[QUOTE=Blackcomb;20575721]they ran the cards at the same settings and they performed around the same for the most part except in areas of high tessellation usage in which the 480 shined, what are you going on about
[/QUOTE]
really? They tested AA and AF too? show me.
Also, one more thing I forgot to mention. Uniheaven benchmark isn't very shader intensive, its main focus is tessellation, so we'll see a mix in between with tessellation enabled games, meaning that my prediction of 20% increase in tessellation enabled games against the 5870 would probably be true, well, hopefully I'm right. Now, 20% isn't a whole lot, but it is a difference maker and you might notice it, but we have to remember that the 480 will be priced in between the 5870 and 5970(it's quite obvious actually, not a whole lot of people question that).
If this is true, ATI won't need to drop prices.
[QUOTE=ShaRose;20582742]Nice job guys, but charlie WAS right on this one.
Perhaps you should stop acting like idiots and agree that he was right.[/QUOTE]
Personally, I agree.
Charlie may be the most biased reviewer I've ever seen, but quite frankly he does seem to have a fairly great deal of insider contacts, and his information/research generally seem to hold up.
Granted he's said quite a few incorrect things in the past, and you have to take a lot of his predictions with a grain of salt, but lately he's been pretty accurate. For all his hate against Nvidia, he makes quite the effort to do his research and find his information.
I Guess Tesselation nearly triples the amount of surface triangles which makes the outermost polygons more complex. so i guess as Dx11 becomes more prevalent in games the average performance of the 4xx line will get better over time.
Until then this 2009 to Q2 2010 is definitely ATi's golden age.
[QUOTE=PunchedInFac;20585655]I Guess Tesselation nearly triples the amount of surface triangles which makes the outermost polygons more complex. so i guess as Dx11 becomes more prevalent in games the average performance of the 4xx line will get better over time.
Until then this 2009 to Q2 2010 is definitely ATi's golden age.[/QUOTE]
Ah yes but what if the 4xx series aren't worth the money? As in they give a 20% increase in JUST DX11 games, but run the same in all others as their ATI equivalents BUT are 30% more in price?
I don't see ATi's golden age stopping anytime soon
[QUOTE=thrawn2787;20598736]Ah yes but what if the 4xx series aren't worth the money? As in they give a 20% increase in JUST DX11 games, but run the same in all others as their ATI equivalents BUT are 30% more in price?
[/QUOTE]
That's not how GPUs work
[QUOTE=johanz;20598796]That's not how GPUs work[/QUOTE]
Then you have no idea what I'm saying
If a 480 is 30% more than the 5870, so $520, but gives you 20% more FPS in DX11 games (when the 480 has no AA or AF :v: ), is it worth it? Probably not
[QUOTE=thrawn2787;20599006]Then you have no idea what I'm saying
If a 480 is 30% more than the 5870, so $520, but gives you 20% more FPS in DX11 games (when the 480 has no AA or AF :v: ), is it worth it? Probably not[/QUOTE]
It doesn't give 20% more fps in DX11 games, it gives you 200% tessellation performance when there's nothing else (Like AA or AF) going on.
I have yet to be impressed by these cards, I would like to see an exact side by side comparison pitting cards of the same price.
you guys need to stop putting AF in the same performance-hog category as AA
[editline]08:38PM[/editline]
[QUOTE=ShaRose;20599966]It doesn't give 20% more fps in DX11 games, it gives you 200% tessellation performance when there's nothing else (Like AA or AF) going on.[/QUOTE]
how do you know this
[editline]08:38PM[/editline]
[QUOTE=thrawn2787;20599006]Then you have no idea what I'm saying
If a 480 is 30% more than the 5870, so $520, but gives you 20% more FPS in DX11 games (when the 480 has no AA or AF :v: ), is it worth it? Probably not[/QUOTE]
how do you know this
[QUOTE=Blackcomb;20600670]you guys need to stop putting AF in the same performance-hog category as AA
[editline]08:38PM[/editline]
how do you know this
[editline]08:38PM[/editline]
how do you know this[/QUOTE]
[img]http://i.imgur.com/9ROAO.jpg[/img]
Are you dense? That's obvious as hell, what I'm asking is how you know how the 480 is going to drop below the 5870 with anti aliasing.
In the end, if i'm looking to upgrade my motherboard, and i run 2 or 3 screens, should i go with I-finity or wait for Firmy?
Cause i'm really liking 5800's after all those reviews and comparisons, but i've been using Nvidia my whole life Dx
[QUOTE=Dark-Energy;20568563]Saw this earlier and did a lot of thinking about it.
I think that benchmark is meaningless.
The only time you see a significant performance difference is when the dragon is rendered. Now, the dragon has so many fucking polygons (i've hit f2 and actually seen it, it's fucking unbelievable), it's at the extreme level, and cause of this, it's nothing to do with the 480, it's the 5870 that struggles. Why? Because I seriously doubt we will see that many polygons being rendered in an actual real world game that uses tessellation. That benchmark is the absolute best case scenario for Nvidia. Too bad we'll never see that signifcant performance difference in a real game. Now, if you look, sometimes the cards are at par. Tessellation is still enabled, but there aren't as much polygons on the screen. So, most likely, in tessellation enabled games, we'll see something in between, like maybe 10-20 percent difference. What is worrysome for Nvidia, is that they were actually at par at a few points, because there weren't nearly as much polys being rendered. This leads me to believe that if tesselation was disabled and they tested it, or they tested a regular game, the performance would be the same.
Also, no AA or AF was used in that benchmark. Lemme guess, they probably used AA and AF and found the performance was very close, oh geez. And one more thing, that benchmark was made by Nvidia, in which they could have easily biased it. I say this because I'm suspicious of them because they've done it before, and they've also done that driver hack thing where the performance was increased but the image quality was reduced. They are creatures of habit.
Not to sound a like a fanboy in the last paragraph, but I just hate some of the shit that they try and pull off and hope that no one like me notices it.
[/QUOTE]
You're misunderstanding the graph completely.
First off, the benchmark was [b][u]not[/b][/u] made by Nvidia, nor anyone affiliated with them, so there goes about one third of your rant already. (unless you're not talking about the software and rather the graph)
Secondly, the entire point of fermi is to provide unmatched tessellation and polygon throughput, which you can clearly see happening.
Third, the only time the fermi card was at/below the 5870's framerate (even then, just by a tiny margin of 1-2 frames) was during some very brief spikes, as well as the beginning. For the other three fifths of the benchmark, the fermi card is well above the 6870's performance level. And don't exaggerate to me that "the dragon is rendered most of the time" because the dragon is only in close-up view in like 3 scenes, moderately far away in 6 more scenes, and the rest it's either incredibly far away or not drawn at all. And yes I have ran that benchmark some time ago countless times myself, so I do in fact know what I'm talking about.
Fourth, AF will do literally nothing to the framerate unless there's some sort of crippling driver bug, because even a 6600/ati equivalent can handle 16x AF while only loosing like 5 frames. AA, on the other hand, is of course going to shorten the distance between the speed of the two cards, because when you divide two values by the same divisor, the difference of those two values is also divided by that exact same divisor(eg 60 - 40 = 20, while 30 - 20 = 10). Although I will admit, ATI has been proven in the past to have more performance when you change the AA type manually in the control panel and not in-game.
Finally, the fermi card delivered an overal smoother framerate. Even if the graph is some form of average and not raw framerate data, the fermi card does in fact have a much smoother time, and if you were to divide the number of frames the two cards pushed out by the amount of time it took, you would see that the fermi card would have a quotient of approximately 50% more. The range is also a lot different, which I have noticed as a trend between the two manufacturers: AI cards usually deliver higher top framerates, but nvidia cards have a lot less valleys and hills to them.
Also, as an afterthought, here's something that begrudged me a bit.
[quote]Also, no AA or AF was used in that benchmark. Lemme guess, they probably used AA and AF and found the performance was very close, oh geez.[/quote]
Of course they are going to use the settings that show off the biggest difference between the cards, they're a fucking business. Also refer to the fourth argument of mine.
Nvidia has till the end of March for this consumer. I've bought my 3 DP monitors for Eyefinity so I am leaning towards a 5850/5870 but if they can prove their product is significantly better than ATI's in a selection of [I]real games[/I] that I care about, then my money will travel there.
I've been using my GTX 260 for awhile now and it is time for an upgrade.
The Uniengine benchmark is more a proof-of-concept for DX11 and tesselation, rather than a proper tool to test GPU's with. I personally trust the results about as much as 3DMark...that is, 'not at all'.
Fortunately, this isn't AvP; "Whoever wins, we lose". More like, "As long as they compete, we win".
[QUOTE=ChristopherB;20603551]"As long as they compete, we win, unless you're a mentally deficient fanboy".[/QUOTE]
fixed
Fuck this, I'm going to argue.
Come on Nvidia get your shit together.
[quote=heise online]3DMark Vantage (Performance-preset)
ATI Radeon HD 5870: P17303
GeForce GTX 470: P17156
ATI Radeon HD 5850: P14300
3DMark Vantage (Extreme-preset)
ATI Radeon HD 5870: X8730
GeForce GTX 470: X7511
ATI Radeon HD 5850: X6430
GeForce GTX 285: X6002
Unigine Heaven (4xAA)
GeForce GTX 470: 29 FPS
ATI Radeon HD 5870: 27 FPS
ATI Radeon HD 5850: 22 FPS
Unigine Heaven (8xAA)
ATI Radeon HD 5870: 23 FPS
GeForce GTX 470: 20 FPS
ATI Radeon HD 5850: 19 FPS[/quote]
.
If thats true and the 470 can at least come within 3 FPS of a 5870 we wouldnt be blown away by the performance of the 480. So i guess we have to wait for both companies to pull out their aces (the 495 and the 5990) so that we can know who is better.
And many companies like giving their game "the way its meant to be played" optimizations, i truly havent seen many that choose ATi ove Nvida
I understand why nvidia uses these abstract bus types like 384-bit. The 5870 uses 256 fine. can someone explain this to me?
[QUOTE=BaconMan_lol;20609181]I understand why nvidia uses these abstract bus types like 384-bit. The 5870 uses 256 fine. can someone explain this to me?[/QUOTE]
48x8 vs 32x8
[QUOTE=Dark-Energy;20604359]Edit: Removed. It's pointless to throw pebbles against a piece of steel armor. No matter how you throw em, they'll just keep bouncing off. Obviously a lot of people just aren't as smart as me and don't put things together (like there's a fucking Nvidia logo on the graph, obviously it's made by Nvidia or a 3rd part assosiated with them) God. Oh well. I'll just wait till the card comes out, then my points will be properly proven. Fermi, apparently, is now a huge hype up cause of this benchmark. Funny how it manipulates so many people. Until it comes out I'll just stop posting in this thread, don't wanna start any flame wars. Get the last word, I don't care.[/QUOTE]
"I'm not going to try and provide a counter-argument to defend my stupidity because it's too difficult. Go ahead and try and continue but I'm calling you out on it so it'll only make you look like a giant prick."
Also, there absolutely no final word on pricing. though i do doubt it to be cheaper than the 5870 (the 480, that is) But a man can always hope and dream
[QUOTE=Robbazking;20609152]Ah fuck if 470 really is that fucking shitty i see no reason in buying 2 of them since i could had picked up 2 5870 at release like fucking 4+ months ago?[/QUOTE]
did you not see the benchmarks or are you trolling
[QUOTE=FunnyGamer;20609947]"I'm not going to try and provide a counter-argument to defend my stupidity because it's too difficult. Go ahead and try and continue but I'm calling you out on it so it'll only make you look like a giant prick."[/QUOTE]
LOL. My stupidity? You're the one who doesn'tunderstand the concept and logic I'm trying to make. Lol...how fucking dare you say that....trying to act all smug in shit huh...just like the rest of them....a confidence booster I guess...makes them feel better about themselves...I can't believe how stuck up some people are...fucking terrible. The world's an awful place to live in....and it's difficult to counter what you say, cause you will never understand cause you're not smart enough...like I said...it's like throwing pebbles at a piece of steel armor......but let me try anyways. I could easily call you stupid as well. Like how you say it wasn't made by Nvidia. You see the fucking logo right on it, and the case even has Nvidia on it. Either it's them or a 3rd part associated with them..pfff..and what I meant by the dragon part was that there's quadruple the amount of polygons being rendered, so if the graph is similiar in other parts, the poly count is very similiar. What I was trying to get to was the amount of polys. God..how much do I have to explain myself?
AA and AF does matter. They have no reason not to use it unless trying to hide something. Once again..like a business thing like you said....cause if the 5870 can handle it better...that closes the gap...yes I know about the 60 - 40 = 20, while 30 - 20 = 10 stuff it's all to do with the percentage difference...so once again if it can handle AA and AF better...the percentage increase on the 480 would be less. So hard to understand, isn't it?
I don't care if Fermi produced a better framerate. The benchmark is meaningless. You know why? Let me explain some things. First of all Fermi was made to use tessellation better than it's rival. The way they did this is that it's software/hardware based (I think). But anyways, it's able to dynamically allocate resources away from the shaders and other parts of the GPU and concentrate it all on the tessellators. Now, the 5870 uses a static, fixed tessellator. So it's basically at a fixed performance.
So, take uniheaven benchmark, which is 90% tessellation, hardly any shaders/anything else. No wonder Fermi performs so well, there's nothing else to process other than tessellation. While this is good, it's not as effective in real games. First of all, I doubt we will see that many polygons being rendered at one time in a tessellation enabled game. Secondly, there is much more to games than just tessellation. This is just a simple benchmark. Games are much more complex graphically (not performance wise) due to much more shaders and variance of it, and variance of textures, post processing, the list goes on. My predication is that the 5870 will be respectively behind the 480 by 20% in tessellation enabled games, and being priced 20% less or more.
So yeah, basically, Uniheaven is a big show off for Fermi, to show it's powerful ability to use tessellation. However, it's not as effective in real life gaming.
Also,
inb4 fanboy
inb4 overly sarcastic joke
inb4 "quote" rape
inb4 every single one of my arguments gets countered by a misunderstood statement, or simply because a user simply just cannot comprehend and understand logic and science.
[QUOTE=Dark-Energy;20604359]Edit: Removed. It's pointless to throw pebbles against a piece of steel armor. No matter how you throw em, they'll just keep bouncing off. [b]Obviously a lot of people just aren't as smart as me and don't put things together (like there's a fucking Nvidia logo on the graph, obviously it's made by Nvidia or a 3rd part assosiated with them) God. Oh well. I'll just wait till the card comes out, then my points will be properly proven.[/b] [b]Fermi, apparently, is now a huge hype up cause of this benchmark. Funny how it manipulates so many people. Until it comes out I'll just stop posting in this thread, don't wanna start any flame wars. Get the last word, I don't care.[/b][/QUOTE]
[QUOTE=Dark-Energy;20613372]LOL. My stupidity? [b]You're the one who doesn'tunderstand the concept and logic I'm trying to make.[/b] Lol...[b]how fucking dare you say that[/b]....[b]trying to act all smug in shit huh...just like the rest of them....a confidence booster I guess...makes them feel better about themselves...I can't believe how stuck up some people are...fucking terrible.[/b] [b][i]The world's an awful place to live in....and it's difficult to counter what you say, cause you will never understand cause you're not smart enough[/i][/b]...like I said...it's like throwing pebbles at a piece of steel armor......but let me try anyways. I could easily call you stupid as well. Like how you say it wasn't made by Nvidia. You see the fucking logo right on it, and the case even has Nvidia on it. Either it's them or a 3rd part associated with them..pfff..and what I meant by the dragon part was that there's quadruple the amount of polygons being rendered, so if the graph is similiar in other parts, the poly count is very similiar. What I was trying to get to was the amount of polys. [b]God..how much do I have to explain myself?[/b]
AA and AF does matter. They have no reason not to use it unless trying to hide something. Once again..like a business thing like you said....cause if the 5870 can handle it better...that closes the gap...yes I know about the 60 - 40 = 20, while 30 - 20 = 10 stuff it's all to do with the percentage difference...so once again if it can handle AA and AF better...the percentage increase on the 480 would be less. So hard to understand, isn't it?
I don't care if Fermi produced a better framerate. The benchmark is meaningless. You know why? Let me explain some things. First of all Fermi was made to use tessellation better than it's rival. The way they did this is that it's software/hardware based (I think). But anyways, it's able to dynamically allocate resources away from the shaders and other parts of the GPU and concentrate it all on the tessellators. Now, the 5870 uses a static, fixed tessellator. So it's basically at a fixed performance.
So, take uniheaven benchmark, which is 90% tessellation, hardly any shaders/anything else. No wonder Fermi performs so well, there's nothing else to process other than tessellation. While this is good, it's not as effective in real games. First of all, I doubt we will see that many polygons being rendered at one time in a tessellation enabled game. Secondly, there is much more to games than just tessellation. This is just a simple benchmark. Games are much more complex graphically (not performance wise) due to much more shaders and variance of it, and variance of textures, post processing, the list goes on. My predication is that the 5870 will be respectively behind the 480 by 20% in tessellation enabled games, and being priced 20% less or more.
So yeah, basically, Uniheaven is a big show off for Fermi, to show it's powerful ability to use tessellation. However, it's not as effective in real life gaming.[/QUOTE]
hahahahahahhahahahahahahahahahahahahahahhahahahahahahhahahahahahahahahahhaahahahahahhahahahahahah
Hell I even partially agree with your statement/point, but the way you argue this; oh god this is hilarious.
[QUOTE=Shogoll;20613556]hahahahahahhahahahahahahahahahahahahahahhahahahahahahhahahahahahahahahahhaahahahahahhahahahahahah
Hell I even partially agree with your statement/point, but the way you argue this; oh god this is hilarious.[/QUOTE]
Yeah sorry, I kinda forgot to fix it before I posted that...ahh I knew it was floating around the back of mind and someone would call me out on it...my prophecy was correct.
[QUOTE=Dark-Energy;20613372]inb4 every single one of my arguments gets countered by a misunderstood statement, or [b][i]simply because a user simply just cannot comprehend and understand logic and science.[/i][/b][/QUOTE]
Ahahahahahahhahaahahahahahhahahahahhahahahahahahahahahahahahahahahahaha
[QUOTE=Dark-Energy;20613372]LOL. My stupidity? You're the one who doesn'tunderstand the concept and logic I'm trying to make. Lol...how fucking dare you say that....trying to act all smug in shit huh...just like the rest of them....a confidence booster I guess...makes them feel better about themselves...I can't believe how stuck up some people are...fucking terrible. The world's an awful place to live in....and it's difficult to counter what you say, cause you will never understand cause you're not smart enough...like I said...it's like throwing pebbles at a piece of steel armor......but let me try anyways. I could easily call you stupid as well. Like how you say it wasn't made by Nvidia. You see the fucking logo right on it, and the case even has Nvidia on it. Either it's them or a 3rd part associated with them..pfff..and what I meant by the dragon part was that there's quadruple the amount of polygons being rendered, so if the graph is similiar in other parts, the poly count is very similiar. What I was trying to get to was the amount of polys. God..how much do I have to explain myself?
AA and AF does matter. They have no reason not to use it unless trying to hide something. Once again..like a business thing like you said....cause if the 5870 can handle it better...that closes the gap...yes I know about the 60 - 40 = 20, while 30 - 20 = 10 stuff it's all to do with the percentage difference...so once again if it can handle AA and AF better...the percentage increase on the 480 would be less. So hard to understand, isn't it?
I don't care if Fermi produced a better framerate. The benchmark is meaningless. You know why? Let me explain some things. First of all Fermi was made to use tessellation better than it's rival. The way they did this is that it's software/hardware based (I think). But anyways, it's able to dynamically allocate resources away from the shaders and other parts of the GPU and concentrate it all on the tessellators. Now, the 5870 uses a static, fixed tessellator. So it's basically at a fixed performance.
So, take uniheaven benchmark, which is 90% tessellation, hardly any shaders/anything else. No wonder Fermi performs so well, there's nothing else to process other than tessellation. While this is good, it's not as effective in real games. First of all, I doubt we will see that many polygons being rendered at one time in a tessellation enabled game. Secondly, there is much more to games than just tessellation. This is just a simple benchmark. Games are much more complex graphically (not performance wise) due to much more shaders and variance of it, and variance of textures, post processing, the list goes on. My predication is that the 5870 will be respectively behind the 480 by 20% in tessellation enabled games, and being priced 20% less or more.
So yeah, basically, Uniheaven is a big show off for Fermi, to show it's powerful ability to use tessellation. However, it's not as effective in real life gaming.
Also,
inb4 fanboy
inb4 overly sarcastic joke
inb4 "quote" rape
inb4 every single one of my arguments gets countered by a misunderstood statement, or simply because a user simply just cannot comprehend and understand logic and science.[/QUOTE]
The best part of this thread is you thinking that other people actually care about your opinion and outlook.
[editline]01:44PM[/editline]
... ... ... ... ... ... ... .
It's the internet.
You're supposed to be an asshole.
Sorry, you need to Log In to post a reply to this thread.
