Linus favors Nvidia over AMD?

[Surgeon]

It is said that Nvidia very much cherry picks higher overclockers for reviewers, and AMD having a shortage of cards this time around could not do this, even though I'm sure they normally would as well.  This in itself makes the overclocked benchmarks biased.  Overclocked benchmarks should never be the ones shown first because they're just not reliable.  It's random - except in the case of cherry picking - where it's even more unfair.  I agree that the performance data for overclocking is somewhat important, but for most people, it's irrelevant, and stock performance numbers are far more valuable and what should be shown up front instead of having to dig through the forums for those results.

[Saladin]

It is said that Nvidia very much cherry picks higher overclockers for reviewers, and AMD having a shortage of cards this time around could not do this, even though I'm sure they normally would as well.  This in itself makes the overclocked benchmarks biased.  Overclocked benchmarks should never be the ones shown first because they're just not reliable.  It's random - except in the case of cherry picking - where it's even more unfair.  I agree that the performance data for overclocking is somewhat important, but for most people, it's irrelevant, and stock performance numbers are far more valuable and what should be shown up front instead of having to dig through the forums for those results.

A couple of points. First it is important to look at more than one set of benchmarks for the reason you listed (variation of cards). If you watch Tiny Tom Logan's review he shows you stock speeds and discusses overclock results. Second, the reviews I have seen have all said the the 290x isn't a good overclocker given the high base temps, and poor stock cooler. Third, Nvidia and AMD should be cherry picking their cards for reviewers, its just good marketing, you cant blame Nvidia for sending good cards when AMD doesn't (also we don't know that AMD didn't send a beast card). Finally, if you don't like the way Linus does his review thats fine but that doesn't make him biased, as a consumer it is our responsibility to find out about the products we buy, we should be checking out multiple reviews. For me the overclocked numbers are very helpful as that is how I use my graphics cards. As for 'digging through the forums' how else should he display that information in a way that doesn't make the charts illegibly cluttered, if you have a good suggestion I am sure he would like to know it as he has discussed this problem in the past.

[TheRuke]

It is said that Nvidia very much cherry picks higher overclockers for reviewers, and AMD having a shortage of cards this time around could not do this, even though I'm sure they normally would as well.  This in itself makes the overclocked benchmarks biased.  Overclocked benchmarks should never be the ones shown first because they're just not reliable.  It's random - except in the case of cherry picking - where it's even more unfair.  I agree that the performance data for overclocking is somewhat important, but for most people, it's irrelevant, and stock performance numbers are far more valuable and what should be shown up front instead of having to dig through the forums for those results.

My point exactly. It's funny because the 780 only beat the AMD card by like 7 FPS or less.

[Saladin]

My point exactly. It's funny because the 780 only beat the AMD card by like 7 FPS or less.

No offense but that doesn't really make the 290x look great. GK110 is a year old piece of silicon (was originally planned as the 680). I, as with many, was hoping the 290x crushed the 780 (as it was hyped by AMD to do) that didn't happen, it is a bit of an underwhelming card given that Nvidia has the 780 ti coming out in the next 4-6 weeks (along with and expected price drop of the 780 which would improve the price to performance ratio of the 780). Further a 7 FPS margin indicates that the 290x Linus had was a very good card (or that the 780 he had wasn't) given that in similar test Tiny Tom Logan was seeing a 15 FPS average margin for the OC'd 780 over the 290x.

[TheRuke]

No offense but that doesn't really make the 290x look great. GK110 is a year old piece of silicon (was originally planned as the 680). I, as with many, was hoping the 290x crushed the 780 (as it was hyped by AMD to do) that didn't happen, it is a bit of an underwhelming card given that Nvidia has the 780 ti coming out in the next 4-6 weeks (along with and expected price drop of the 780 which would improve the price to performance ratio of the 780). Further a 7 FPS margin indicates that the 290x Linus had was a very good card (or that the 780 he had wasn't) given that in similar test Tiny Tom Logan was seeing a 15 FPS average margin for the OC'd 780 over the 290x.

Point of Diminishing Return much? "I want 7 moar fps for $100 durrrrrrr". It was planed but it didn't happen. I planed on going to mars but I didn't. The price drop will be $50 at MAXIMUM so the Price:Performance will still be lower than the 290X. ti is a joke, if it's any good it'll just replace the Titan spec wise. 15 FPS? dang. What was he testing on? Games that are "The Way It's Meant to Be Played" or "Gaming Evolved" titles? If they were benchmark software then that's a good indicator but does not compare to real-world testing, the practical testing.

[Leonard]

It has come to my attention that linus favors nvidia over amd for the past 1.5 year.Before that time i think linus was more objective.i can listen to your opinions.

Linus's loyalty seems to be to his son and his wife, as they should be. To me he tell it as he feels it at that moment. If Nvidia is doing something he likes he will say Nvidia is doing great with this and if AMD is doing some thing good he will say AMD did great with this.  

[Saladin]

Point of Diminishing Return much? "I want 7 moar fps for $100 durrrrrrr". It was planed but it didn't happen. I planed on going to mars but I didn't. The price drop will be $50 at MAXIMUM so the Price:Performance will still be lower than the 290X. ti is a joke, if it's any good it'll just replace the Titan spec wise. 15 FPS? dang. What was he testing on? Games that are "The Way It's Meant to Be Played" or "Gaming Evolved" titles? If they were benchmark software then that's a good indicator but does not compare to real-world testing, the practical testing.

You can watch either of the TWO videos he did on the 290x launch, he benched multiple cards (AMD and Nvidia) on multiple synthetics and multiple games at various resolutions. He also has a full write up (like 9 pages, including all graphs of the benchmarks) on the 290x at OC3D. My point is this, the 290x is a pretty good card, but its not great and we the consumer wanted/needed it to be great and crush Nvidia, this didn't happen. Because the 290x is only comparable instead of the clear winner, Nvidia has no incentive to innovate, they can just release incremental improvements like the 780 ti. I don't think that Linus is/was biased, it would be stupid, bad for business. I continue to think that overclocked results are valid given that that is how I run my cards. I had been planning to upgrade to a 290x if it was the beast card it was hyped to be, given that it isn't I will hold off until the 780 ti and possibly until the next generation of card come out sometime next year.

[eshbop]

No offense but that doesn't really make the 290x look great. GK110 is a year old piece of silicon (was originally planned as the 680). I, as with many, was hoping the 290x crushed the 780 (as it was hyped by AMD to do) that didn't happen, it is a bit of an underwhelming card given that Nvidia has the 780 ti coming out in the next 4-6 weeks (along with and expected price drop of the 780 which would improve the price to performance ratio of the 780). Further a 7 FPS margin indicates that the 290x Linus had was a very good card (or that the 780 he had wasn't) given that in similar test Tiny Tom Logan was seeing a 15 FPS average margin for the OC'd 780 over the 290x.

If you've actually looked at multiple reviews you'd know that the 290x and 780 are on par (if not the 290x has the edge) with each other. These are the quotes from the conclusions of the first few articles you get when Googling 290X reviews, I quoted every article I opened.

 

PCPer - "AMD was able to put enough horsepower into the R9 290X to push past both the GeForce GTX 780 and GTX Titan graphics cards in enough of our game tests to declare it the fastest single GPU card for PC gaming. "

AnandTech - "Consequently against NVIDIA’s pricing structure the 290X is by every definition a steal at $549. Even if it were merely equal to the GTX 780 it would still be $100 cheaper, but instead it’s both faster and cheaper"

Techspot - "Nvidia's offering clearly won in Far Cry 3 and Tomb Raider, while AMD's claimed an easy victory in Crysis 3, Sleeping Dogs and Metro: Last Light. But the two heavyweights were separated by less than 5% in Battlefield 3, Dirt 3, Max Payne 3, Medal of Honor and Hitman: Absolution. What's particularly impressive about this comparison is the R9 290X's shockingly low price of $550. That's 15% cheaper than the GTX 780 despite being 10% faster, "

Guru3D - "The product is often as fast or faster then a GeForce GTX Titan, whilst it is going to be priced a good chunk below the GeForce GTX 780. "

Techreport - "The Radeon R9 290X is a bit faster than the GeForce GTX 780 and costs a hundred bucks less. Beats the Titan for nearly half the price, too."

Techpowerup - "Compared to NVIDIA's lineup, the R9 290X in "Quiet" mode is slightly faster than the GTX 780 and 5% slower than the GTX Titan. With the "Uber" BIOS, the card ends up a bit faster than the GTX Titan. At higher resolutions, like 2560x1600, we see the card outperform NVIDIA's offerings by another few percent."

Hexus - "But, let's be fair, AMD is really back in the high-end graphics game with the R9 290X's arrival. It's at least as good as Nvidia's best cards"

Bit-tech - "The most important thing to say about the new Radeon R9 290X 4GB is that on the price performance scale it's a clear winner against Nvidia's GTX 780 3GB, at the time of writing. It undercuts it by at least £50 in the UK and $100 stateside, yet outperforms it almost universally"

[Saladin]

If you've actually looked at multiple reviews you'd know that the 290x and 780 are on par (if not the 290x has the edge) with each other. These are the quotes from the conclusions of the first few articles you get when Googling 290X reviews, I quoted every article I opened.

I'm not disagreeing that STOCK the 290x is slightly faster, just that that isn't that impressive given the age of the GK110 chip (remember it was originally supposed to be the 680), the overclocking potential of both the 290x and the 780, the hype surrounding the 290x, and the fact that AMD needs a big win. The 290x is a good card, but as I have been saying it isn't a great card, it isn't mind bogglingly powerful or crazy efficient, its just on par with the 780. This doesn't drive the market and force innovation, all it does is ensure we continue with small incremental improvements. As to the original topic of bias I continue to point out that a bias would hurt Linus not help him (any upfront payoff would be detrimental in the long run) and that there is no actual evidence of bias other that people saying he has a bias (this isn't proof, just an Ad Hominem attack).

[DeViLzzz]

Gotta love that physX ...

Nvidia all the way

Linus's preference is the right preference.

[eshbop]

I'm not disagreeing that STOCK the 290x is slightly faster, just that that isn't that impressive given the age of the GK110 chip (remember it was originally supposed to be the 680), the overclocking potential of both the 290x and the 780, the hype surrounding the 290x, and the fact that AMD needs a big win. The 290x is a good card, but as I have been saying it isn't a great card, it isn't mind bogglingly powerful or crazy efficient, its just on par with the 780. This doesn't drive the market and force innovation, all it does is ensure we continue with small incremental improvements. As to the original topic of bias I continue to point out that a bias would hurt Linus not help him (any upfront payoff would be detrimental in the long run) and that there is no actual evidence of bias other that people saying he has a bias (this isn't proof, just an Ad Hominem attack).

I was just pointing out that what TTL and LTT result show aren't supported by a lot of other reviews. I would not put this down to either of them being bias, because I know neither of them are.

 

This is the performance AMD were aiming for, which is how it's been for the past few generations from both sides, equal tier products always perform very close, the 600 series did this, the 6000 series did it, etc. I disagree that GK110 could of ever been or was meant to be the 680, but assuming that it was this is why Nvidia released GK104 as their top tier card, because it performed roughly the same as Tahiti XT. They don't aim to be mindbloggingly more powerful, but slightly more powerful or on par at a lower price.

 

There is nothing more to be had from GPUs using the 28nm process and they know Nvidia will not release a new product until 20nm is ready and that's when we will see a big jump in performance from both sides. There's no more innovation to be had on 28nm. That's why we have seen each side release a refresh. They are waiting for TSMC and GF to innovate. If AMD really wanted to they could of aimed for a Titan sized GPU (this chip is 20% smaller than GK110), it would of been faster, but would of been closer to Titan priced, but that is senseless, especially because they don't have the luxury of a contract for the GPUs development for it's place in a supercomputer (re. Titan) and the ability to sell off chips that don't make the cut. The fact that GK110 is older is irrelevant as it's only been introduced into the mainstream line with the release of the 780. AMD have released an equal performing product at lower cost, and at a much lower cost than when GK110 first became a consumer product (though I'd argue that a $1000 GPU with it's additional compute power isn't really a consumer product).

[LinusTech]

*yawn*

 

https://www.youtube.com/playlist?list=PL8mG-RkN2uTzFS_ljRvTdL9rF6aAWf_Dx

 

AMD wins almost every one of these, and we shower them with praise.

 

The best part is in the comments you can read the complaints about how we overclocked the cards and the AMD ones come out on top. You can read the complaints about how we "only test Gaming Evolved titles", and how we're obviously paid AMD shills.

 

Overclocking is free. We've done guides on it. We overclock our cards. Don't like it? Go read any one of LITERALLY 100 other reviews. Overclocked settings is OUR unique spin, and it's actually a lot more work for us. We wanted to do something a little bit different from everyone because otherwise we could save ourselves a lot of time by just reading someone else's review and talking about their numbers.

 

*yawn*

[LinusTech]

On a personal level both Linus & Slick prefer Nvidia, but they remain very fair in their testing & do not let their personal opinions affect their objectivity.

I asked Linus about the last AMD/ATI card he used in his personal system, this is when he received the Nvidia sponsorship, & he told me that it was the 9700 Pro some 10-11 years ago.

 

Actually the last time I used an ATI card in my personal system was a 3870, and my wife's system runs a 5870.

[Lays]

@LinusTech

 

I think people will just see whatever they want to see, there are moments where you lean towards certain parts, but then you always release other videos, or reasons behind why you lean towards certain products, with complete explanations of why.

 

I've always loved that about your vids, you've always gone the extra mile to say why you recommend things over other things. 

 

Your videos are what helped me build my first PC, and I've continued to recommend them to others to help as-well. 

[FellTheSky]

so what. The 780 is just a better card 

 

yes, you pay 100$+ but look at the build quality, look at the features, for me its worth the 100$ hands down (also the msi gaming version overclocks like a beast!)

 

You can go with a r290x, but i think it would have been better for amd to release it together with mantle. I know  a lot of people that were waiting, and saw the numbers/build quality and just bought the 780...

 

On topic, linus can have whatever he wants to have, but his numbers remains consistent with the rest of the reviewers. Thats objective review for me, not like some tek syndicate guys that pulled some numbers out of their asses and where like here fanboys, give us subscribers. I love the tek, but i do not trust their reviews

[TechFan@ic]

Actually the last time I used an ATI card in my personal system was a 3870, and my wife's system runs a 5870.

I already know your wife still runs a 5870, that's your wife however not you & I believe I've already told you that.

Did you actually use the 3870 for gaming, or was it just there as a temporary solution ?

I bet you would have had one of the 8800s in your actual gaming system by then.

I know you had an HD 2800 XT in your office system at NCIX for a while too.

 

[eshbop]

Overclocking is free. We've done guides on it. We overclock our cards. Don't like it? Go read any one of LITERALLY 100 other reviews. Overclocked settings is OUR unique spin, and it's actually a lot more work for us. We wanted to do something a little bit different from everyone because otherwise we could save ourselves a lot of time by just reading someone else's review and talking about their numbers.

The way you guys do it is fine, overclocked benchmarks give me the information I care about. Though do you guys know whether the GPU has been hand-picked or hasn't or if it overclocks well or poorly, especially at launch when I'd assume no reviewers know what a typical or average attainable overclock is?

[LinusTech]

I already know your wife still runs a 5870, that's your wife however not you & I believe I've already told you that.

Did you actually use the 3870 for gaming, or was it just there as a temporary solution ?

I bet you would have had one of the 8800s in your actual gaming system by then.

I know you had an HD 2800 XT in your office system at NCIX for a while too.

 

 

I picked up a 3870 the last time I tried to downsize my rig and go SFF. It was my primary gaming machine for a while in a Silverstone SG01 and the reason I went for that card was it was a very economical card to liquid cool due to the integrated copper cooler that covered the RAM and (iirc) VRM.

 

I shortly bailed on the whole thing and went back to a full size tower with GTX 260 SLI.

 

I buy/use/recommend whatever I think will work best. I'm not always right, but if I didn't have an opinion at all then there wouldn't be much point listening to anything I say anyway.

[TechFan@ic]

I picked up a 3870 the last time I tried to downsize my rig and go SFF. It was my primary gaming machine for a while in a Silverstone SG01 and the reason I went for that card was it was a very economical card to liquid cool due to the integrated copper cooler that covered the RAM and (iirc) VRM.

 

I shortly bailed on the whole thing and went back to a full size tower with GTX 260 SLI.

 

I buy/use/recommend whatever I think will work best. I'm not always right, but if I didn't have an opinion at all then there wouldn't be much point listening to anything I say anyway.

Well yeah, like the mCubed T-Balancer ?

[Crion]

This is a ridiculous comment and goes along the same lines as calling TTL a Nvidia fanboi because he said the 780 was the better card (his Orca has dual 7970s).  If the numbers say one card is better than the other, then saying it like it is cannot be called favoritism.  Favoritism would be saying one card is better regardless of what the data told you. 

 

The enthusiast community wants to see great cards come from both AMD and Nvidia.  Why?  It pushes the industry as a whole. 

[Zeo2002]

This is a ridiculous comment and goes along the same lines as calling TTL a Nvidia fanboi because he said the 780 was the better card (his Orca has dual 7970s).  If the numbers say one card is better than the other, then saying it like it is cannot be called favoritism.  Favoritism would be saying one card is better regardless of what the data told you. 

 

The enthusiast community wants to see great cards come from both AMD and Nvidia.  Why?  It pushes the industry as a whole. 

I would say favoritism, like statistics, is to keep changing the tests until the numbers say what you want them to say.

 

The question, though I don't know the answer and highly doubt that Linus does this, is if we unfairly skew the test to one side or another. If we max out the cards OC abilities as much as possible (to a standard, such as stable under Prime 95 testing for 24 hours) and anymore causes instability or is capped, than that would show max OC for that card. If you're going to OC only X percentage on one card and x percentage on another (or even the same percentage across both cards) that does not necessarily show equality.

[TheMissxu]

Here's my take on the situation:

 

You didn't overclock the 290x because the cooler was bad, but you still overclocked the other cards. I think that gives an unfair advantage to the other cards.

That phrase "because the cooler was bad" means you are essentially comparing the two cards coolers. This is why LTT and TTL both recommended the 780 over the 290x.

 

As we've seen with other reviewers, the 290x holds it's own against other cards (when they aren't overclocked). I know your philosophy is overclocking equals free performance, but when a cooler on a given card is poor, you shouldn't overclock the other competing cards. When you do and recommend the other card, you are basically recommending a $100 cooler so you get a better overclock. This makes perfect sense, but...

 

What bothers me is that there are $100 aftermarket coolers you could have applied to the 290x to show its true unbottlenecked (by temps.) overclocking performance. I'm curious how a DIY cooler would effect the 290x overclocking.

 

Another case would be a custom watercooled CPU. While this is a very unique case, if both the NVidia and AMD had a water cooling loop installed for the CPU, adding block to the AMD card would make it equal in price to the 780 (roughly, actually $20 more). Then there wouldn't be that bottleneck. In fact, for the same priced system, you could compare the 780 and 290x evenly. The benchmark numbers are yet to be determined.

 

It's unfortunate that AMD have released the R9 290x with such a poor cooler. The average user won't be overclocking (arguably) as the average user wont be using a DIY cooler (a little less arguable), thus, shouldn't you recommend the 290x over the 780 (remember, from stock vs stock run by the majority of reviewers, the 780 falls short of the 290x while having a difference of $100)? If you argue the 290x shouldn't be recommended because of the heat, show me a $100 aftermarket cooler wont give a better experience over the competitors.

 

My two cents.

 

PS, with higher res monitors becoming more cheap, maybe you could suck up the heat and use the $100 to upgrade your monitor.

[Saladin]

Here's my take on the situation:

 

You didn't overclock the 290x because the cooler was bad, but you still overclocked the other cards. I think that gives an unfair advantage to the other cards.

That phrase "because the cooler was bad" means you are essentially comparing the two cards coolers. This is why LTT and TTL both recommended the 780 over the 290x.

 

As we've seen with other reviewers, the 290x holds it's own against other cards (when they aren't overclocked). I know your philosophy is overclocking equals free performance, but when a cooler on a given card is poor, you shouldn't overclock the other competing cards. When you do and recommend the other card, you are basically recommending a $100 cooler so you get a better overclock. This makes perfect sense, but...

 

What bothers me is that there are $100 aftermarket coolers you could have applied to the 290x to show its true unbottlenecked (by temps.) overclocking performance. I'm curious how a DIY cooler would effect the 290x overclocking.

 

Another case would be a custom watercooled CPU. While this is a very unique case, if both the NVidia and AMD had a water cooling loop installed for the CPU, adding block to the AMD card would make it equal in price to the 780 (roughly, actually $20 more). Then there wouldn't be that bottleneck. In fact, for the same priced system, you could compare the 780 and 290x evenly. The benchmark numbers are yet to be determined.

 

It's unfortunate that AMD have released the R9 290x with such a poor cooler. The average user won't be overclocking (arguably) as the average user wont be using a DIY cooler (a little less arguable), thus, shouldn't you recommend the 290x over the 780 (remember, from stock vs stock run by the majority of reviewers, the 780 falls short of the 290x while having a difference of $100)? If you argue the 290x shouldn't be recommended because of the heat, show me a $100 aftermarket cooler wont give a better experience over the competitors.

 

My two cents.

 

PS, with higher res monitors becoming more cheap, maybe you could suck up the heat and use the $100 to upgrade your monitor.

A couple of points if I may. First as to your point that the 290x cooler is bad therefore other cards shouldn't be overclocked, this is nonsensical, just reread this sentence. The cooler is part of the card, comparing the ability of the coolers to control temperatures and allows overclocking is valid. Even when you only compare stock coolers the Nvidia cards cool better. In my opinion, given that AMD is doing a reference only launch, the 290x should of had a better cooler. Second There are currently no non-reference cards, therefore no non-reference coolers other than aftermarket, at the time of filming there were no water blocks available for the 290x (Tiny Tom Logan has said he will bench them underwater as did Linus on the WAN show tonight). I am very excited to see benchmarks of the 290x underwater compared to a 780 and 780 ti underwater, but that needs to wait. Third I vehemently disagree that the average user wont overclock this card, truth is that the 'average user' wont even BUY this card. The target audience are enthusiasts, a large portion/majority of users that are spending the money on this kind of card probably will overclock their GPU, they want the performance, that is why they are spending so much money in the first place. Finally If you buy an aftermarket cooler for $100 you are now at 780 prices, at this point the price to performance decreases (as you have added $100 to the price), and you would at the very least be better off waiting to see what happens with the 780 ti in a few weeks (AMD/Nvidia price drops?).

[Lukiose]

I would say Linus favours Nvidia while Slick is relatively Neutral, you can kind of detect the tone but nonetheless, they give honest benchmarks & reviews and that is really all that matters. When the reference 770 came out Linus didn't hold back from saying it is bad either, he just didn't bash it blatantly because he has sponsors. When AMD cards perform well they don't say that they suck either. That is professionalism.

[TheMissxu]

@Saladin

 

Hmm, I see the R9 290x day one review as a review of the R9 290x card line. Not the actual reference card. So, not overclocking it wouldn't do it justice. We don't see eye-to-eye on this, and this seems to be the underlying issue then.

edit: don't forget, Linus is a busy man. He cannot do reviews of every non-reference card.

Your second point does make sense.

The waiting game is always to be played isn't it? Why stop waiting at the 780Ti?

 

Please don't forget the opinion of other reviewers. (Cheesy title, it wasn't me. Also, most of them did the stock vs stock comparison, but we know they are smart enough to look past that. I think.)

Edited October 26, 2013 by TheMissxu