SIS Xabre 400, is this the right way to optimise?

[legion88]

Looks to be a real winner.

Afterall, SiS is paying attention to consumers by "using a few optimisations in there to assist performance" under OpenGL.

One of these optimizations include dropping down to bilinear at resolutions of 800x600 and greater, no matter what graphical settings you use.

It isn't a bug nor a cheat. It is an optimisation.

More hardware companies should use optimizations like this. Benchmarks are never misleading with these optimizations. They are always measurable (obviously) and reproducible. This is good stuff.

Now if the evil Satan Microsoft would to leave SiS alone, these same optimisations would be available under D3D.

Should more companies optimise the hardware in a similar fashion (not necessarily use the same trilinear optimisation)?

[The Baron]

Uh, if nVidia or ATI did something similar, there would be mass killings, riots, death in the streets, etc. al. Then you'd have the Congressional hearings... :p

Really, this crap about trilinear and no 32-bit textures in OGL... it's not optimisations. It IS a cheat. As soon as it affects image quality.... it's a cheat.

[legion88]

Originally posted by =SSC=The Baron
Uh, if nVidia or ATI did something similar, there would be mass killings, riots, death in the streets, etc. al. Then you'd have the Congressional hearings... :p

Really, this crap about trilinear and no 32-bit textures in OGL... it's not optimisations. It IS a cheat. As soon as it affects image quality.... it's a cheat.

How can it be a cheat? ;)

After all, it doesn't have code in the drivers that goes something like this: "if game = 'quake' then run trilinear optimisation and use 16-bit textures". Their optimisation affects all OpenGL games and it is not selective to a widely-used "benchmark software".

SiS decided to optimise the OpenGL performance that differs from some people's expectations. That's all.

edit: It wasn't like SiS was trying to hide anything. They simply didn't bother to make it widely known that they had these optimisations in place.

[The Baron]

edit: It wasn't like SiS was trying to hide anything. They simply didn't bother to make it widely known that they had these optimisations in place.

ATI didn't hide the whole Quake/Quack thihng, they just didn't make it widely known that they had these optimisations in place.

[legion88]

Originally posted by =SSC=The Baron
ATI didn't hide the whole Quake/Quack thihng, they just didn't make it widely known that they had these optimisations in place.

Exactly!!!

So if the Quack benchmark-specific optimisation isn't a cheat, then neither is SiS' optimisation.

For consistency, a certain number of people on this board should vote "yes" in the poll, unless of course, they have since changed their stance regarding the Quack optimisation.

[The Baron]

Apparently you didn't note the sarcasm of my post. :rolleyes:

[StealthHawk]

well, from the B3D review, it seems like SiS has at least stopped that TexTurbo crap as talked about in this thread: http://www.nvnews.net/vbulletin/showthread.php?s=&threadid=1213

they have now given users a choice to go to standard IQ by setting the driver to "Quality." even though it still seems they are playing with some low ball moves in that Trilinear apparently will be forced off in resolutions above 640X480 even if it is selected in-game if the driver is not set to "Quality."

and it seems like there are some problems exposed fully by B3D in that "4x FSAA" is only comparable to 2x from other vendors. and 2x doesn't do squat for jaggies but instead blurs the whole screen.

why can't SiS just stop this IQ degradation stuff and conform to standards. it's just ridiculous.

[ASCI Blue]

nVidia did the same thing to boost 3dmark scores after the Radeon 9700 was released. But since that's nVidia it's not a cheat right? :rolleyes:


why can't SiS just stop this IQ degradation stuff and conform to standards. it's just ridiculous.


A friends prof made a comment about this and I'll post it for you StealthHawk. ""That's the great thing about standards, so many to choose from." And that is a very true statement.

[legion88]

Originally posted by StealthHawk
well, from the B3D review, it seems like SiS has at least stopped that TexTurbo crap as talked about in this thread: http://www.nvnews.net/vbulletin/showthread.php?s=&threadid=1213


why can't SiS just stop this IQ degradation stuff and conform to standards. it's just ridiculous.

Isn't standards part of the problem, as in, there isn't any standards?

What is the standard mipmap level? What is the standard "performance" aniso? What is the standard "quality" aniso? What is the standard "2X FSAA"?

When 3dfx still ran the show, their OpenGL quake benchmarks were based on the use of their famous quake-optimised miniGLs while all other vendors were using the slower OpenGL ICDs.

In the past, it is clear that reviewers participated enthusiastically in treating one vendor differently from all others, resulting in favorable performance results in benchmarks. Reviewers, for instance, did not request from ATi a quake-enhanced miniGL to even the playing field.

When ATi obviously had code in their drivers for the Radeon 8500 to recognize Quake III, fans jumped on the "it's a bug" bandwagon about a week before ATi made the announcement (in an interview) that there's this alleged "bug".

Why not give SiS the same treatment?

SiS may not be a big name player (were they ever?) but they should at least get the same treatment as all other graphics card manufacturers of the past.

[lagadu]

the thing would be that Sis doens't have enough market to pull this off in a good way for them, let's façe it, if you go around asking people at random if they know Sis, they won't, so they won't sell cards based on the "name". so most people left who'd be willing to buy one are more into the "know-how" so when they go and read things about it and talk to those who can claim understanding about vid cards, the xabre just gets slaughtered because of these "optimizations" known by those who offer advice of those who investigate and they end up buying something from nvidia or ati.

this is why i think we won't see much from sis after all this xabre thing...
just my €0.02

[jbirney]

When ATi obviously had code in their drivers for the Radeon 8500 to recognize Quake III, fans jumped on the "it's a bug" bandwagon about a week before ATi made the announcement (in an interview) that there's this alleged "bug".


Why do you keep bringing this up over and over? The Q3 references were in the drivers 8 months before the 8500 was even shipped. They were created to "optimize" the oringal Radeon cards.

[-=DVS=-]

SIS is crap lets forget it an move on :rolleyes:

[poursoul]

Originally posted by legion88
Looks to be a real winner.

Afterall, SiS is paying attention to consumers by "using a few optimisations in there to assist performance" under OpenGL.

One of these optimizations include dropping down to bilinear at resolutions of 800x600 and greater, no matter what graphical settings you use.

It isn't a bug nor a cheat. It is an optimisation.

More hardware companies should use optimizations like this. Benchmarks are never misleading with these optimizations. They are always measurable (obviously) and reproducible. This is good stuff.

Now if the evil Satan Microsoft would to leave SiS alone, these same optimisations would be available under D3D.

Should more companies optimise the hardware in a similar fashion (not necessarily use the same trilinear optimisation)?

god, doesn't anybody else notice that this sounds like a major plug?

C'mon, "the evil satan microsoft" "Looks to be a real winner" and of course "It isn't a bug nor a cheat. It is an optimisation."

Hey, DVS, if you were to post about the same thing (in a favorable manner) wouldn't you just put:

hey, people look at this. Your thoughts.

or something similar?

This guy works for SiS or a is a SiStroll *spits bad taste out of mouth*

[-=DVS=-]

Originally posted by poursoul
god, doesn't anybody else notice that this sounds like a major plug?

C'mon, "the evil satan microsoft" "Looks to be a real winner" and of course "It isn't a bug nor a cheat. It is an optimisation."

Hey, DVS, if you were to post about the same thing (in a favorable manner) wouldn't you just put:

hey, people look at this. Your thoughts.

or something similar?

This guy works for SiS or a is a SiStroll *spits bad taste out of mouth*

Heh ok My Thoughts

Sis Hardware is bad its a fact
Sis Drivers bad also its a fact
Or any other company for a fact there is no other companys who make good gameing cards , Matrox makes good 2d whatever
:rolleyes: Triple head is usefull

ATI & Nvidia make BEST High Ends and Low end parts

now why anyone should by inferior product when Nvidia and Ati have alot of Low ends with superb features/performance

SIS have zero new features to offer , they don't have speed to offer only scam :o its not even full DX8 :rolleyes:

There was also some company who had p10 card with flexible shaders and full DX 8 :rolleyes: we now have DX9 and still they didn't release the card :rolleyes:


EDIT: lowering image quality to get extra points they call it optimization isn't this pathetic :rolleyes: and by defoult with no chance to run a game as you like it :o SiS does Su*k

[The Baron]

Isn't standards part of the problem, as in, there isn't any standards?

The STANDARD is that when you say that you are using trilinear instead of bilinear, you actually USE trilinear. The STANDARD is when you are supposed to use 32-bit textures, you don't automatically reduce them to 16-bit textures.

In the past, it is clear that reviewers participated enthusiastically in treating one vendor differently from all others, resulting in favorable performance results in benchmarks. Reviewers, for instance, did not request from ATi a quake-enhanced miniGL to even the playing field.

Were they supposed to request something special that the general public did not have access to in order to get "more accurate" benchmarks? You get benchmarks with what's publicly available. You don't get benchmarks using Top-Secret Stuff That Only The Reviewer Can Use. If ATI wanted to provide a Quake-specific GL library in order to improve its performance in Quake, then fine, they should release it.

Yeah, this is a real Sis fanboy thread.... blah.

[Joe DeFuria]

Lol...

Roscoe to this day accuses me of being some throw-back 3dfx fan, and he STILL starts new threads about the Quake/Quack thing?

:rolleyes:

For the record....and in hopes this will be the last time:

ATI had a driver BUG. YES, IT IS BAD for a driver to 'ignore' the settings of a game or the user. ATI's BUG caused certain textures to not use the correct LOD in Quake. That is a legitimate problem and gripe. That BUG was fixed within two weeks. And performance did NOT SUFFER when it was fixed. One would think it would have been forgotten at that point...

In short: ATI'S BUG WAS BAD. This does not mean it is a 'cheat', as that is an intentionally doing something bad to increase performance.

What SIS is doing is:
1) Definitely bad. As I said...any time a driver does something not requested by the game or user, it's bad.

2) It's probably cheating. Can't say for sure at this time because we don't know if the drop from trilinear to bilinear is intentional or not. If it doesn't get "fixed" in the next driver release or two, I'd say it's definitely intentional, and along the lines of cheating.

No inconsistencies between this situation and Quack. Only with Roscoe...

[StealthHawk]

Originally posted by ASCI Blue
nVidia did the same thing to boost 3dmark scores after the Radeon 9700 was released. But since that's nVidia it's not a cheat right? :rolleyes:

what are you talking about? when did nvidia lower image quality to get higher 3dmark scores?

i recall hearing nothing about this. unless you're talking about the splash screen buffer flushing(which didn't lower IQ), and gave minimal performance gains.

[AnteP]

Quack was a freakking bug, the next driver after the quack driver came out with
BETTER GRACPHICS and BETTER PERFORMANCE
now that's the type of optimization I'd like

not having bilinear and 16 bit textures shoved down my throat...

[legion88]

Originally posted by =SSC=The Baron
The STANDARD is that when you say that you are using trilinear instead of bilinear, you actually USE trilinear. The STANDARD is when you are supposed to use 32-bit textures, you don't automatically reduce them to 16-bit textures.



If there was such a "standard" as you speak, then why was there not an uproar when the Voodoo5's "high quality" quake3 scores were compared to other card's "high quality" quake3 scores? You know, the V5 only uses bilinear in Quake 3 (and any other multi-textured game).



Were they supposed to request something special that the general public did not have access to in order to get "more accurate" benchmarks? You get benchmarks with what's publicly available. You don't get benchmarks using Top-Secret Stuff That Only The Reviewer Can Use. If ATI wanted to provide a Quake-specific GL library in order to improve its performance in Quake, then fine, they should release it.

Yeah, this is a real Sis fanboy thread.... blah.

Where's the logic that "request" means "Top Secret Stuff". Watching too many James Bond movies, have we?

Your argument on "standard" is not reality at all and has never been. It's the ideal, not the reality.

When one video card uses an OpenGL ICD while the other uses a quake-enhanced miniGL for quake benchmarks, how is that a "standard"? It is not a standard. It is favoritism. Favoritism is the reality.

This is not a SIS fanboy thread. I have never bought a SIS graphics card (unless, of course, it was "attached" to a standard computer package such as Dell of years ago but that's different).

This thread is to show the hypocrisy shown by fanboys. It proves that smaller companies like SIS (smaller in terms of making graphics cards) can not get away with the same BS that larger companies like ATI can.

There's not a single supporter of SIS here. SIS is being shut-out in terms of support. ATI, on the other hand, had plenty of fanboys. Remember, this is an NVIDIA fansite. SIS is being pounded while ATI has supporters.

Let's take the quack.

Fanboys think that the "quake 3" code in the drivers showing up 8 months before the Radeon 8500 hit the stores somehow supports their little religious cause. And that the "quake 3" code was used to optimise the original Radeon 256.

Amusing isn't it? The "quake 3" code as pointed out by fanboys of that time in defense of ATI claimed that it had no affect on the original Radeons, only the Radeon 8500. Thus, they reasoned, it was a bug.

Now it's been turned around 180 degrees. Today, it was claimed that the quack did had an effect since it was used to "optimise" the original Radeon. Because of that, it ended up having a negative impact on the 8500--a "bug".

Further, 8 months means that they had 8 months to "work" the code and then fix it but they didn't. That's months of incompetence that the fanboys are unknowingly admitting. Might not have been the full 8 months of incompetence but it is still months, especially considering the obvious fact that they were able to fix the "bug" so quickly after being caught (i.e. remove the IQ problems while keeping nearly the same performance).

How hard is it to simply wait to release a "fix" after being caught to make it appear that a "fix" was being worked on? Not hard at all.

And considering its impact on image quality, we all know that one can not use the idiotic excuse that the ATI driver team simply didn't see the image quality problems--they did. One has to be incredibly stupid to really believe software engineers being paid 40,000 dollars (US) per year or more didn't see the IQ problems of the world's most widely used benchmark game for 8 months. That's 8 months, not 8 minutes, not 8 days but 8 months.

(Of course, we all know that the reviewers saw the same image quality problem of the 8500 as well when it came out. The majority did not say anything and many avoided using quake 3 pics. They, instead, used pics from Serious Sam and others to show off the image quality of the 8500. Why was that? Was there a "suggestion", not knowing that they were being manipulated? Who knows. Perhaps they were simply giving a new product the benefit of the doubt, which would be dumb.)

Other than hardware sites (see comment below), it would appear that no user during that 8 months even knew about that "quake 3" code. Thus, it proves that ATI can hide something in "plain view" for months without the user even knowing about it. Of course, it might not look like "hiding" to a few coders but most users do not go around searching for strings in drivers.

Naturally, the fanboys conveniently forget the fact that hardware sites actually already knew about this so-called "bug" around mid-summer 2001, months before the public knew about it in October (or November) of 2001. NVIDIA was spreading the "good news" about what was discovered in ATI's drivers--at least twice, once in the summer and again in October.

And we all have to like the nonlogic of fanboys. Fixing the problem after being caught using it somehow proves that it was an unintentional bug.

[The Baron]

Um, yeah, apparently you haven't noticed, but the days of Quake and OGL ICDs are over. DirectX and actual OpenGL support from every major manufacturer kinda fixed all that.

So, the moral is, yes, there are standards. No, there are not standards to actual image output due to differences in hardware, but yes, there are standards in what the card performs to what it is fed with. If you're SUPPOSED to use trilinear, you use trilinear. If you don't, you should be yelled at. Voodoo5 example is too long ago for me to remember, but oh well.

If you're a reviewer, yes, request DOES mean stuff that is not available to the general public. So no, you should not request extra drivers or what-have-you if they are not ready for primetime or they are not going to be released to the general public anytime soon for whatever reason.

And don't call me a fanboy. I've used Matrox cards, I've used S3 cards, I've used nVidia cards, the only card I ever put in myself were 3dfx and ATI cards. I'm NOT a fanboy by any stretch of the imagination. I just believe that if you say you're doing something on the chip, you should actually be doing it. If SiS wanted my support, they should make a chipset that competes on the same level as a Radeon 9500 with similar IQ but with cards that cost $50 less. THEN they'd have my support.

But this whole Sabre thing... it's a non-entity as far as I'm concerned. It's the wrong way to optimize. Image quality is image quality. It's like if you bought CDs with MP3s instead of Redbook audio on it. Sure, they could use CDs with a smaller diameter and potentially make it cheaper, but you sacrifice a great deal of quality. So, if you appreciate quality, you will regard those CDs as a non-entity.

Blah.

[lagadu]

what part of the quack issue changes the fact that sis has a crap card with crap drivers? :confused: :confused:
[edit]oh, and ati made it in one set of drivers, they were caught and never done it again and in the following driver set they not only corrected that but they had increased performance, but sis is doing this for several driver revisions on purpose, does this mean they admit that they have CRAP hardware and they only way they can keep competitive is by doing this?
ati didn't get away with the quack issue, even today (read this thread for example) they get criticized for doing it, why should sis be able to do this and get away with it???

[Bigus Dickus]

Jesus Christ, I swear I'll be 90 years old one day and Roscoe will still be spouting the same tired rhetorical bull**** about Quack and XOR image quality.

Some idiots never get any wiser I suppose. :rolleyes: