GTX 480 vs HD 5870!

[bogas04]

TechYudh

We already informed you about Fermi long back before its release , thanks to Tech PowerUp Forums for the pivotal information provided by them. Now that the two Green Devils , GTX 470 and 480 are out , we come up with a video benchmark , performed by a enthusiast gamers youtube channel , The Trubritar. In this Video benchmark , the gamers' malicious game Crysis , old yet powerful 3DMark Vantage and Unigine benchmark which tests DirectX11 performance , especially Tessellation which we earlier reported , is key feature of Fermi. Let's begin then!

Read More

[citan]

shame that a 480 costs twice as a 5870

[Jason S. Global Moderator]

shame that a 480 costs twice as a 5870

i know youre joking, but the 480 is $85 more than the cheapest 5870. some of those 5870s are selling for $450 (the overclocked models)

im curious to see what nvidia does w/ their drivers in the next month or so... the 200 series should be right around the corner.

good video review. it's refreshing to see videos instead of just staring at graphs

[bogas04]

i know youre joking, but the 480 is $85 more than the cheapest 5870. some of those 5870s are selling for $450 (the overclocked models)

im curious to see what nvidia does w/ their drivers in the next month or so... the 200 series should be right around the corner.

good video review. it's refreshing to see videos instead of just staring at graphs

Tell me where do you find 480 easily available? Also if you go to a store , according to Deman and Supply rule , if supply is low , demand is high and so is the price. Hence the Internet Prices are still not actual prices.

Anyways thanks , do digg / tweet or share it on facebook plz. :D

[Got3n]

Nothing I havent seen on the major review sites, tried to read the site you posted but its pretty bad, sorry to say, but the layout and the amount of content per page was just blah.

[tsupersonic]

Tell me where do you find 480 easily available? Also if you go to a store , according to Deman and Supply rule , if supply is low , demand is high and so is the price. Hence the Internet Prices are still not actual prices.

Anyways thanks , do digg / tweet or share it on facebook plz. :D

Guess you don't remember the low supply issues ATI had when they released the 5xxx series... :whistle: I'm sure the prices will settle down after a few months.

I'm sticking with my 260 for now, still handles every game I throw at it so no need to upgrade.

[bogas04]

after a few months.

right when 6xxx will be about to release

[Dmented]

Nevermind. Misread.

[Jason S. Global Moderator]

Tell me where do you find 480 easily available? Also if you go to a store , according to Deman and Supply rule , if supply is low , demand is high and so is the price. Hence the Internet Prices are still not actual prices.

Anyways thanks , do digg / tweet or share it on facebook plz. :D

heh - thanks for the lesson :rolleyes:

i was merely basing my comparison on MSRP - i cant find any 480s for sale, but once theyre available from reputable websites, they wont be inflated, but MSRP or lower.

[treemonster]

so 400 series cards are widely available. where are the 6000 series cards?

haha JK. that would be mean.

[adson]

im curious to see what nvidia does w/ their drivers in the next month or so... the 200 series should be right around the corner.

:shifty: :shifty: :shifty: :shifty: :shifty:

[Knife Party]

right now Nvidia really can't stand against ATi on any basis, not price, performance, power efficiency or even heat given by their Fermi range. And with the 6000 range around the corner, Nvidia better start practicing voodoo or something. :yes:

[treemonster]

im curious to see what nvidia does w/ their drivers in the next month or so... the 200 series should be right around the corner.

:shifty: :shifty: :shifty: :shifty: :shifty:

256 series drivers have been out for a while...

right now Nvidia really can't stand against ATi on any basis, not price, performance, power efficiency or even heat given by their Fermi range. And with the 6000 range around the corner, Nvidia better start practicing voodoo or something. :yes:

look at the 470 vs the 5870. the 470 going to to toe with the 5870 in more recent reviews and costing ~$100 less now. then there the 460, that in an sli configuration costs about the same as a 5870 and destroys it.

also i'm sure the 6000 sereis will start shipping next week and be totally magical in every way.

[Nagisan]

look at the 470 vs the 5870. the 470 going to to toe with the 5870 in more recent reviews and costing ~$100 less now. then there the 460, that in an sli configuration costs about the same as a 5870 and destroys it.

It gets close to the 5870 in terms of performance, but every benchmark I have seen shows the 5870 outperforming the 470 by a fair margin (which varies by game obviously), probably not worth the extra $100, but it is higher performance than the 470 period.

[treemonster]

It gets close to the 5870 in terms of performance, but every benchmark I have seen shows the 5870 outperforming the 470 by a fair margin (which varies by game obviously), probably not worth the extra $100, but it is higher performance than the 470 period.

http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/34207-evga-gtx-460-768mb-superclocked-single-sli-review-7.html

except for bc2, where the 470 does pretty well on a price vs performance ratio, and when aa is turned off completely in games, the 470 meets or beats the 5870, and 470 sli blows 5870 cf away.

[GreenDamGirl]

From what i have seen the 480gtx is teh top single gpu card by a pretty fair margine. Here is just one example

http://tech.icrontic.com/news/nvidias-first-official-radeon-5870-vs-gtx-480-benchmark/

Now if you throw the 5990 in there ATI reigns supreme. I also believe ATI has lost its lower cost advantage over Nvidia. Both make great products in my opinion.

I plan on replacing my current 8800gts (about 4 years old now, but still runs like a champ) soon and will be sticking with Nvidia.

[treemonster]

597- is a dual gpu card and suffers from all the issues of crossfire. it's about equal to 5870 cf so based on the review i linked you can quickly guess how it rates ona cost to performance ratio comparison with 470sli.

now the 470 doesn't do well in heat or pwoer compared to it's 5000 series counterparts, but if you're buying these cards you tend to care less about power, and if you really care about heat you are probably using watercooling anyway.

[Nagisan]

http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/34207-evga-gtx-460-768mb-superclocked-single-sli-review-7.html

except for bc2, where the 470 does pretty well on a price vs performance ratio, and when aa is turned off completely in games, the 470 meets or beats the 5870, and 470 sli blows 5870 cf away.

http://www.guru3d.com/article/geforce-gtx-460-sli-review/5

Quite different results, the 5870 does similar or better than the 470 in almost all tests, granted we are both more likely to pick out reviews that favor our opinions, I have actually heard of guru3d before, which I cannot personally say about hardwarecanucks.com.....

I will agree the 470 is a better price vs performance than the 5870, though, but I just do not see the 470 beating the 5870 in a majority of tests like you are trying to claim.

[treemonster]

you haven't heard of hwc? ok too bad i guess.

i don't like the lack openess in the review you posted on methodology of each game. what level/etc did they test. how did they get fps results? it does have a vague description of their methodology in an early page, but doesn't go in to detail. maybe you can link an article where they outline in detail their methodology for each game the test? how do they "foprce the settings" as they claim, through the driver controls panels? or through in game settings? all of this factors into a sound review.

i also loled at this piece of obnoxious FUD in the conclusion

Crossfire has gotten so much better and is still getting better and better with each quarter of the year that passes. Especially with the help of the new CrossfireX profile patches things really look much better and games get supported more quickly.

when in teh wild and else i've seen nothing but the opposite, including poorly made cf profiles that are several months later after game launches.

[Nagisan]

i also loled at this piece of obnoxious FUD in the conclusion when in teh wild and else i've seen nothing but the opposite, including poorly made cf profiles that are several months later after game launches.

I cannot speak for how good or bad CF is as I have never used it and do not plan to anytime soon. But I do know ATI implemented CF profiles which can be downloaded separately, with that, ATI can push out CF updates for specific games by releasing a new CF profile. They can do these without a new driver release which means they can push them out much more often. It does however seem that they are not updating these on a per-game basis, instead, they update a few games at a time. So while they are not updating the CF profiles the day a game comes out, they are updating them faster than waiting a few months for a driver that supports the game and they are releasing more than one at a time.

As for forcing settings, 99% of the time that has to be done through the driver or third-party software. Games generally cannot force a driver to do something, its the other way around, so I do not know why you think you need to know how they are forcing settings, its done on a driver level, not in-game.

Also, level does not matter, they are using the SAME EXACT timedemo on EACH card, which means performance between the cards is relative to that timedemo. You cannot take Guru3D reviews and compare them to HWC (as you call them) because you do not know the exact timedemo level and timeframe and what not. But Guru3D uses the same timedemo in every test, which means its the same exact level and time period in all their tests, which means their benchmarks are just as accurate as HWC would be, even if they used a different level.

In the same manner, how they measured FPS doesn't matter much either as long as they used the same method, and I am quite positive they did.

You seem to be trying to break down a review website because they don't tell you "We launched the game", "We started the timedemo we previously recorded" etc. You need to look less at specifics and more at the fact that every test they ran was ran using the SAME timedemo, measuring performance the SAME way, which means if they showed the 5870 as being faster than the 470, then using their setup, the 5870 was faster than the 470. It wasn't faster because they used a less intense level on the 5870, it wasn't faster because they measured FPS different between the two, it was faster under the same exact conditions.

Let me give you an example.

You have 2 buckets of water and you want to know which one is colder. You use a thermometer to measure each bucket, you use the same thermometer on each one, by dropping it so it settles into the bucket. YOU are trying to say "Well, that doesn't matter because you do not specify how much water is in each bucket or the material each bucket is made out of, etc etc.", sorry to say, but in this case, the amount of water and material of the bucket doesn't matter at all. You are simply trying to find out which bucket contains the coldest water. In the same sense, in the Guru3D benchmarks, you are trying to figure out which GPU is faster, the testing method is the SAME for both, the measurement method is the SAME for both, the ONLY variable is the GPU and associated drivers, in which case, the 5870 is generally the same or faster than the 470.

Again, we are both likely to find reviews to favor our opinions/thoughts, but really, why would knowing the level they used in the timedemo or the software used to measure FPS make ANY bit of difference? Unless they changed those factors for each test they are the exact same and therefore nullify each other out.

So no, none of that factors into a sound review, it is all just information you would like to know that would not change the outcome of the review even if you knew it.

Stop grasping for straws, 90% of your posts that I have seen you are attempting to prove somebody wrong and in doing so you bring up useless information and insist you are right while ignoring the other person. That is not how you prove a point, instead, try to point out facts that actually make a difference instead of information that would not change the outcome and saying lack of that information makes my facts invalid. Fact is unless you have proof that Guru3D uses different timedemos or measurement methodologies for each test, it is JUST as valid as any other review you can find. So why don't you take your own advice, quit coming up with obnoxious FUD that makes no difference in the final outcome, and back up your words with fact or at the very least reasons as to why lack of useless information would make a difference.

[bluefisch200]

597- is a dual gpu card and suffers from all the issues of crossfire. it's about equal to 5870 cf so based on the review i linked you can quickly guess how it rates ona cost to performance ratio comparison with 470sli.

now the 470 doesn't do well in heat or pwoer compared to it's 5000 series counterparts, but if you're buying these cards you tend to care less about power, and if you really care about heat you are probably using watercooling anyway.

I have not noticed micro lags since the HD4000 series anymore...

[treemonster]

I cannot speak for how good or bad CF is as I have never used it and do not plan to anytime soon. But I do know ATI implemented CF profiles which can be downloaded separately, with that, ATI can push out CF updates for specific games by releasing a new CF profile. They can do these without a new driver release which means they can push them out much more often. It does however seem that they are not updating these on a per-game basis, instead, they update a few games at a time. So while they are not updating the CF profiles the day a game comes out, they are updating them faster than waiting a few months for a driver that supports the game and they are releasing more than one at a time.

As for forcing settings, 99% of the time that has to be done through the driver or third-party software. Games generally cannot force a driver to do something, its the other way around, so I do not know why you think you need to know how they are forcing settings, its done on a driver level, not in-game.

Also, level does not matter, they are using the SAME EXACT timedemo on EACH card, which means performance between the cards is relative to that timedemo. You cannot take Guru3D reviews and compare them to HWC (as you call them) because you do not know the exact timedemo level and timeframe and what not. But Guru3D uses the same timedemo in every test, which means its the same exact level and time period in all their tests, which means their benchmarks are just as accurate as HWC would be, even if they used a different level.

In the same manner, how they measured FPS doesn't matter much either as long as they used the same method, and I am quite positive they did.

You seem to be trying to break down a review website because they don't tell you "We launched the game", "We started the timedemo we previously recorded" etc. You need to look less at specifics and more at the fact that every test they ran was ran using the SAME timedemo, measuring performance the SAME way, which means if they showed the 5870 as being faster than the 470, then using their setup, the 5870 was faster than the 470. It wasn't faster because they used a less intense level on the 5870, it wasn't faster because they measured FPS different between the two, it was faster under the same exact conditions.

Let me give you an example.

You have 2 buckets of water and you want to know which one is colder. You use a thermometer to measure each bucket, you use the same thermometer on each one, by dropping it so it settles into the bucket. YOU are trying to say "Well, that doesn't matter because you do not specify how much water is in each bucket or the material each bucket is made out of, etc etc.", sorry to say, but in this case, the amount of water and material of the bucket doesn't matter at all. You are simply trying to find out which bucket contains the coldest water. In the same sense, in the Guru3D benchmarks, you are trying to figure out which GPU is faster, the testing method is the SAME for both, the measurement method is the SAME for both, the ONLY variable is the GPU and associated drivers, in which case, the 5870 is generally the same or faster than the 470.

Again, we are both likely to find reviews to favor our opinions/thoughts, but really, why would knowing the level they used in the timedemo or the software used to measure FPS make ANY bit of difference? Unless they changed those factors for each test they are the exact same and therefore nullify each other out.

So no, none of that factors into a sound review, it is all just information you would like to know that would not change the outcome of the review even if you knew it.

Stop grasping for straws, 90% of your posts that I have seen you are attempting to prove somebody wrong and in doing so you bring up useless information and insist you are right while ignoring the other person. That is not how you prove a point, instead, try to point out facts that actually make a difference instead of information that would not change the outcome and saying lack of that information makes my facts invalid. Fact is unless you have proof that Guru3D uses different timedemos or measurement methodologies for each test, it is JUST as valid as any other review you can find. So why don't you take your own advice, quit coming up with obnoxious FUD that makes no difference in the final outcome, and back up your words with fact or at the very least reasons as to why lack of useless information would make a difference.

i was going to write something here, but i'll link a relevant article about what i was talking about with regards to time demos(not every game has one, and premade demos are not always ideal for reviews) and forcing settings at the driver level.

http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/33227-your-guide-hardware-canucks-gpu-benchmarking-process.html

http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/33190-gpu-benchmarking-methods-investigated-fact-vs-fiction.html

and a lot of things factor into the final results of reviews. that is why reviews come with such different results even if they use the same games and hardware.

I have not noticed micro lags since the HD4000 series anymore...

i've heard plent of reports of microsttueter in 5000 series cards. though in videos that claim it's obvious i can't see it.

[Nagisan]

i was going to write something here, but i'll link a relevant article about what i was talking about with regards to time demos(not every game has one, and premade demos are not always ideal for reviews) and forcing settings at the driver level.

http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/33227-your-guide-hardware-canucks-gpu-benchmarking-process.html

http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/33190-gpu-benchmarking-methods-investigated-fact-vs-fiction.html

and a lot of things factor into the final results of reviews. that is why reviews come with such different results even if they use the same games and hardware.

Funny, from reading that (the second link in particular) I gathered that actual gameplay represents the FPS you can expect, while timedemos are a fairly accurate method of measuring the difference between two GPUs, sounds oddly familiar to what I mentioned before.

We all know that actual gameplay results in values readers can expect, but gameplay is TOO variable for it to be a good benchmark when comparing GPUs, there are simply too many variables that can change the results. And, according to your second link, they actually praise timedemos because they give repeatable results, but as mentioned not every game supports them. As for the premade demos, it seems they still give fairly accurate and repeatable results which are GREAT for comparing different hardware, but a few games (out of the ones listed) give incorrect results (that is, the demo results do not agree with actual in-game performance differences).

So, thank you for the read that proves what I said to be accurate, it was an interesting read and I understand a little better at what you are trying to get at. But what you posted seems to agree with what I was saying, timedemos (which are used by Guru3D where available) give good comparisons between GPUs because of their consistency, which throws out your argument of the review being crap because they do not mention what level and everything they are doing the test in.

You seem to be missing the point, they did a test with repeatable variables, it showed the 5870 generally outpacing the 470, according to Guru3D tests, the 5870 is better than the 470.

I do not care what method they used for testing, or what level they did it on, if they did it the same on each GPU, it is a fair comparison of the two GPUs. You need to understand that how Guru3D and MOST other review sites test cards is going to be different, but as long as you compare the results within that same review, they will be comparable. You seem to think because they used a different method than HWC, that it is automatically invalid and wrong, which is just stupid and condescending.

Probably done posting in this topic, personally, but WHY can't you understand that a GPU vs a GPU with the SAME EXACT TEST is a fair comparison of their performance, regardless of whether or not that test shows realistic real-world performance you can expect?

Its like trying to compare the MPG two vehicles get, compare them both on the highway and the one that does better generally gets better MPG in the city, regardless of whether or not the highway MPG is the same as the city MPG.