Let's Talk about Fairly Testing GPU Performance R.E. Warming Up GPUs

Brent_Justice

Moderator
Joined
Apr 17, 2000
Messages
17,755
Intro

This topic is a pet peeve of mine, but is also relevant in the soon to come AMD Nano reviews. It also applies to all NVIDIA and AMD video card testing with today's latest power management and GPU Boost features. I am talking about, warming up your video card prior to testing performance. Let's talk about that.

How GPU Clock Speed Varies

In today's modern GPUs the clock speed is in essence a variable. The clock speed in AMD and NVIDIA GPUs is dynamic and able to change based on many factors including thermals, power and even down to small differences in binning. We aren't going to get specific into how that works today, today I want to talk to you about testing video cards, and the right way to do it knowing how GPU clock speed works in modern GPUs.

For NVIDIA, this is called GPU Boost. You start with a base and boost clock, and the GPU dynamically changes the clock speed upwards as needed. However, it is also capable of going the opposite direction, and lowering the clock speed if needed. Typically, in gaming, you will find the real-world in-game frequency higher than the boost clock. For AMD, it works differently.

AMD provides an "Up To" clock speed rating. This is what the GPU is capable of reaching, but it may often not. Instead, the clock speed is below the rated clock speed, and has the ability to reach up to it, if possible. These are two very different ways of looking at clock speed, but just note that in both methods the clock speed can change over time while gaming.

Clock Speed over Time

This is the heart of the matter. The Clock Speed can change over time while gaming. For NVIDIA, this means a lower boost clock speed. For AMD, this can mean the clock speed is under the rated "up to" clock speed. This time is also variable, it could be after 5 minutes of gaming, or after 15 or 30 minutes of gaming. The reason why it changes is due to the thermal properties, mainly. As you play games over time, the video card heats up, the components in your system heats up, the ambient air temperature heats up. This heating up, can cause your clock speed, to go down over time.

For example, your video card may start out at 950MHz for the first 5 minutes of playing a game. However, after 15 minutes it may drop to 900MHz. After 30 minutes it may drop to a much lower 800MHz. This means you are losing performance over time, but it also means that if you tested for performance at the 5 minute mark, versus the 30 minute mark, your results would be very different.

In modern GPUs, it is IMPERATIVE to test performance after the video card as "warmed up" after say 15-30 minutes of playing a game. Else, you are only reporting the higher performance, and ignoring the real-world performance after playing games for a while.

How to Test

This is how GPU testing should be done:

A.) The very first thing should be to test and find out what the real-world in-game clock frequency is while gaming over time, not the base clock, not the boost clock. This requires usage of a third party utility like Afterburner. In Afterburner, you can enable an OSD on your screen that shows the actual GPU clock frequency. The first thing you need to do is play a game for 30 minutes and monitor this GPU frequency. You need to see what it starts off as, and how it changes over time, and the final clock speed it settles to consistently. It is the consistent clock speed you are looking for. This then needs to be reported in the review. This is the clock speed you need to test at.

B.) Then, when you do your testing, benchmarks, gaming, whatever, do it after this time period of playing games that have warmed up your GPU to its consistent real-world clock speed. This then results in your performance indicating the real-world performance regular gamers who install this video card, will get while gaming.

IF you do a short benchmark, say 5 minutes, this will show an inflated result. 5 minutes will not be long enough to warm up the card to its real-world frequency. 5 minutes will show you the high end 950MHz performance, but will not show you the performance when after gaming for 15-30 minutes it drops to 800MHz. The benchmark isn't long enough. Therefore, you must also have a test that lasts long enough to show the real-world performance. Testing from a "cold start" will always show inflated results.

Summary

Therefore, A.) the reviewer must find out what the dynamic clock speeds are and report the actual clock speed after a long period of gaming, B.) test video cards after this warm up period, C.) use tests that are long enough to realize these real-world clock frequencies and never do a cold start test.

This is how "FAIR" review sites, test cards. This is how we, at HardOCP, as a "FAIR" website, test video cards. This is how we would have tested Nano.

As you look forward to reviews this week, keep this topic in mind.
 
If they want to use [H]ardOCP's seal of approval on their marketing they need to pass the tests. Letting the cards warm up to obtain a real world sense of frame rates makes sense.

AMD made a huge PR blunder here.
 
MAWFfOp.jpg
 
I come on Brent, you have to use canned benchmarks. That's the only way to be fair.

Sarcasm aside. It is a sad state when you have to make a post for people to explain how you guys do actually have fair reviews.

It's just 1 company doesn't like that you play games. I guess gamers don't care about playing the games, just want to run benchmarks?

Keep up the good work!
 
If that's your takeaway from the above article, then this is probably not the review site for you .

I read the entire post and I completely understand it and agree with it. I was jiving right up until the end when the Nano is brought up again.

We all know its a sore topic for Hardforum but all the bitching won't change it. You didn't get a card, move on.
 
Last edited:
This is an important topic, and should not be swept under the rug, and something to look for in all reviews in regards to NVIDIA and AMD GPUs. It makes a big difference in the performance results shown. When video cards can dynamically change their clock speed, and it can vary as much as 200MHz in 30 minutes, you betcha it can make a big difference. It is easy to inflate results, it is harder and takes more time to show proper results.
 
I come on Brent, you have to use canned benchmarks. That's the only way to be fair.

Sarcasm aside. It is a sad state when you have to make a post for people to explain how you guys do actually have fair reviews.

It's just 1 company doesn't like that you play games. I guess gamers don't care about playing the games, just want to run benchmarks?

Keep up the good work!

They also cut out TR that does extremely technical analysis during reviews and TPU which does a huge number of games in their benchmarks. I'm sure AMD knew they could not blackmail the larger sites so they just bullied a few smaller sites instead.
 
So is the assumption here that the Nano is going to throttle a lot - and only hardocp would have properly tested the card and shown this throttling in practice? Isn't that sort of selling other hardware sites a bit short? I believe Anandtech is getting a card for review - and they haven't always been particularly kind to AMD (R9 290 review caption was "HOT AND LOUD") - but they have been pretty good at doing fairly thorough reviews in the past. I wouldn't assume that every review of the Nano is going to be extremely positive just because AMD didn't send cards to some sites.
 
So is the assumption here that the Nano is going to throttle a lot - and only hardocp would have properly tested the card and shown this throttling in practice? Isn't that sort of selling other hardware sites a bit short? I believe Anandtech is getting a card for review - and they haven't always been particularly kind to AMD (R9 290 review caption was "HOT AND LOUD") - but they have been pretty good at doing fairly thorough reviews in the past. I wouldn't assume that every review of the Nano is going to be extremely positive just because AMD didn't send cards to some sites.

I'm just telling you how we do it, and it is important to look for when evaluating performance in GPUs.

If you don't, it doesn't exactly represent the real-world and what the gamer gets when they game with it.
 
So is the assumption here that the Nano is going to throttle a lot - and only hardocp would have properly tested the card and shown this throttling in practice? Isn't that sort of selling other hardware sites a bit short? I believe Anandtech is getting a card for review - and they haven't always been particularly kind to AMD (R9 290 review caption was "HOT AND LOUD") - but they have been pretty good at doing fairly thorough reviews in the past. I wouldn't assume that every review of the Nano is going to be extremely positive just because AMD didn't send cards to some sites.

Several sites have been cut out inlcuding Techreport and TPU. Anandtech was recently sold to the same corporation that bought Tom's Hardware. So I no longer consider them an independent review site.
 
They also cut out TR that does extremely technical analysis during reviews and TPU which does a huge number of games in their benchmarks. I'm sure AMD knew they could not blackmail the larger sites so they just bullied a few smaller sites instead.

TR I can understand. They are very heavily Bias.

TPU And [H] make no sense to me. It's AMD loss either way you look at it.
 
GPU thermal throttling is obnoxious, I wish the cards would just have a single frequency mode for games to provide more stable framerates. I don't care if the frame rate is slightly higher if it fluctuates all over the place it's hugely more distracting. It's not ever easy to disable it if you want to, in fact it's nearly impossible.
 
The testing methodology makes sense.
However, in my real world experience, I don't see a difference.
As you can see the specs in my sig, we all know how the clocks on these cards work. Even before swapping out the coolers on my cards, I could play BF4 for hours on end with the GPU's maintaining "Boost" clocks. the 280x when it was all on its lonesome always stayed at 1GHz. Never once did I have observe throttling (studied via GPUz logging).

So, while you claim this is a good idea (and I tend to agree), have you actually witnessed a bouncing of clock frequency due to temp throttling between "Boost" and standard clocks?
 
The testing methodology makes sense.
However, in my real world experience, I don't see a difference.
As you can see the specs in my sig, we all know how the clocks on these cards work. Even before swapping out the coolers on my cards, I could play BF4 for hours on end with the GPU's maintaining "Boost" clocks. the 280x when it was all on its lonesome always stayed at 1GHz. Never once did I have observe throttling (studied via GPUz logging).

So, while you claim this is a good idea (and I tend to agree), have you actually witnessed a bouncing of clock frequency due to temp throttling between "Boost" and standard clocks?

The 290x was a culprit of this. When not warmed up, it would run full clocks in benchmarks but when actually gaming on it for a period of time - it would throttle back.

Some review sites noticed this - others did not -- all due to the testing method.

Of course custom coolers/WC fixed it but it still was an issue.

NVIDIA cards throttle as temperatures go up but stay above the stated stock clock speed.
 
Last edited:
This is an important topic, and should not be swept under the rug, and something to look for in all reviews in regards to NVIDIA and AMD GPUs. It makes a big difference in the performance results shown. When video cards can dynamically change their clock speed, and it can vary as much as 200MHz in 30 minutes, you betcha it can make a big difference. It is easy to inflate results, it is harder and takes more time to show proper results.

Do you ever tell people how much the difference is between warm and cold clocks? I think that would be very important info.
 
So is the assumption here that the Nano is going to throttle a lot - and only hardocp would have properly tested the card and shown this throttling in practice? Isn't that sort of selling other hardware sites a bit short? I believe Anandtech is getting a card for review - and they haven't always been particularly kind to AMD (R9 290 review caption was "HOT AND LOUD") - but they have been pretty good at doing fairly thorough reviews in the past. I wouldn't assume that every review of the Nano is going to be extremely positive just because AMD didn't send cards to some sites.

Assumption? [H] has all but guaranteed it. They are also all but guaranteeing that the other sites are going to cover it up.

I've seen sites jump through hoops to point out and create throttling with Hawaii. Running cards in 40° ambient "hot boxes" and even holding their hand over the blower.

Don't worry, if it throttles, it will be known. If it can be made to throttle, they will do it. nVidia will figure out how to get the worst performance out of Nano and spread the word.
 
§kynet;1041846500 said:
All cards should be tested in a real world environment, like this.

Meanwhile, in outside of the matrix real world...
 
O yea everyone runs a Computer without a case and has it open like that. It has the best airflow!
 
Yeah.. that's sort of the funny thing about this "guide" about fairly testing GPU performance.

B.) Then, when you do your testing, benchmarks, gaming, whatever, do it after this time period of playing games that have warmed up your GPU to its consistent real-world clock speed. This then results in your performance indicating the real-world performance regular gamers who install this video card, will get while gaming.

except in an open bench scenario most cards probably won't throttle - and in a closed case they possibly will. Most gamers are probably going to be installing these GPU's in closed case systems. Most of these aftermarket cards don't use blower type coolers, and just dump all the air back into the case. It kind of seems like this guide went over something pretty important. 15-30 mins of warm up won't make much difference if you aren't testing in closed case environment.
 
Yeah.. that's sort of the funny thing about this "guide" about fairly testing GPU performance.



except in an open bench scenario most cards probably won't throttle - and in a closed case they possibly will. Most gamers are probably going to be installing these GPU's in closed case systems. Most of these aftermarket cards don't use blower type coolers, and just dump all the air back into the case. It kind of seems like this guide went over something pretty important. 15-30 mins of warm up won't make much difference if you aren't testing in closed case environment.

Is this where i point out my reference 290x's never pass like 84C under CONSTANT full load for hours in my Raven RV01 and now that I undervolted never hit 80C at all?
 
Summary

Therefore, A.) the reviewer must find out what the dynamic clock speeds are and report the actual clock speed after a long period of gaming, B.) test video cards after this warm up period, C.) use tests that are long enough to realize these real-world clock frequencies and never do a cold start test.

This is how "FAIR" review sites, test cards. This is how we, at HardOCP, as a "FAIR" website, test video cards. This is how we would have tested Nano.

As you look forward to reviews this week, keep this topic in mind.

True. :)

But, what would you do with the benchmark suit when reviewing the Nano? You guys spend a lot of time playing the games and one of the reasons why I value [H] reviews so much is because you write about the gaming experience which means more then the FPS numbers IMHO. You do however have a limited selection of games due to this, compared to other sites, so I would have preferred that your benchmark suit would have been less vendor biased then it is today. Your reviews makes a general statement about the cards according to the benchmark suit and it IS a card review, not a game review.

Doesnt matter much that you warm up the cards so much that your asses catches fire, if the benchmark suit is biased and not neutral to begin with. I understand that you select your games according to popular titles, but you already make reviews spesifically of those games of interest and how different cards perform in those.

In my opinion, [H] is one the most unbiased sites on the internet and have no issues giving different vendors crap if they deserve it no matter who that is. Your choice of games today when you review CARDS paints another picture though. For those that are not familiar with [H] could with good reasons think that [H] scewed the reviews to begin with due to this to affect the outcome.
 
If you have a problem with the features the game developers implement in their games, take that issue up with them.

Games that either use features from NVIDIA. or AMD, are not "vendor biased" or biased toward either GPU. We have dispelled this myth on several occasions. One of the most popular game titles out right now, The Witcher 3, for example, takes the same burden enabling HairWorks on NVIDIA or AMD GPUs, as we have tested, dispelling the myth. It is one of the most neautral games in terms of performance, trading blows between NV and AMD. Do not eat the hype, stick with the facts. Just because it has GameWorks features doesn't mean it is vendor biased. Just because it has TressFX, doesn't mean it is vendor biased. It goes both ways. We had no complaints when we used the TressFX in Tomb Raider all its time during use.

Do not confuse sponsorship's with game features either. A game sponsored by AMD or NVIDIA does also not mean it will automatically be faster on either GPU. A sponsorship is very different from the game developer chosing to implement NVIDIA or AMD supplied technology.

Selecting games, based on who's technology it implements, is in itself a biased act. The game suite is an ever changing suite of games, as new games come out, we will implement new games. It takes time to do so, so be patient.
 
If you have a problem with the features the game developers implement in their games, take that issue up with them.

Games that either use features from NVIDIA. or AMD, are not "vendor biased" or biased toward either GPU. We have dispelled this myth on several occasions. One of the most popular game titles out right now, The Witcher 3, for example, takes the same burden enabling HairWorks on NVIDIA or AMD GPUs, as we have tested, dispelling the myth. It is one of the most neautral games in terms of performance, trading blows between NV and AMD. Do not eat the hype, stick with the facts.

Do not confuse sponsorship's with game features either. A game sponsored by AMD or NVIDIA does also not mean it will automatically be faster on either GPU. A sponsorship is very different from the game developer chosing to implement NVIDIA or AMD supplied technology.

Selecting games, based on who's technology it implements, is in itself a biased act. The game suite is an ever changing suite of games, as new games come out, we will implement new games. It takes time to do so, so be patient.

Witcher 3? I love the game, have 46H in it so far, playing it on a 980 on my HTPC. Should have been excluded from the benchmark suit by the fact alone that the game used 64x tessellation with 8X MSAA on Hairworks by default. Should ring a little warning bell when selecting that game for a review.

That a game sponsored by a vendor doesn´t mean its automatically faster on their GPU. However, any game where vendors doesn´t have an equal oportunity to optimize for the game should be excluded from a benchmark when doing a card review.

Another example for argument sake would have been Tomb Raider earlier. Nvidia didn´t have a chance to optimize for the game, since they got the latest game code rather late. Should automatically have been excluded from any card review then.

Reading a review where the game selection can almost give you a prediction of the conclusion without ever reading the review is a bit boring.
 
If you play that game (no pun intended), the game list will be pretty narrow.

Besides, if the game has neither GameWorks or TressFXetc, how do we know the game engine itself is not biased against one of the card vendors?

You are just cherry picking games based on the Tin, and taking absolutely nothing else into account, which is by itself unscientific.

If complete impartial game selection is required, one of the following must be done:

1. Have a deep analysis of the game code to ensure that both AMD and nVidia cards, with absolutely equal GPU power (a statistical impossibility), run exactly the same frame rates on both (another impossibility).

2. Just choose the games and ignore everything else.

Neither of which would directly involve GW and TressFX, as there is absolutely no scientific proof that using one would gimp the otherside, and there is no scientific proof to prove otherwise.

Under these circumstances, where effects of GW etc are completely unknown on any specific GPU, it makes zero sense to cherry pick the games based on that alone.
 
Last edited:
Games are games, if we chose not to use Witcher 3, a game everyone and their grandma was looking forward to, and is playing (it is quite a good game), what kind of value would be delivering to the gamer via video card reviews?

We cannot and will not base our game choices on who's technology the game developer has chosen to use, and or how it is implemented.

We cannot ignore games based on who's tech is in use.

What we will do is explore that game, and let you know how video cards perform in it, how image quality looks, and which one provides the best value.

The only way to know how game A performs on video cards, is to test game A.

If a lot of games are using NVIDIA supplied technology, that is just the way it is right now, we cannot ignore those games. It would be exactly the same if all the games were using TressFX or an AMD technology. If you see one vendor's tech being used a lot right now in new games by game developers, it has to mean something. If you have issue with what they have chosen to use in their own games, take it up with them.

Here's what is going to happen, a year from now a lot of games might be using AMD tech, and that is just the thing at the time, and we use a lot of those games. Come back in a year, and make the same complaints that we have too many "AMD biased" games being used. I urge you to, so I can point you back to these kind of comments.

I suggest trying to read each review without bias, and come to a conclusion based on the data presented, as we do.
 
Last edited:
Think of it like this, it is a cycle.

You may have a bunch of games that game developers have chosen to use NVIDIA tech right now. But a year from now, we may be in a cycle where they chose to use AMD tech.

At any given time you are going to have your games that use NV tech, and or AMD tech. It is up to the developer.

It just is, it is just the way things are, we cannot ignore games either way, else we'd have nothing to use!

As a poster above stated, how are you to know the game engine itself isn't biased toward a vendor?

You just cannot do that when it comes to playing games and reviewing video cards. You just have to say, it is what it is, the game dev used that tech, so let's just see how it performs and looks and how video cards compare and tell the end user what the best value is to play that game. It is that simple.

How things are now, in this cycle, may not always be. Right now game devs are choosing to use NV 3D effects, that means something, and it is just the way it is, that's the cycle we are in. Perhaps AMD should up its game in terms of evangelizing its features to game developers. AMD needs to make an impact with game devs and prove why their stuff is better. In the end, the game developer chooses what to use.
 
Real world testing for real world games.

[H]ard gives you all the data (FPS graph) besides settings.

Might add a GPU temp data to the chart too, but besides that [H]ard's reviews gives you all the data needed to gauge performance.

And for "controversy"...
They called out Infinium Labs.
They have called bad hardware/bad behavour with NO bias as long as they have produced hardware reviews.

I will bet you...in a few years, [H]ard will once again be called "biased towards AMD"....by other people.
It's the way the "pendulum" has swung for a small (but always very highpitched and vocal) fangroup since the dawn of the forums.

Botomline:
Warm GPU = how I game.
I seldom game for just 10-15 minute sessions...often it is +2 hours continued.
 
Games are games, if we chose not to use Witcher 3, a game everyone and their grandma was looking forward to, and is playing (it is quite a good game), what kind of value would be delivering to the gamer via video card reviews?

We cannot and will not base our game choices on who's technology the game developer has chosen to use, and or how it is implemented.

What we will do is explore that game, and let you know how video cards perform in it, how image quality looks, and which one provides the best value.

The only way to know how game A performs on video cards, is to test game A.

If a lot of games are using NVIDIA supplied technology, that is just the way it is right now, we cannot ignore those games. It would be exactly the same if all the games were using TressFX or an AMD technology. If you see one vendor's tech being used a lot right now in new games by game developers, it has to mean something. If you have issue with what they have chosen to use in their own games, take it up with them.

I suggest trying to read each review without bias, and come to a conclusion based on the data presented, as we do.

I´ve probably read every review you´ve made on [H] and never had any doubt that you are unbiased and I read the reviews more then just looking at the numbers. Problem with reviews and a limited selection of games, is that the choice of benchmark suit can make the review biased by itself. A different selection of games would change the conclusion, since the conclusion is build upon the benchmark suit.

Some buy cards based upon how they perform in spesific games. Others, like me, buy cards according to how they perform in general, since I have more then a few games. If I wanted to know how a card perform in a spesific game like Witcher 3, I wouldnt be disappointed since [H] have their own Witcher 3 game performance spesific review. Its enough to extrapolate how it will perform if I choose a higher or lower performance card then those reviewed there.

When I read a card review, I prefer that the benchmark suit is made as vendor neutral as possible, since I need to know how the card performs in general and not just in a selected few number of games. Some review sites choose a benchmark suit which is an even mix between games that lean towards the different vendors to balance them out, while some tries to find games where vendors both have optimized.

If a review site have a small selection of games, most of those games are biased towards a single vendor and that review site makes a general conclusion of a CARDS performance without reservation that the review is based upon vendor biased benchmarks, the conclusion becomes worthless and a joke no matter how well the review is written. In those cases, its no longer a CARD review, but a bunch of game performance reviews bundled together.

To bring it back to topic, warming up the GPU before taking numbers ARE the fair way to do a review. First then you get a fair shake of how the cards really perform. Problem is that all that fairness goes away the moment the benchmark suit is biased and you make a general conclusion based upon a biased benchmark. I see your arguments about why you sometimes (and not to affect the outcome) choose a benchmark suit that is biased, but you should at least then acknowledge that and say in the conclusion that you are aware that the benchmark is biased and that the conclusion must not be taken as a general statement of the cards performance. Its not a card review anymore then, just a bunch of game performance reviews bundled up.
 
Last edited:
Back in 2013 HardOCP's test suite was dominated by AMD Gaming Evolved titles -

GTX 780ti, 780, 770 and r9 290 launch reviews with 3 out 4 being AMD titles -
http://www.hardocp.com/article/2013/11/07/nvidia_geforce_gtx_780_ti_video_card_review/1
http://www.hardocp.com/article/2013/05/23/nvidia_geforce_gtx_780_video_card_review#.VfFDUJdSI8I
http://www.hardocp.com/article/2013...770_lightning_video_card_review/#.VfFDX5dSI8I
http://www.hardocp.com/article/2013/11/04/amd_radeon_r9_290_video_card_review#.VfFDbpdSI8I

I'm curious how many of the same people who are complaining about the testing suite now being a small sample and Nvidia title's heavy can actually cite their comments back in 2013 also taking the same stance?

The problem with a lot of "bias" accusations is that there is tendency for them to be brought out simply due to a disagreement over conclusions.
 
Last edited:
lol the evil undercover nvidian agent.

He just appears to do more harm than good, AMD and especially their PR dept seem to be their own worst enemy at times. The boasting and exaggerated Fury X launch benchmarks really rubbed me the wrong way, I really thought they learned their lesson after the Bulldozer launch but seems they are doubling down on the same failed marketing tactics.
 
If a review site have a small selection of games, most of those games are biased towards a single vendor and that review site makes a general conclusion of a CARDS performance without reservation that the review is based upon vendor biased benchmarks, the conclusion becomes worthless and a joke no matter how well the review is written. In those cases, its no longer a CARD review, but a bunch of game performance reviews bundled together.

Small game selection is a direct result of how [H] review GPU performance: They play games. The reason many review sites use so many games is that a lot of those have their own canned benchmarks.

Many benchmarks is useful only for relative performance comparison (since you are actually comparing the same benchmarks), but is not entirely accurate for actual performance, which is what [H] values more. The reviews prefer absolute numbers over relative ones.

Benchmarks are short and sweet, and is reproducible, but it has a shortcoming that it does not reflect actual gaming performance, because you are not actually gaming.

Both are important to make a concrete decision on GPU. The benchmarks is a very useful indicator for value for money in, but only actual gameplay will tell you if the value is 'good enough'.

Back on topic, I find [H] method of 'warming up' cards before actually recording its data to suit my decision making routine, if you have better cooling than what is reviewed, you can expect better performance due to lower throttling. However if [H] did cold GPU reviews on open benches (open benches is useful for other scientific data, such as recording heat distribution), then they lead to the cards having higher performance than they were actually capable of when operating hot, leading to misleading numbers.

I, for one, prefer to see the worst case numbers than best case numbers.
 
Last edited:
Small game selection is a direct result of how [H] review GPU performance: They play games. The reason many review sites use so many games is that a lot of those have their own canned benchmarks.

Many benchmarks is useful only for relative performance comparison (since you are actually comparing the same benchmarks), but is not entirely accurate for actual performance, which is what [H] values more. The reviews prefer absolute numbers over relative ones.

Benchmarks are short and sweet, and is reproducible, but it has a shortcoming that it does not reflect actual gaming performance, because you are not actually gaming.

Both are important to make a concrete decision on GPU. The benchmarks is a very useful indicator for value for money in, but only actual gameplay will tell you if the value is 'good enough'.

Absolutely. [H] is probably the first place I go to when it comes to reviews, which is why I care about this. Canned benchmarks are worthless and even FPS numbers are worth little to me compared to how [H] does it by writing about their gaming experience. [H] does however have the fewest games in their benchmark suit, since they spend much time and effort on each game. Having more focus on making the benchmark suit as vendor neutral as possible would increase the value of the review if to use the review as a base on how the card performs in general.

Edit: Which btw. is a good reason why we NEED a [H] Nano review.
 
Last edited:
So is the assumption here that the Nano is going to throttle a lot - and only hardocp would have properly tested the card and shown this throttling in practice? Isn't that sort of selling other hardware sites a bit short? I believe Anandtech is getting a card for review - and they haven't always been particularly kind to AMD (R9 290 review caption was "HOT AND LOUD") - but they have been pretty good at doing fairly thorough reviews in the past. I wouldn't assume that every review of the Nano is going to be extremely positive just because AMD didn't send cards to some sites.

We don't like to be called unfair, but everyone is unfair but us...
 
Back
Top