Overclock.net › Forums › Industry News › Hardware News › [Hexus] Review: Sapphire Radeon R9 Fury Nitro
New Posts  All Forums:Forum Nav:

[Hexus] Review: Sapphire Radeon R9 Fury Nitro - Page 4

post #31 of 179
Quote:
Originally Posted by Tojara View Post

They're not. Running many tests doesn't help when your testing methodology is crap.
Stop spreading FUD.

TPUs test suite and methodology for graphic cards is viewed like one of the best out there, if not the best.
It give by far more accurate results due to number of games vs reviews that test 5 games and makes up an average out of that tiny data.

There are many other aspects of TPU that give them the edge and the only reason why mtcn77 desperately critizise them is because he dont like the result.
Quote:
Originally Posted by Tojara View Post

Hardly my issue that you can't spot glaring flaws in a testing methodology, but mixing results of games running at anything between 30 and 200FPS makes as much sense as mixing 800x600 results with 4k results.
They are presenting 1080p/1440p/4K results individually.

So what do you suggest? Capping the games at 60FPS and leaving those who game with 144Hz behind?
Neither does it matter if its 200 or 30FPS because its the percentage that is used.
Edited by iLeakStuff - 1/31/16 at 5:32am
post #32 of 179
Quote:
Originally Posted by iLeakStuff View Post

So what do you suggest? Capping the games at 60FPS and leaving those who game with 144Hz behind?

Yeah, don't forget those who play at 20 FPS too.

God's sake...
post #33 of 179
Quote:
Originally Posted by mcg75 View Post

So out of curiosity, who does testing better and why is it better?

His game suite has a good mix of a lot of the latest titles with a couple older favorites thrown in.

He does not mix results. Each resolution has it's own average at the end of the test which makes some titles hitting 200 fps at lower resolutions a non issue.

And testing 1600x900 might seem silly until you look at Steam's hardware survey and see that 1600x900 still has more users than 4K and 1440p.

He's the only one that retests every card in that lineup in every test with newest drivers. Majority of others recycle results.

So don't tell us it's flawed, tell us why it's flawed.
It is an issue, since realistically you will not be running some of your games at 30FPS and some at 150FPS, you'll change the settings to get something close to your monitor's refresh rate, with a few exceptions. Well past 144FPS you're only testing how CPU bottlenecked a game is and with a 5960X that's entirely useless info for 99.5% of people.

They should separate the results for 60FPS and 144FPS since changing settings changes how cards stack up, you see manufacturers do this all the time with their own results. Increasing a tessellation heavy feature can have little performance impact on an Nvidia card while it can significantly cut the performance on an AMD card, and vice versa on shader heavy effects.

I haven't mentioned 1600x900 once or said that it's pointless, only that you shouldn't mix results since it messes up the results. Also, doing more work doesn't make his results any better than anyone else's. Working smarter is far better than working harder. If you're not testing the absolute latest games new drivers rarely have more than very marginal 2-3% improvements, and for those they have to test all the cards either way.
post #34 of 179
Quote:
Originally Posted by Tojara View Post

It is an issue, since realistically you will not be running some of your games at 30FPS and some at 150FPS, you'll change the settings to get something close to your monitor's refresh rate, with a few exceptions. Well past 144FPS you're only testing how CPU bottlenecked a game is and with a 5960X that's entirely useless info for 99.5% of people.

They should separate the results for 60FPS and 144FPS since changing settings changes how cards stack up, you see manufacturers do this all the time with their own results. Increasing a tessellation heavy feature can have little performance impact on an Nvidia card while it can significantly cut the performance on an AMD card, and vice versa on shader heavy effects.

I haven't mentioned 1600x900 once or said that it's pointless, only that you shouldn't mix results since it messes up the results. Also, doing more work doesn't make his results any better than anyone else's. Working smarter is far better than working harder. If you're not testing the absolute latest games new drivers rarely have more than very marginal 2-3% improvements, and for those they have to test all the cards either way.

Please suggest a site thats results are more accurate then?

If I'm understanding what you are saying correctly, youre suggesting they disable features that do not suit one side or the other.

I would say it's more accurate to leave all features enabled in order to give a fairer average of how the cards perform across a wide range of all games and settings. It would be unfair to compare one side with higher settings then another and present a graph of FPS as equal or closer when the image quality is not the same.
post #35 of 179
Quote:
Originally Posted by Slink3Slyde View Post

I would say it's more accurate to leave all features enabled in order to give a fairer average of how the cards perform across a wide range of all games and settings. It would be unfair to compare one side with higher settings then another and present a graph of FPS as equal or closer when the image quality is not the same.

Take for example GTA V and the GTX 770. Do you really think that 11.5 FPS at 1080p or 12 FPS at 900p is an actual portrait of its true performance at that resolution? If it represented any actual real world data you would lower whatever setting(s) that is obliterating that GTX 770 and play just fine.

Leaving the same settings going from a GTX 750 to a GTX 980 Ti is so messed up I don't even.

So yeah, TPU charts are the best around and that's quite sad.
post #36 of 179
Quote:
Originally Posted by Slink3Slyde View Post

Please suggest a site thats results are more accurate then?

If I'm understanding what you are saying correctly, youre suggesting they disable features that do not suit one side or the other.

I would say it's more accurate to leave all features enabled in order to give a fairer average of how the cards perform across a wide range of all games and settings. It would be unfair to compare one side with higher settings then another and present a graph of FPS as equal or closer when the image quality is not the same.
Anandtech, Tom's, Guru3d, Hardwarecanucks, etc. They're by no means perfect, but for the majority of time run benchmarks that make sense. Many other sites do great job from time to time.

Features like that shouldn't be disabled, as long as the game is actually playable with them. If it isn't, you should start disabling or lowering them to make the game playable, which will often change how the cards stack up, and that's my entire point.

And absolutely, apples-to-apples comparisons are still the most relevant for most people and because of that the cards tested shouldn't be massively apart in performance. Comparing a 750 ti to a, say, 390 is rarely worthwhile as you would usually turn the settings up on the latter as the framerate would typically be well over the refresh rate if the 750 ti was benchmarked at settings that were playable. The same also applies the other way around when the slower cards could be chocked to death by a whole number of things. Due to that comparing cards at entirely different performance ranges is utterly pointless if the other one isn't capable of running the benchmark properly, it is useless information.
Edited by Tojara - 1/31/16 at 7:27am
post #37 of 179
Quote:
Originally Posted by Klocek001 View Post

I always take a review like this with a shovel of salt,like every review that shows min fps very close to avg fps. in TW3 they probably just load a scene then turn the mouse around and voila!

Exactly. People reading these benchs need to comb through the settings. Basing any sound reasoning off numbers on a graph without the setup info to take into account is bad. Even then, it may not be a true reading of the performance. More tests are always welcome when i comes to a measure of perfromance.
Quote:
Originally Posted by BiG StroOnZ View Post

Is this a joke? The Fury is a terrible value, it often gets outpaced by an overclocked 980. Comparing it to 980 Ti is about the most senseless thing I've ever seen done on the internet.
Warning: Spoiler! (Click to show)


It is 26.6% slower than a non-reference 980 Ti @ 4K. Basically any 980 Ti people buy today is going to be non-reference. After you overclock the 980 Ti it is then 36.5% slower than a 980 Ti:


At a $500 price point its not competiting against the Ti. Its compettion is against the 980...and its beating it soundly. At 4k though, its comparable to the Ti and the consumer gets to keep about $120.
3770K
(14 items)
 
R7 1700
(8 items)
 
 
CPUMotherboardGraphicsRAM
RYZEN 1700 ASROCK AB350 GAMING K4 ZOTAC 1080TI REFERENCE MODEL G SKILL 8 GIG (2X) 
Hard DriveOSPowerCase
WESTERN DIGITAL 250 M.2 2280 WINDOWS 10 64 CORSAIR SF600 LIAN LI PC-O10 
  hide details  
Reply
3770K
(14 items)
 
R7 1700
(8 items)
 
 
CPUMotherboardGraphicsRAM
RYZEN 1700 ASROCK AB350 GAMING K4 ZOTAC 1080TI REFERENCE MODEL G SKILL 8 GIG (2X) 
Hard DriveOSPowerCase
WESTERN DIGITAL 250 M.2 2280 WINDOWS 10 64 CORSAIR SF600 LIAN LI PC-O10 
  hide details  
Reply
post #38 of 179
Thread Starter 
I will risk saying, this is the coolest most awesome GPU AMD and Sapphire have made so far.
This review is done with latest drivers for this card. Relative performance of TPU pretty much confirms that Fury is close to reference 980Ti.
Edited by Pro3ootector - 1/31/16 at 7:38am
post #39 of 179
Quote:
Originally Posted by Imouto View Post

Take for example GTA V and the GTX 770. Do you really think that 11.5 FPS at 1080p or 12 FPS at 900p is an actual portrait of its true performance at that resolution? If it represented any actual real world data you would lower whatever setting(s) that is obliterating that GTX 770 and play just fine.

Leaving the same settings going from a GTX 750 to a GTX 980 Ti is so messed up I don't even.

So yeah, TPU charts are the best around and that's quite sad.


Quote:
Originally Posted by Tojara View Post

Anandtech, Tom's, Guru3d, Hardwarecanucks, etc. They're by no means perfect, but for the majority of time run benchmarks that make sense. Many other sites do great job from time to time.

Features like that shouldn't be disabled, as long as the game is actually playable with them. If it isn't, you should absolutely start disabling or lowering them to make the game playable, which will often change how the cards stack up, and that's my entire point.

And absolutely, apples-to-apples comparisons are still the most relevant for most people and because of that the cards tested shouldn't be massively apart in performance. Comparing a 750 ti to a, say, 390 is rarely worthwhile as you would usually turn the settings up on the latter as the framerate would typically be well over the refresh rate if the 750 ti was benchmarked at settings that were playable. The same also applies the other way around when the slower cards could be chocked to death by a whole number of things.

I see where you are coming from guys, mid range and older cards could be tested at lowered settings, I can understand that point.

Where I was getting lost is how that's relevant to the Nitro Fury review which is AMD's second best card.

I still stand by TPU as the best site for comparing the high end.
post #40 of 179
Quote:
Originally Posted by Pro3ootector View Post

I will risk saying, this is the coolest most awesome GPU AMD and Sapphire have made so far.
This review is done with latest drivers for this card. Relative performance of TPU pretty much confirms that Fury is close to reference 980Ti.

Its definitely a top contender on amd's side for the high end. The 2nd half of 2016 is a long wait for some people for polaris. It'd be a good stopgap card in the interim.
3770K
(14 items)
 
R7 1700
(8 items)
 
 
CPUMotherboardGraphicsRAM
RYZEN 1700 ASROCK AB350 GAMING K4 ZOTAC 1080TI REFERENCE MODEL G SKILL 8 GIG (2X) 
Hard DriveOSPowerCase
WESTERN DIGITAL 250 M.2 2280 WINDOWS 10 64 CORSAIR SF600 LIAN LI PC-O10 
  hide details  
Reply
3770K
(14 items)
 
R7 1700
(8 items)
 
 
CPUMotherboardGraphicsRAM
RYZEN 1700 ASROCK AB350 GAMING K4 ZOTAC 1080TI REFERENCE MODEL G SKILL 8 GIG (2X) 
Hard DriveOSPowerCase
WESTERN DIGITAL 250 M.2 2280 WINDOWS 10 64 CORSAIR SF600 LIAN LI PC-O10 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [Hexus] Review: Sapphire Radeon R9 Fury Nitro