Of course PNG stands on it's merit. It is clear that it is far superiour to the gif format in important ways, anyone who understands the data that these files can represent knows this. With richer content and full color display hardware now common, and more PNG support than ever in applications and browsers PNG will be around for a long time to come.
Anyone can optimize if it means degrading quality, the real hypocricy is degrading quality and not doing what a benchmark is trying to mention, while boasting about superior quality rendering. It was NVIDIA saying "hey we do high quality 8X aniso" when they didn't. Any fool can get that performance benefit if they want, without some NVIDIA hack lying to them. Just go to display->properties->advanced->D3D and crank down your iniso filtering to override the application's settings, voilla, so don't pretend this was an "optimization". It was a cheat. The fundamental point here is they were telling reviewers they were doing 8X aniso when some driver engineer had hotwired it under the covers do NOT do 8X anisotropic filtering. That's fraudulent where I come from.
Bollocks, anisotropic filtering is a simple integer, even programmatically when it's not set in the driver. However you can override it in the driver, again with a simple integer that specifies state. There's no special rendering 'path' it's done in the low level texture filtering hardware. Fragment shaders request the texture fetch and may have a 'path' but the actual aniso filtering is a low level pipelined hardware operation where the application specifies the state from a fixed function set of filter parameters. What NVIDIA did here was to ignore all those numbers, and simply cheat by reducing the anisotropic filter taps, basically they said they were doing 8X aniso, but they lied and did less, or pushed out the MIP LOD to help cache behaviour, either one is a flagrant cheat. There is no special rendering path, it's one integer specified for the texture filter. The screenshot does not show anisotropic filtering in action (actually texturing at an oblique angle), if it did and the author has simply picked a better view from the demo the quality difference would be apparent.
Anisotropic filtering is a big win when viewing images under certain conditions. The conditions are not met in the screenshot, that's all. When viewed at an oblique angle anisotropic filtering becomes expensive and benefits quality most, and this is where it makes a huge difference.
Dude, the evidence is there, and it's clear, I know a fair bit about anisotropic filtering. I also know that a conditional on an application name that mysteriously accelerates only for anisotropic filtering for one application is dickering with the filter taps and texture LODs and that is supported by the visual differences when viewed with a skilled eye. As an aside there are issues with temporal aliasing quality here too that you can exploit by tapping fewer samples and pushing the spacing and/or LOD that won't show up in a still. You can see the difference in places on the screenshot, it's just not as obvious as it would be if the reviewer had the sense to take a screenshot from a more illustrative position. The evidence is there, you can choose not to see it, or somehow reject it as not enough, but a sneaky anisotropic filtering lie where 8x is claimed but 8x isn't delivered is exactly the kind of dishonest shortcut that is subtle and difficult to detect. This is not just some ambiguous issue. When the dial says 8x aniso it isn't doing 8x aniso, it's a lie and it's it's a cheat.
We have 9 smoking guns with NVIDIA prints on them. This is THE most extensive set of cheats perpetrated in the history of graphics (and to heck with the joint press releases, I know what a cheat is) I've seen Futuremark's Q&A where they clearly say these "application specific optimizations" were unacceptable, even after the joint press release.
The impact anisotropic settings are heavily dependent on the viewing angle of the texture. That particular screenshot does not appear to have particularly onerous viewing angles on any of the polygons. Given that the test setting is 8:1 anisotropic filtering you really require a view from the same demo where there's a steep slope on a few polygons to see the real visual impact. It doesn't look like any portion of that scene that's illuminated has more that 2:1 or maybe 3:1 anisotropic texture derivitive. I suspect what NVIDIA has done here is crank their aniso filtering down to maybe 4:1 in the driver, giving lower quality at steep viewing angles (blurrier textures on sloping surfaces), saving time through fewer texture filter taps and much better texture cache behaviour (due to lower resolution MIP LOD fetches).
Exactly, they changed the fragment code and broke the shader test and 'optimization'. There are shades here, if you tune a fragment program to rearrange intructions it's borderling but this apparently is what is done for real games. If you rewrite it wholesale and change what it draws then you're really cheating, ATI did the former, and it's not clear exactly how they detected the fragment code, or hos generic it was, probably not very, but it's not cased on the application.
Yes, they say that, and they're WRONG. It's the fragment *code* that's tested. They also say changing the executable name breaks the performance optimizations, WRONG again, don't believe everything you read.
No, that is NOT TRUE. The ATI optimization applies generically to any shader with similar instructions. Futuremark made a CHANGE TO THE SHADER CODE to uncover this. I do think the acceptability depends upon just how generic the optimization was, there is some wiggle room, unlike the blatant NVIDIA cheat.
If ATI had triggered on the shader name, application name or modified the results of the shader to be functionally different, I'd be right with you calling it a cheat, but they didn't.
I haven't forgotten, it should be remembered, I don't think it's worse though. I have mentioned it in other posts and The Register mentioned it in their article. Maybe you should read it before posting?
I believe they trigger their optimization on the shader CODE itself, so it is not a cheat. This is a close call because ideally you want those optimizations to be as general purpose as possible. So I'm still on the fence on this one, it's not as clear cut as the NVIDIA cheats or triggering based on the application name as ATI did with Quake->Quack.
Look there is a clear difference between what NVIDIA and ATI have done here. ATI are not cheating, they look at a sequence of instructions and reorder them if they fit a pattern, but they do exactly the same thing as before. This is central to the kinds of things optimizing compilers and CPUs do. Maybe you thing it's too narrow a path, but it's a minor infraction at best compared to the blatant cheats of NVIDIA, who not only rewrote shaders but did several otehr really heinous things, like disableing screen clear and adding their own hidden clip planes.
It's a real shame that The Register obscured the truth here with an article that attacks ATI for conservatively removing optimizations while giving the real miscreant gets a free pass. ATI should leave their optimizations in IMHO, but maybe you disagree because their mathematically equivalent optimization is not general enough, it's a close call, but they don't deserve what the distorted treatment given in The Register.
Oh I agree that earlier example was a cheat, the Quake->Quack IMHO was a cheat although there are apologists, that's what I meant when I said they've been caught cheating in the past. What's being discussed here is a separate issue w.r.t. ATI cheats in Futuremark where a ~2% difference with shaders was uncovered where they detect instructions and 'optimize'. As I said, ATI were caught cheating earlier (IMHO, I'm not interested in an argument, you're entitled to disagree). This is purely about whether they were cheating in optimizing the futuremark shaders by reordering instructions to something that was mathematically and functionally identical. Like I said it's borderline, it's a close call and I understand the position of anyone on either side of the debate. However, anyone would agree that this is not as flagrant as the NVIDIA cheats. I don't really want to take sides here, but we need to call a cheat a cheat to discourage this stuff in future, that's all, and NVIDIA cheated. ATI... well make your mind up, I don't think they really cheated this time, although they did in the Quake->Quake thing IMHO.
Well it really gets into the realms of opinion here. I'm not going to debate this, like I say it's borderline and your conclusion depends on your outlook we agree at least that any infraction was minor compared to NVIDIA and that ATI are already tainted by earlier cheats. I'm no fanboy of either, although I love their work.
ATI did not cheat (although they have cheated in the past), they reordered the instructions in a shader but it remained mathematically and functionally identical. This is what optimizing compilers do all the time. It's called optimization, not cheating, and it is legitimate, they looked at the shader INSTRUCTIONS to see if they were suitable for this optimization. It made about a 2% difference, and IMHO ATI should leave this optimization in and broaden the scope if possible.
If it was a really narrow path optimization then it's borderline, but far different from the wholesale cheating of NVIDIA who: rewrote shaders with completely different results, didn't clear the screen at certain points and added hidden clip planes to eliminate pixel fill. All very underhanded and why you would make excuses for this I don't know. Sure it's a synthetic benchmark, but if it don't matter then don't cheat.
Dunno about you, but my spam has skyrocketed recently. I mean *really* gone through the roof. I used to get a few spam, but now it's epidemic, I get about 5 or 6 spam for every legit email, it's a pain in the ass. At least I know I can blame dumb articles like this for the problem.
They've chosen 2x aa on everything which favours teh NVIDIA. All benchmarks where Radeon underperforms FX where other reviews have a mix that shows the exact opposite results. Their review is a disgrace.
The dumbasses even call Doom III a DX9 title. Frikin morons.
Do not buy a Lexmark printer. Do not buy one under any circumstances. If you have influence in any company over this kind of decision, veto Lexmark there too.
This is chicken and egg, Google is a trademark and a valuable one. Someone decided one day to call it a verb, not the other way around. It's one thing to be unable to calim proprietary rights to a verb, quite another when someone starts calling your incredibly popular and valuable trademark a verb. You can legitimately claim proprietary rights to your trademark and stop people coopting it in a way that dilutes it's value.
Microsofts XBox business relies on making money of game sales and losing money on console sales. If the XBox can run a general purpose operating system and do other useful things, it breaks Microsoft's business plan in a fundamental way if people start buying the console without buying the games.
Microsoft will not authenticate Linux for the XBox, this is one of the reasons they deliberately put the hardware protection in place, and why there is such a bounty cracking the console without 'chiping' it.
The EU citizens don't need to jump through hoops to get a Visa to come here. You fill in a form ON THE PLANE and get in. Given this lax approach it is reasonable that they run some checks and have some background info on who is arriving.
Perhaps you would rather that you have to go to the consulate weeks in advance and apply for a visa, with an interview and background checks as is common with most citizens of countries visiting other countries (not just the US).
The code for tile scrollers is NOT advanced. The effort is in the tiles and the artistic property. This whole business of saying "Please give me free tiles for *MY* tile scroller" is flawed.
The tiles ARE the tile scroller, the code that does the scrolling is trivial by todays standards. This is the underlying problem with free stuff like this, the content tends to define these projects.
What are you actually going to write? A background scroller and sprite muxer? That's old hat, and all you want is the content (the hard part) so you can go claim a project as your own. Join one of the existing efforts out there, talent is diluted enough as it is on these projects.
If you have physical access to a Unix system you can get root access using similar bootable media approaches and edit password files to your heart's desire.
If you have physical access you can defeat security.
Slashdot Mirror
Of course PNG stands on it's merit. It is clear that it is far superiour to the gif format in important ways, anyone who understands the data that these files can represent knows this.
With richer content and full color display hardware now common, and more PNG support than ever in applications and browsers PNG will be around for a long time to come.
Anyone can optimize if it means degrading quality, the real hypocricy is degrading quality and not doing what a benchmark is trying to mention, while boasting about superior quality rendering. It was NVIDIA saying "hey we do high quality 8X aniso" when they didn't. Any fool can get that performance benefit if they want, without some NVIDIA hack lying to them. Just go to display->properties->advanced->D3D and crank down your iniso filtering to override the application's settings, voilla, so don't pretend this was an "optimization". It was a cheat. The fundamental point here is they were telling reviewers they were doing 8X aniso when some driver engineer had hotwired it under the covers do NOT do 8X anisotropic filtering. That's fraudulent where I come from.
Bollocks, anisotropic filtering is a simple integer, even programmatically when it's not set in the driver. However you can override it in the driver, again with a simple integer that specifies state. There's no special rendering 'path' it's done in the low level texture filtering hardware. Fragment shaders request the texture fetch and may have a 'path' but the actual aniso filtering is a low level pipelined hardware operation where the application specifies the state from a fixed function set of filter parameters. What NVIDIA did here was to ignore all those numbers, and simply cheat by reducing the anisotropic filter taps, basically they said they were doing 8X aniso, but they lied and did less, or pushed out the MIP LOD to help cache behaviour, either one is a flagrant cheat. There is no special rendering path, it's one integer specified for the texture filter. The screenshot does not show anisotropic filtering in action (actually texturing at an oblique angle), if it did and the author has simply picked a better view from the demo the quality difference would be apparent.
Anisotropic filtering is a big win when viewing images under certain conditions. The conditions are not met in the screenshot, that's all. When viewed at an oblique angle anisotropic filtering becomes expensive and benefits quality most, and this is where it makes a huge difference.
Dude, the evidence is there, and it's clear, I know a fair bit about anisotropic filtering. I also know that a conditional on an application name that mysteriously accelerates only for anisotropic filtering for one application is dickering with the filter taps and texture LODs and that is supported by the visual differences when viewed with a skilled eye. As an aside there are issues with temporal aliasing quality here too that you can exploit by tapping fewer samples and pushing the spacing and/or LOD that won't show up in a still. You can see the difference in places on the screenshot, it's just not as obvious as it would be if the reviewer had the sense to take a screenshot from a more illustrative position. The evidence is there, you can choose not to see it, or somehow reject it as not enough, but a sneaky anisotropic filtering lie where 8x is claimed but 8x isn't delivered is exactly the kind of dishonest shortcut that is subtle and difficult to detect. This is not just some ambiguous issue. When the dial says 8x aniso it isn't doing 8x aniso, it's a lie and it's it's a cheat.
We have 9 smoking guns with NVIDIA prints on them. This is THE most extensive set of cheats perpetrated in the history of graphics (and to heck with the joint press releases, I know what a cheat is) I've seen Futuremark's Q&A where they clearly say these "application specific optimizations" were unacceptable, even after the joint press release.
The impact anisotropic settings are heavily dependent on the viewing angle of the texture. That particular screenshot does not appear to have particularly onerous viewing angles on any of the polygons. Given that the test setting is 8:1 anisotropic filtering you really require a view from the same demo where there's a steep slope on a few polygons to see the real visual impact. It doesn't look like any portion of that scene that's illuminated has more that 2:1 or maybe 3:1 anisotropic texture derivitive. I suspect what NVIDIA has done here is crank their aniso filtering down to maybe 4:1 in the driver, giving lower quality at steep viewing angles (blurrier textures on sloping surfaces), saving time through fewer texture filter taps and much better texture cache behaviour (due to lower resolution MIP LOD fetches).
Exactly, they changed the fragment code and broke the shader test and 'optimization'. There are shades here, if you tune a fragment program to rearrange intructions it's borderling but this apparently is what is done for real games. If you rewrite it wholesale and change what it draws then you're really cheating, ATI did the former, and it's not clear exactly how they detected the fragment code, or hos generic it was, probably not very, but it's not cased on the application.
Yes, they say that, and they're WRONG. It's the fragment *code* that's tested. They also say changing the executable name breaks the performance optimizations, WRONG again, don't believe everything you read.
No, that is NOT TRUE. The ATI optimization applies generically to any shader with similar instructions. Futuremark made a CHANGE TO THE SHADER CODE to uncover this. I do think the acceptability depends upon just how generic the optimization was, there is some wiggle room, unlike the blatant NVIDIA cheat.
If ATI had triggered on the shader name, application name or modified the results of the shader to be functionally different, I'd be right with you calling it a cheat, but they didn't.
I haven't forgotten, it should be remembered, I don't think it's worse though. I have mentioned it in other posts and The Register mentioned it in their article. Maybe you should read it before posting?
I believe they trigger their optimization on the shader CODE itself, so it is not a cheat. This is a close call because ideally you want those optimizations to be as general purpose as possible. So I'm still on the fence on this one, it's not as clear cut as the NVIDIA cheats or triggering based on the application name as ATI did with Quake->Quack.
Look there is a clear difference between what NVIDIA and ATI have done here. ATI are not cheating, they look at a sequence of instructions and reorder them if they fit a pattern, but they do exactly the same thing as before. This is central to the kinds of things optimizing compilers and CPUs do. Maybe you thing it's too narrow a path, but it's a minor infraction at best compared to the blatant cheats of NVIDIA, who not only rewrote shaders but did several otehr really heinous things, like disableing screen clear and adding their own hidden clip planes.
It's a real shame that The Register obscured the truth here with an article that attacks ATI for conservatively removing optimizations while giving the real miscreant gets a free pass. ATI should leave their optimizations in IMHO, but maybe you disagree because their mathematically equivalent optimization is not general enough, it's a close call, but they don't deserve what the distorted treatment given in The Register.
Oh I agree that earlier example was a cheat, the Quake->Quack IMHO was a cheat although there are apologists, that's what I meant when I said they've been caught cheating in the past. What's being discussed here is a separate issue w.r.t. ATI cheats in Futuremark where a ~2% difference with shaders was uncovered where they detect instructions and 'optimize'. As I said, ATI were caught cheating earlier (IMHO, I'm not interested in an argument, you're entitled to disagree). This is purely about whether they were cheating in optimizing the futuremark shaders by reordering instructions to something that was mathematically and functionally identical. Like I said it's borderline, it's a close call and I understand the position of anyone on either side of the debate. However, anyone would agree that this is not as flagrant as the NVIDIA cheats. I don't really want to take sides here, but we need to call a cheat a cheat to discourage this stuff in future, that's all, and NVIDIA cheated. ATI... well make your mind up, I don't think they really cheated this time, although they did in the Quake->Quake thing IMHO.
Well it really gets into the realms of opinion here. I'm not going to debate this, like I say it's borderline and your conclusion depends on your outlook we agree at least that any infraction was minor compared to NVIDIA and that ATI are already tainted by earlier cheats. I'm no fanboy of either, although I love their work.
ATI did not cheat (although they have cheated in the past), they reordered the instructions in a shader but it remained mathematically and functionally identical. This is what optimizing compilers do all the time. It's called optimization, not cheating, and it is legitimate, they looked at the shader INSTRUCTIONS to see if they were suitable for this optimization. It made about a 2% difference, and IMHO ATI should leave this optimization in and broaden the scope if possible.
If it was a really narrow path optimization then it's borderline, but far different from the wholesale cheating of NVIDIA who: rewrote shaders with completely different results, didn't clear the screen at certain points and added hidden clip planes to eliminate pixel fill. All very underhanded and why you would make excuses for this I don't know. Sure it's a synthetic benchmark, but if it don't matter then don't cheat.
Dunno about you, but my spam has skyrocketed recently. I mean *really* gone through the roof. I used to get a few spam, but now it's epidemic, I get about 5 or 6 spam for every legit email, it's a pain in the ass. At least I know I can blame dumb articles like this for the problem.
Have you read their review? It's a Joke.
They've chosen 2x aa on everything which favours teh NVIDIA. All benchmarks where Radeon underperforms FX where other reviews have a mix that shows the exact opposite results. Their review is a disgrace.
The dumbasses even call Doom III a DX9 title. Frikin morons.
This is not ballanced, the Register is absolutely unqualified to review any graphics card.
Do not buy a Lexmark printer. Do not buy one under any circumstances. If you have influence in any company over this kind of decision, veto Lexmark there too.
This is chicken and egg, Google is a trademark and a valuable one. Someone decided one day to call it a verb, not the other way around. It's one thing to be unable to calim proprietary rights to a verb, quite another when someone starts calling your incredibly popular and valuable trademark a verb. You can legitimately claim proprietary rights to your trademark and stop people coopting it in a way that dilutes it's value.
Buy Microsoft Stock!
Microsofts XBox business relies on making money of game sales and losing money on console sales. If the XBox can run a general purpose operating system and do other useful things, it breaks Microsoft's business plan in a fundamental way if people start buying the console without buying the games.
Microsoft will not authenticate Linux for the XBox, this is one of the reasons they deliberately put the hardware protection in place, and why there is such a bounty cracking the console without 'chiping' it.
The EU citizens don't need to jump through hoops to get a Visa to come here. You fill in a form ON THE PLANE and get in. Given this lax approach it is reasonable that they run some checks and have some background info on who is arriving.
Perhaps you would rather that you have to go to the consulate weeks in advance and apply for a visa, with an interview and background checks as is common with most citizens of countries visiting other countries (not just the US).
The code for tile scrollers is NOT advanced. The effort is in the tiles and the artistic property. This whole business of saying "Please give me free tiles for *MY* tile scroller" is flawed.
The tiles ARE the tile scroller, the code that does the scrolling is trivial by todays standards. This is the underlying problem with free stuff like this, the content tends to define these projects.
What are you actually going to write? A background scroller and sprite muxer? That's old hat, and all you want is the content (the hard part) so you can go claim a project as your own. Join one of the existing efforts out there, talent is diluted enough as it is on these projects.
This is no different for any decent OS.
If you have physical access to a Unix system you can get root access using similar bootable media approaches and edit password files to your heart's desire.
If you have physical access you can defeat security.