first place once we find out the truth.
For those who are semi-informed about the card, they might see a discrepency in the benchmarks compared to what the card should do, but arn't going to automatically assume that the specs are wrong. The most likely source of blame is the drivers, especially when hardware sites are doing previews of cards that arn't yet available to the public and make that claim themselves. Most of this group will probably be smart enough to wait for performance to improve, but the opinions of the reviewers holds a lot of sway, and if the reviewers are misinformed, their readers will be too.
For the last group, it probably hurts them the most, because they buy the card based on the specs on the back of the box. This is certainly *not* a good way to buy graphics cards, but it doesn't mean that this group of people deserve to be lied to. They'll probably never know that they got cheated, because they'll just assume that any card with the same specs as theirs would perform in a similar manner.
I guess I agree with Joe, in that I don't understand why you (and other people) seem to be indifferent to it. Do you agree with us in that it causes a lot of confusion, and will end up misleading people? I guess I'm just trying to understand where you are coming from, but it's hard because I myself have been confused by the lies from some of these companies, and it makes me angry to know that I was tricked.
Nite_Hawk
I had about 3 different responses written, but I ended up deleting them because I think it was really just a rehash of the same arguements being made. I would like to address a couple of things though.
There are a lot of different people who buy videocards. The smallest minority are people like us who actively engage in discussions about them, and who have a pretty good handle on what's happening in the industry. Another are those who have read a review or two about videocards from places like Tom's or Anandtech, and take the time to look at benchmarks to compare cards. The largest majority by far, are those who go out to buy a new videocard, and look at the specifications listed on the back, and look at the box art and product names to make their purchasing decisions.
I believe lieing (or simply being misleading) about the specs hurts all three groups, but in different ways. Those of us who are technically inclined and of a curious nature end up wasting time figuring out why the performance of a card doesn't live up to it's theoretical speed. We also are made to feel naive and rather stupied for having believed the PR of that company in the first place once we find out the truth.
For those who are semi-informed about the card, they might see a discrepency in the benchmarks compared to what the card should do, but arn't going to automatically assume that the specs are wrong. The most likely source of blame is the drivers, especially when hardware sites are doing previews of cards that arn't yet available to the public and make that claim themselves. Most of this group will probably be smart enough to wait for performance to improve, but the opinions of the reviewers holds a lot of sway, and if the reviewers are misinformed, their readers will be too.
For the last group, it probably hurts them the most, because they buy the card based on the specs on the back of the box. This is certainly *not* a good way to buy graphics cards, but it doesn't mean that this group of people deserve to be lied to. They'll probably never know that they got cheated, because they'll just assume that any card with the same specs as theirs would perform in a similar manner.
I guess I agree with Joe, in that I don't understand why you (and other people) seem to be indifferent to it. Do you agree with us in that it causes a lot of confusion, and will end up misleading people? I guess I'm just trying to understand where you are coming from, but it's hard because I myself have been confused by the lies from some of these companies, and it makes me angry to know that I was tricked.
Nite_Hawk
Marketing is the art of bending the truth to suit your purposes. Facts are not created, they are:
[1] exaggerated if they are good
[2] glossed over if they are bad
Companies do this all the time, and it's a clear case of caveat emptor when it comes to interepreting such claims.
Making up untrue facts is, well, just outright lying. Does anyone here think this is a good thing? If it's not a good thing, then why even bother trying to justify it?
[1] exaggerated if they are good
[2] glossed over if they are bad
Companies do this all the time, and it's a clear case of caveat emptor when it comes to interepreting such claims.
Making up untrue facts is, well, just outright lying. Does anyone here think this is a good thing? If it's not a good thing, then why even bother trying to justify it?
Is anybody trying to justify it?
RussSchultz said:
Well, as an example, you'd expect an 8x1 to be faster than a 4x2 for compute-bound shaders (all else being equal). On the other hand, if the shader performance was bound by texture fetch, you'd expect the 4x2 to be able to keep up.
If you know what the architecture is, you have an idea of what to expect in certain synthetic benchmarks. Then you can decide how truthful a vendor is likely to be being when they talk about future performance improvements from optimized drivers.
As an artificial example, now that NVidia has become a little more truthful about the NV30 architecture, we know that it will never top about 1600 Mpixels/second, regardless of the available bandwidth. If we were still thinking in terms of 8x1, we might expect that some future driver improvement could boost the single-texture fillrate closer to the theoretical maximum for that arrangement.
no, but you are saying it doesnt matter.RussSchultz said:
And yes, as joe said, your stance on this issue indicates a passive support of untruthfullness.
Here is a simple question: Do you think its OK for a company to lie on spec sheets?
If the answer is no, then why are you willing to let them do so?
If the answer is yes, then all i can say is i think you are absolutely wrong, and a fool to boot.
I said it doesn't matter to ME, when evaluating the product.
I said it doesn't matter to me what their spec says when I'm evaluating a product, and particularly this spec, since it has zero impact on me directly. Anything I'm interested in has to do with how much it costs, how well it performs, and other directly applicable factors like driver stability, etc.
I also stated that this particular spec has near zero impact on the layman, because he hasn't a clue how it affects him anyways, but in no way did I say "hey, lying is ok".
I never said I support them lying; I said the exact opposite.
So take your moralistic accusations and ....
I said it doesn't matter to me what their spec says when I'm evaluating a product, and particularly this spec, since it has zero impact on me directly. Anything I'm interested in has to do with how much it costs, how well it performs, and other directly applicable factors like driver stability, etc.
I also stated that this particular spec has near zero impact on the layman, because he hasn't a clue how it affects him anyways, but in no way did I say "hey, lying is ok".
I never said I support them lying; I said the exact opposite.
So take your moralistic accusations and ....
Joe DeFuria said:
Perhaps they are, but I imagine only until you see something more than the specs. So I imagine that if you're the type to preorder months in advance, then that's all you have to go with, and you might suffer more if the specs are misleading. On the other hand, I don't think anyone who preorders when only specs are available makes an educated buying decision.
I agree that not having the full and correct information can impact the buying decisions of a few thousand people who actually understand and care about these issues. So I can understand when someone who is one of these people gets annoyed. However, millions of people have been tricked by NVIDIA when it named a chip GeForce4 MX, and that wasn't a spec lie at all. Yet I see it as far worse, because almost everyone I meet has no idea that the GeForce4 MX is worse than the GeForce3. That's not a lie, and it won't trick anyone who cares about the specs, but the vast majority of people don't care or understand the technical specs, and they fall for it. And I hate NVIDIA for doing it, since they keep the market back and force developers to develop for DX7 hardware for a little while longer.
Nite_Hawk said:
I think you're wrong. For the vast majority of people, specs are gibberish, and they don't look at them at all. They buy their cards based on several things: what the salespeople tell them to buy, advice from others (who often don't know much), and the numbers in the card's name. They know that if the number is higher, then the card must be better. So Radeon 9000 must be better than 8500, GeForce4 MX must be better than GeForce3, and ASUS V9180 must be better than ASUS V8460.
Those who do look at the back will see a list of very impressive theoretical throughputs along with meaningless PR speak such as "Lightspeed Memory Architecture" and very impressive "Single-pass multitexturing" and "32-bit colors, Z/stencil buffer". (This is taken from the back cover of my ASUS V8200 GeForce3 card, BTW. The scan is blurry, so I can't make out all the wonderful phrases.) I'm pretty sure that no uninformed person can make a purchasing decision based on the back cover specs, misleading or not (and of course they always are, if they contain maximum theoretical throughputs).
Althornin said:
I wrote: "I personally don't care if..." I means me, personally means especially me. There's no us in there.
Joe DeFuria
Legend
OK Russ, you haven't answered my question, which I've repeated once and asked for an answer so I can have your stance clarified. I'll assume it was an oversight. Again:
I raise it again because in your last post you are precisely raising a point about it being "personally" (as in "especially you").
I can say that I'M speaking not only for myself, and what I care about, but what I think is better for everyone else as well.
Do you think everyone else would be better off with more accurate specs, or that it doesn't matter, in the same way it doesn't matter to you?
I raise it again because in your last post you are precisely raising a point about it being "personally" (as in "especially you").
I can say that I'M speaking not only for myself, and what I care about, but what I think is better for everyone else as well.
Do you think everyone else would be better off with more accurate specs, or that it doesn't matter, in the same way it doesn't matter to you?
How many times do I have to repeat myself?
I said "I personally don't care ..."
There's no nobody else in there. Did I say "I think that everybody should not care"? No! I said "I personally don't care..."
I intentionally didn't answer your question because the answer is completly self evident with what I wrote the very first time.
I said "I personally don't care ..."
There's no nobody else in there. Did I say "I think that everybody should not care"? No! I said "I personally don't care..."
I intentionally didn't answer your question because the answer is completly self evident with what I wrote the very first time.
Joe DeFuria
Legend
Russ, I don't ask questions that I feel have already been answered clearly. I certainly don't REPEAT them. 
So great. YOU personally don't care. And you also apparently don't care about anyone else. As long as YOU are satisfied / comfortable ignoring specs (because you will get data elsewhere), everyone else should be comfortable ignoring specs too.
Personally, I think that's an arrogant position to have.
And I can only hope you don't respond with something to the effect of "NO, that's not what I mean!", because you just told me that the question was answered....
So great. YOU personally don't care. And you also apparently don't care about anyone else. As long as YOU are satisfied / comfortable ignoring specs (because you will get data elsewhere), everyone else should be comfortable ignoring specs too.
Personally, I think that's an arrogant position to have.
And I can only hope you don't respond with something to the effect of "NO, that's not what I mean!", because you just told me that the question was answered....