AMD vs Nvidia; GTX 1060 vs. RX 480 - An Updated Review.

Never seen [H]ardocp use the same settings when doing reviews unless it is the Apples to Apples section. If some patch or driver made the game run better than before, then they use the better settings, and highlight that video card A could run this and video card B could only operate at these settings. This occurred in the latest Watch Dogs review on the front page if I remember right. All of the people complaining here should go to that comment section and tell Brent why he's wrong too.

So what's the difference in Hardware Canucks using the most up to date settings for the games? If both video cards are running them Apples to Apples then what is the big deal? Hardware gets refined, drivers are streamlined for better efficiency, and settings are adjusted. At least that's how I've been treating my PC hardware for the past 35 years.
 
cageymaru said:
Never seen [H]ardocp use the same settings when doing reviews unless it is the Apples to Apples section. If some patch or driver made the game run better than before, then they use the better settings, and highlight that video card A could run this and video card B could only operate at these settings. This occurred in the latest Watch Dogs review on the front page if I remember right. All of the people complaining here should go to that comment section and tell Brent why he's wrong too.

So what's the difference in Hardware Canucks using the most up to date settings for the games? If both video cards are running them Apples to Apples then what is the big deal? Hardware gets refined, drivers are streamlined for better efficiency, and settings are adjusted. At least that's how I've been treating my PC hardware for the past 35 years.
Click to expand...


Because without that baseline, to reach such conclusions, like they have, even their % difference tables, just don't work. You no scientific means of measuring

  • Independent Variable: The variable you will test (vary) to answer your hypothesis. In the example above, the independent variable would be the different colors of the light bulbs.
  • Dependent Variable: This is what occurs in response to the changing independent variable.
  • Control: The control should be the part of the experiment where you do not include the Independent Variable.
The control is what is missing (which I call baseline), without that, you cannot draw conclusions based on the other two variables, cause both independent and dependent variables have correlations to the Control but might not have correlations between each other.
 
razor1 said:
The control is what is missing (which I call baseline),
Click to expand...

So since it is missing your baseline you cant believe the benchmarks?....The benchmarks look legit no matter how people try to make it fit their agenda.

99% of websites never go back to see how much better drivers/patches/updates do to a video card. Might not be the same benchmarks, but that is to be expect since so many news games did come out.

You want these cards for future gaming, not older games
 
I have no problem with an updated review of both cards, both using the same new set of games and then making a comparison with new drivers.

If you change out the games, though, it completely invalidates any comparison of margins between previous and current tests.
 
MavericK said:
I have no problem with an updated review of both cards, both using the same new set of games and then making a comparison with new drivers.

If you change out the games, though, it completely invalidates any comparison of margins between previous and current tests.
Click to expand...

Totally agree. Still though the benchmarks don't lie, the RX 480 looks to be the better buy for future games and current gen games.
 
MavericK said:
Not if you want to do VR. ;)
Click to expand...

that is SOOOOO fucking true. AMD is so bad in VR right now. VR is a niche scene right now. So AMD has time to fix that....What AMD needs to go is take of that VR PREMIUM READY bullshit on those boxes!
 
Brackle said:
that is SOOOOO fucking true. AMD is so bad in VR right now. VR is a niche scene right now. So AMD has time to fix that....What AMD needs to go is take of that VR PREMIUM READY bullshit on those boxes!
Click to expand...

Maybe some of that Intel money will allow them to create a better VR experience. I doubt that it's the hardware. Probably need to write more optimizations that can be added to game engines. The old school AMD problems have been corrected within the new regime for basic gaming, but shifted to VR. Luckily I won't be purchasing a VR headset soon due to the new factor of them. I am seeing some killer apps though. Just not "Crysis" killer yet. ;) So that's the only thing keeping my finger off the buy button.
 
Brackle said:
that is SOOOOO fucking true. AMD is so bad in VR right now. VR is a niche scene right now. So AMD has time to fix that....What AMD needs to go is take of that VR PREMIUM READY bullshit on those boxes!
Click to expand...

The problem is they are once again being defined as a poor product for another category of use. AMD is their own worst enemy. They better fix it fast.
 
Brackle said:
So since it is missing your baseline you cant believe the benchmarks?....The benchmarks look legit no matter how people try to make it fit their agenda.

99% of websites never go back to see how much better drivers/patches/updates do to a video card. Might not be the same benchmarks, but that is to be expect since so many news games did come out.

You want these cards for future gaming, not older games
Click to expand...


You can't draw a 100% absolute conclusion and this causes the ability to not abstinate from drawing other conclusions, read what I posted in the previous page. Simple 6th grade science class should tell you these things.

The ability to isolate one factor and test upon that one factor is paramount to understanding and drawing a conclusion without that you don't really have anything, you are guessing.
 
Last edited:
razor1 said:
You can't draw a 100% absolute conclusion and this causes the ability to not abstinate from drawing other conclusions, read what I posted in the previous page. Simple 6th grade science class should tell you these things.

The ability to isolate one factor and test upon that one factor is paramount to understanding and drawing a conclusion without that you don't really have anything, you are guessing.
Click to expand...
I can only speak for myself, but I don't think most of us are trying to draw any conclusions about the two cards being X percent faster at the same settings than they were at launch. Again, the point of the article as stated on the first page was to compare current performance of the two cards in the latest games.

I do think the tables on the conclusion page that show old and current performance deltas should have been left out of the article as they could lead one to view the article as more of a driver comparison article than a current performance article. Otherwise I don't see anything wrong with it.
 
Well I can agree with you Schmave , but I can also agree with the writer of the article. If you think of it as two articles that compare the best gaming on those cards; I don't have an issue with the end results.

Article 1: Shootout using the best games available at the time.
Article 2: Revisit the shootout with a new best of gaming list.
Conclusion: Compare results from shootout 1 to shootout 2.

Games fall off the face of the earth in terms of popularity. To run a successful testing website it is imperative to test the newest toys that people are currently playing. Like I said earlier, nobody wants to read about games that they have long finished and none of their buddies are talking about. As an addendum he could go back and add those older titles in to make a few people happy I suppose. I see no point in it myself. I would stick to what's hot and current.
 
cageymaru said:
Never seen [H]ardocp use the same settings when doing reviews unless it is the Apples to Apples section. If some patch or driver made the game run better than before, then they use the better settings, and highlight that video card A could run this and video card B could only operate at these settings. This occurred in the latest Watch Dogs review on the front page if I remember right. All of the people complaining here should go to that comment section and tell Brent why he's wrong too.

So what's the difference in Hardware Canucks using the most up to date settings for the games? If both video cards are running them Apples to Apples then what is the big deal? Hardware gets refined, drivers are streamlined for better efficiency, and settings are adjusted. At least that's how I've been treating my PC hardware for the past 35 years.
Click to expand...
[H] though is not running a game review at multiple times and then comparing to previous reviews to point out how good or bad a card compares. The Apples to Apples comparisons are made within the self-same review.

(Sorry for the big-ass quote but I'm on my phone and editing posts really sucks on it.)
 
They're benching quantum break dx12.

Lol.

Anyway in the OP of this thread it says something about looking at driver improvements, where is the data comparing old and new in the same games?
 
Ieldra said:
They're benching quantum break dx12.
Click to expand...
Sites got criticized for including Project Cars in their benches because it's a 'broken game' but somehow Quantum Break is just fine.
It's even better once we realized Nvidia's DX11 mode outperforms DX12 on both brands... Whoopsy.

Oh right, something something GameWorks. My mistake.
 
Algrim said:
[H] though is not running a game review at multiple times and then comparing to previous reviews to point out how good or bad a card compares. The Apples to Apples comparisons are made within the self-same review.

(Sorry for the big-ass quote but I'm on my phone and editing posts really sucks on it.)
Click to expand...

Could have sworn that [H]ardocp has done driver improvement shootouts over the years. I would link some, but the video card reviews links are broken for me after about a year of testing. I think I remember one when around the time that this website discovered that AMD had runt frames and then AMD fixed it.

But since I can't look it up it's a moot point. ;) You could be right though.
 
Schmave said:
But is that the point of the review? On the first page he states:


He then goes on to say:


To me (and if you read the rest of the first page) it sounds like the point of the article is to compare the two cards as they stand in the market today, against the current crop of relevant games. This includes performance in the latest games as well as the latest prices for each card. It is not to see how drivers have improved performance relative to launch drivers.

I don't understand why people are getting so bent out of shape about this.
Click to expand...
Yes you got it. Sad most of the others cant comprehend what is clearly stated in the article.
 
Schmave said:
I can only speak for myself, but I don't think most of us are trying to draw any conclusions about the two cards being X percent faster at the same settings than they were at launch. Again, the point of the article as stated on the first page was to compare current performance of the two cards in the latest games.

I do think the tables on the conclusion page that show old and current performance deltas should have been left out of the article as they could lead one to view the article as more of a driver comparison article than a current performance article. Otherwise I don't see anything wrong with it.
Click to expand...


err did you the first post that made this thread, and lets keep on reading cause there are many people that have stated the exact same thing.
 
Pieter3dnow said:
You mean all the Nvidia trolls in the AMD forum ?
Click to expand...


what? where did that come from, the only troll here is you man, cause everyone else is discussing the topic.

If you don't like what someones says doesn't make them a troll, but if you call a person a troll because they are talking with facts, that makes you a troll because you don't know how to respond to those facts. which many of your "fans" tend to do, because there isn't much going on upstairs, which yeah until ya get help, that probably won't change.
 
razor1 said:
what? where did that come from, the only troll here is you man, cause everyone else is discussing the topic.

If you don't like what someones says doesn't make them a troll, but if you call a person a troll because they are talking with facts, that makes you a troll because you don't know how to respond to those facts. which many of your "fans" tend to do, because there isn't much going on upstairs, which yeah until ya get help, that probably won't change.
Click to expand...
sorry but when the lions share of the posts ate arguing they didn't use the same games in both tests when the point of the article was more a snapshot of current releases and how they fair currently at each moment in time, then yes they are trolling. As linked the premise of the test was given and thesis established. Just because it doesn't fit the individual concerns of some does not invalidate the results. Besides I am shocked a bit SKYTML actually did this piece.
 
Summary of relevant parts of thread: Article conclusion was fairly misleading. Article should have been broken into two segments; performance changes in original games via driver updates and then performance comparison between the two cards in a range of current, popular titles. Some data still useful, other parts not.

Intersperse these relevant points with 1.5 pages of bickering and people not reading previous posts properly.
 
JustReason said:
sorry but when the lions share of the posts ate arguing they didn't use the same games in both tests when the point of the article was more a snapshot of current releases and how they fair currently at each moment in time, then yes they are trolling. As linked the premise of the test was given and thesis established. Just because it doesn't fit the individual concerns of some does not invalidate the results. Besides I am shocked a bit SKYTML actually did this piece.
Click to expand...

Sorry but when the lion's share of the posts ate arguing that the point of the article isn't to look at performance change over driver releases, but just at performance as it stands now, despite the OP explicitly stating that this is the aim and linking a table showing relative performance of the rx480 and 1060 now and at launch then yes, they are trolling.

There is absolutely nothing special about this review other than the *claim* that it assesses driver improvements, which is false.

I was going to read the hardware Canucks introduction and copy paste it here when I realized cageymaru did the work for me before he abandoned all sense and started arguing against what he wrote in the OP.

gtx-1060-update-101-jpg.11915


If the aim isn't to assess driver improvements why does this exist?

What sense does it make to distinguish between relative performance in May and in November if the same suite of games wasn't used?

If I ran 5 miles in 5 minutes in May, and I ate 6 hot dogs in 5 minutes in November, have I improved 20%?

I don't think so.
 
I saw the video on the youtube. I disagreed with them as soon as they said 'stock settings'. Ignoring overclocking is downright stupid for these cards. The 1060 and RX480 are neck and neck until you start OCing. I know the RX480 was having issues overclocking much more than what it comes as out of the box. Still possible, but not as much as the 1060 was pulling off.

If they wanted to only focus on driver updates, they should be using the reference boards in their tests instead of factory overclocked boards from 3rd parties.

Just my two cents.
 
skeeder said:
I saw the video on the youtube. I disagreed with them as soon as they said 'stock settings'. Ignoring overclocking is downright stupid for these cards. The 1060 and RX480 are neck and neck until you start OCing. I know the RX480 was having issues overclocking much more than what it comes as out of the box. Still possible, but not as much as the 1060 was pulling off.

If they wanted to only focus on driver updates, they should be using the reference boards in their tests instead of factory overclocked boards from 3rd parties.

Just my two cents.
Click to expand...

They included BOTH reference cards and mildly overclocked factory cards from both camps. You do realize that MOST consumers do not overclock their systems. That's why we read and post at places like [H]ardocp instead of Tom's Hardware. :) If you want a balls to the walls OC comparison then this was the WRONG review to read. ;) But to dismiss it because their readership probably doesn't OC their cards is kinda silly. Especially when one cards may or may not OC as well as another. How would you feel if you bought a card based on OCs to the moon and your card struggled to get out of factory OC land?
 
Ieldra said:
Sorry but when the lion's share of the posts ate arguing that the point of the article isn't to look at performance change over driver releases, but just at performance as it stands now, despite the OP explicitly stating that this is the aim and linking a table showing relative performance of the rx480 and 1060 now and at launch then yes, they are trolling.

There is absolutely nothing special about this review other than the *claim* that it assesses driver improvements, which is false.

I was going to read the hardware Canucks introduction and copy paste it here when I realized cageymaru did the work for me before he abandoned all sense and started arguing against what he wrote in the OP.

gtx-1060-update-101-jpg.11915


If the aim isn't to assess driver improvements why does this exist?

What sense does it make to distinguish between relative performance in May and in November if the same suite of games wasn't used?

If I ran 5 miles in 5 minutes in May, and I ate 6 hot dogs in 5 minutes in November, have I improved 20%?

I don't think so.
Click to expand...

To hell with you. How's about that?
 
Ieldra said:
Sorry but when the lion's share of the posts ate arguing that the point of the article isn't to look at performance change over driver releases, but just at performance as it stands now, despite the OP explicitly stating that this is the aim and linking a table showing relative performance of the rx480 and 1060 now and at launch then yes, they are trolling.

There is absolutely nothing special about this review other than the *claim* that it assesses driver improvements, which is false.

I was going to read the hardware Canucks introduction and copy paste it here when I realized cageymaru did the work for me before he abandoned all sense and started arguing against what he wrote in the OP.

gtx-1060-update-101-jpg.11915


If the aim isn't to assess driver improvements why does this exist?

What sense does it make to distinguish between relative performance in May and in November if the same suite of games wasn't used?

If I ran 5 miles in 5 minutes in May, and I ate 6 hot dogs in 5 minutes in November, have I improved 20%?

I don't think so.
Click to expand...

I like Michael's work at HardwareCanuck but discussed this with him offline and others have in the forum there, fair to say he does not agree with how the conclusion is skewed but there are some logical holes.
He explains he removed RoTR DX11 because the game is going to be removed for newer ones, but yet keeps DX12 that works better for 480.
The point as you say was exactly about seeing how gaming has evolved from July to now, and cricitally this was not just about drivers but also settings/options, this is his explanation why he has increased setting on Quantum Break to less game play friendly settings.
Quantum Break now uses settings that is far removed from July, back then Nvidia/AMD were closer but with the new settings that are unplayable even for a 480 without Freesync makes the gap much larger.
This is further skewed that we do not have DX11 of this game, while also DX11 is skewed because it is now omitting RoTR that would had put DX11 performance wider between the two cards.

If he intended to use advanced options to see how well this settings and post-processing effects work and ignoring the crud FPS, then he should also apply to games such as Gears of War 4, and if he had the DX12 gap between 480/1060 lowers - see results on HardOCP that also use PresentMon.
I could go on.
But I think he has taken on too much scope with this project as not only is it about previous games but also a crud load new ones and also revising settings.
In reality the scope should include both DX11 and DX12 APIs if a game has them, and needs much stronger framework on deciding the option settings as there is no consistency (such as what we now have with Quantum Break compared to Gears of War 4).
I could mention more but it is not going to change anything, just to say most need to look at the results individually and also to the previous ones along with settings.
The changes to Quantum Break is enough to increase that % by several points.
And finally there is the analysis needed to see whether optional settings are comparable between AMD/Nvidia and decide whether they should be enabled; such as Fallout 4 where in his setup the 480 is pretty much equal to a 1060 and comes down to godrays with HBAO+, same can be said about PureHair and if enabled creates more work for RoTR as it does more on AMD hardware at max settings.
Cheers
 
JustReason said:
sorry but when the lions share of the posts ate arguing they didn't use the same games in both tests when the point of the article was more a snapshot of current releases and how they fair currently at each moment in time, then yes they are trolling. As linked the premise of the test was given and thesis established. Just because it doesn't fit the individual concerns of some does not invalidate the results. Besides I am shocked a bit SKYTML actually did this piece.
Click to expand...

Wrong look up the definition of a troll

You don't even know that pathethitic

Anyways back to the topic

If you remove all games that were added and the games that have different settings over the different Apis the change is only within 3% in favor of the rx480 Eye balling it btw.

Yeah that is with in the Margin of error!

All this for a max 3%?

Impressive amount of intelligence needed...........
 
Last edited:
razor1 said:
Wrong look up the definition of a troll

You don't even know that pathethitic

Anyways back to the topic

If you remove all games that were added and the games that have different settings over the different Apis the change is only 3% in favor of the rx480 Eye balling it btw.

Yeah that is with in the Margin of error!

All this for 3%?

Impressive amount of intelligence needed...........
Click to expand...
You can keep saying this, as I have been, and people still won't get it.
Then it turns into a debate about the game choices, which is a debate you could have about ANY benchmark. Did they choose too many Nvidia games? Too many AMD games? etc.

The point here is, even with an AMD slant they couldn't produce FineWine results so they had to resort to rigging the benchmark. Frankly this has hurt my opinion of HardwareCanucks quite a bit, although I usually only look at their chassis reviews on YouTube.
 
I completely agree with what ya have stated,

Not sure about Hardware Cunucks not going to take sides on that one, they f'ed up ok, lets see if they fix it.
 
cageymaru said:
Could have sworn that [H]ardocp has done driver improvement shootouts over the years. I would link some, but the video card reviews links are broken for me after about a year of testing. I think I remember one when around the time that this website discovered that AMD had runt frames and then AMD fixed it.

But since I can't look it up it's a moot point. ;) You could be right though.
Click to expand...

Yes, we have...

http://www.hardocp.com/article/2012/11/12/fall_2012_gpu_driver_comparison_roundup

http://www.hardocp.com/article/2013/03/04/2012_amd_nvidia_driver_performance_summary_review

http://www.hardocp.com/article/2013/01/21/2012_nvidia_video_card_driver_performance_review

http://www.hardocp.com/article/2013/02/18/2012_amd_video_card_driver_performance_review

and there are lots more....

and these will not be the last....
 
I doubt HWC will reconcile or even admit to some...massaging of the message here (har har). If we're being honest, it's clear they were looking for some way to portray the RX480 in the best light possible. Between their traditional website content vs YT videos? I think it's clear they value the production side a great deal more. That's understandable, considering how funny some of those videos were (see their Computex/Taiwan vids).

But when it comes to in-depth, hardware reviews, HWC has been inconsistent in general, so it doesn't surprise me that their revisit of Polaris would generate additional questions.
 
Brent beat me to it, but yes, [H] used to do Driver Reviews. I haven't seen any in years (and 2012, yeah, 4 years ago, so there you are).

Arguing statistical analysis aside, the article did one thing for me - it showed me that AMD's driver situation has vastly improved.

Enough so that I'll be grabbing an RX480 for the wife's system :)
 
 
Not sure what the fuss is, but HardOCP has constant updates to current games with most recent patches/updates, most recent drivers etc. as time goes on. In other words current day results when written. I am not even sure you could do a valid driver analysis unless you keep the game at the same state (no updates or patches), same Windows state, same hardware, bios and hardware drivers etc. So what is even the point? It will be likely something has changed, new driver may not even work right without the Window update etc.

Now comparing the two cards, 1060 and 480 to me is virtually a wash, gaming experience are about the same depending upon the game as well.
 
noko said:
I am not even sure you could do a valid driver analysis unless you keep the game at the same state (no updates or patches), same Windows state, same hardware, bios and hardware drivers etc.
Click to expand...
Yes that's pretty easy, I could run DDU right now and install launch day 980 Ti drivers and compare them to today's drivers. Everything else would be the same.
I could test new games (assuming they physically ran on old drivers) as well as improvements to games which came out pre-launch (games that will run on both old and new drivers). I could also compare their day-one drivers for each game to today's current driver.

There are ways to compare driver improvements but swapping games out of the test suites, then directly comparing the two benchmarks, is not a good way to do it.

Their results should go straight into the garbage can, it's not even really a debatable topic imo. The fact that we are still discussing this means too many people are desperately trying to draw conclusions where none exist. Still waiting for HardwareCanucks to pull the review, frankly. Or at least edit the last page since their % change number is off by about three-fold, as we've already proven.
 
Last edited:
One big consideration when doing a driver-to-driver comparison over a moderate amount of time.
At times certain AAA releases will have driver released pertaining to them with specific optimisation, this can create a skew that it looks like it is just driver general update improvements but that is not always the case, it can also be tied to a driver-game optimisation that carries on through to the more recent driver.
This is relevant if any of those games are part of the comparison using old driver pre-game and then the latest.
So need to ensure either no specific game-optimisation was introduced after original driver or such games are excluded or only use games that existed at time of the original driver, which unfortunately really limits the usefulness of the scope.
Such things can skew the conclusion context, tricky to differentiate improvements down to specific game optimisation or those that are more general driver improvements.
Only way to be sure is to use games that existed when the original driver was done, kinda boring and really not helpful for DX11/DX12 that is still immature in terms of game development.

Cheers
 
noko said:
I am not even sure you could do a valid driver analysis unless you keep the game at the same state (no updates or patches), same Windows state, same hardware, bios and hardware drivers etc. So what is even the point?
Click to expand...
good point not a true scientific analysis but thats a pita and I think people are nitpicking. old vs new on same game version is enough I think.

TaintedSquirrel said:
Yes that's pretty easy, I could run DDU right now and install launch day 980 Ti drivers and compare them to today's drivers. Everything else would be the same.
I could test new games (assuming they physically ran on old drivers) as well as improvements to games which came out pre-launch (games that will run on both old and new drivers). I could also compare their day-one drivers for each game to today's current driver.
Click to expand...
this is what noko is saying. you can roll back or change the vid driver but all the other variables would have to be kept constant to make it as scientific and the nitpickers want.
 
noko said:
Not sure what the fuss is, but HardOCP has constant updates to current games with most recent patches/updates, most recent drivers etc. as time goes on. In other words current day results when written. I am not even sure you could do a valid driver analysis unless you keep the game at the same state (no updates or patches), same Windows state, same hardware, bios and hardware drivers etc. So what is even the point? It will be likely something has changed, new driver may not even work right without the Window update etc.

Now comparing the two cards, 1060 and 480 to me is virtually a wash, gaming experience are about the same depending upon the game as well.
Click to expand...

If it weren't for vr I'd have bought an rx480, but it gets destroyed if you factor that in, unfortunately.
 
pendragon1 said:
good point not a true scientific analysis but thats a pita and I think people are nitpicking. old vs new on same game version is enough I think.

this is what noko is saying. you can roll back or change the vid driver but all the other variables would have to be kept constant to make it as scientific and the nitpickers want.
Click to expand...

You would have to rollback all of the Windows 10 updates like Anniversary also. Good luck with that.
 
You must log in or register to reply here.
Back
Top