AdoredTV Dissects the Marketing to Make Nvidia Turing RTX Performance Predictions

[cageymaru]

AdoredTV has a new video where he explains how the marketing from Nvidia works and why it is meant to obfuscate the true performance potential of the upcoming Nvidia Turing RTX lineup. By analyzing previous performance metrics released by other reviewers on the internet, comparing hardware specifications on Nvidia's own website and digging deep into the testing methodology carried out by Nvidia; AdoredTV cuts through the marketing to generate his performance prediction for the upcoming release. He thinks they are fast; very fast indeed, but not quite as fast as the marketing would like you to think.

 

[SickBeast]

Isn't this the same person who thought Turing would only be 15-20% faster than Pascal, even after nVidia released benchmarks showing otherwise? And we are supposed to believe them now why exactly?

 

[cageymaru]

Isn't this the same person who thought Turing would only be 15-20% faster than Pascal, even after nVidia released benchmarks showing otherwise? And we are supposed to believe them now why exactly?

The video was interesting and he made some good points. Just another performance prediction which are always fun to analyze. I don't believe marketing slides from AMD, Nvidia, Intel or anyone else. I still find it fun to read what others think.

I'm a real world benchmarks only type of person.

 

[SickBeast]

The video was interesting and he made some good points. Just another performance prediction which are always fun to analyze. I don't believe marketing slides from AMD, Nvidia, Intel or anyone else. I still find it fun to read what others think.

I'm a real world benchmarks only type of person.

The thing is, he said not long ago that he thought Turing was 15-20% faster. Now he is saying something else. So which is it? What are we supposed to believe? And why are we trusting his speculation as opposed to nVidia's benchmarks, which tend to be accurate?

The last time I checked, AdoredTV was an extremely questionable source. I refuse to watch their videos or support them in any way.

 

[cageymaru]

The thing is, he said not long ago that he thought Turing was 15-20% faster. Now he is saying something else. So which is it? What are we supposed to believe? And why are we trusting his speculation as opposed to nVidia's benchmarks, which tend to be accurate?

The last time I checked, AdoredTV was an extremely questionable source. I refuse to watch their videos or support them in any way.

Oh. Well then his videos aren't for you.

 

[alamox]

Isn't this the same person who thought Turing would only be 15-20% faster than Pascal, even after nVidia released benchmarks showing otherwise? And we are supposed to believe them now why exactly?

his analysis is on point and makes sense, you really think 2080 is going to be 35% to 60% faster than 1080 ? and you switch a setting and you get over 100% n some games ?.... come on
the 2080 would be around 1080Ti 5% give or take, that's the best scenario imo, ppl eyeballing the 50-100% are delusional.
great video as usual, the guy speaks facts, no emotional BS... only when he talks about AMD, he do like them for the same reasons i like them...company politics, but he tries to stay unbias.

 

[Compddd]

Video reviews/articles should die.

 

[ng4ever]

Video reviews/articles should die.

Agreed.

 

[DNMock]

The thing is, he said not long ago that he thought Turing was 15-20% faster. Now he is saying something else. So which is it? What are we supposed to believe? And why are we trusting his speculation as opposed to nVidia's benchmarks, which tend to be accurate?

The last time I checked, AdoredTV was an extremely questionable source. I refuse to watch their videos or support them in any way.

Don't know anything about AdoredTV, but adjusting your opinion/predictions to reflect new information is exactly how it's supposed to be and far more honest than simply dismissing or cherry picking your information so you don't have to change your position.

You should absolutely trust the guy who changes his mind when new data is brought to light over the guy who never changes his mind when new data is brought to light.

 

[alamox]

Video reviews/articles should die.

he doesn't do reviews, it's mostly analysis videos

 

[Derangel]

Isn't this the same person who thought Turing would only be 15-20% faster than Pascal, even after nVidia released benchmarks showing otherwise? And we are supposed to believe them now why exactly?

He estimated based on the specs Nvidia listed for the cards. Taking marketing slides at face value, especially when no information is provided, is stupid. They may be correct, or it may be incredibly cherry picked examples. We know they used some form of AA, but not which type or how much. They also only gave the comparison for one of the three cards. You are making some pretty damn big assumptions based on worthless info.

 

[SickBeast]

He estimated based on the specs Nvidia listed for the cards. Taking marketing slides at face value, especially when no information is provided, is stupid. They may be correct, or it may be incredibly cherry picked examples. We know they used some form of AA, but not which type or how much. They also only gave the comparison for one of the three cards. You are making some pretty damn big assumptions based on worthless info.

Did he take into consideration the fact that the new CUDA cores on the Turing cards are *much* more powerful?

 

[gxp500]

...

 

[Derangel]

Did he take into consideration the fact that the new CUDA cores on the Turing cards are *much* more powerful?

There is really no solid data on exactly how much stronger each core is. As I recall, the information he got for his RTX leak video did put the 2080 at closer to 50% more than the 1080. Can't remember the exact numbers.

Here's the thing; Unless you're an Nvidia engineer none of use know a thing about what was done to the CUDA cores or the exact per-core performance differences compared to Pascal. Until reviews are out everyone is just guessing. Personally, I'd prefer to expect the lower estimates and be surprised if performance is closer to Nvidia's marketing crap instead of expecting the marketing crap to be 100% accurate and be disappointed if it isn't.

 

[SickBeast]

There is really no solid data on exactly how much stronger each core is. As I recall, the information he got for his RTX leak video did put the 2080 at closer to 50% more than the 1080. Can't remember the exact numbers.

Here's the thing; Unless you're an Nvidia engineer none of use know a thing about what was done to the CUDA cores or the exact per-core performance differences compared to Pascal. Until reviews are out everyone is just guessing. Personally, I'd prefer to expect the lower estimates and be surprised if performance is closer to Nvidia's marketing crap instead of expecting the marketing crap to be 100% accurate and be disappointed if it isn't.

Can you just answer my question, please. Did he take into account the fact that the new CUDA cores are more powerful, or did he not? It's yes or no.

 

[tremortj]

Can you just answer my question, please. Did he take into account the fact that the new CUDA cores are more powerful, or did he not? It's yes or no.

I think I'll just wait for actual benchmarks to prove it one way or another.

 

[Californian]

Can you just answer my question, please. Did he take into account the fact that the new CUDA cores are more powerful, or did he not? It's yes or no.

For this video it is mentioned in passing. His video before this looking at Turing did talk about it. His opinion seems to be that per game (per engine?) optimizations would be needed to be able to take advantage of the separate INT and FP units.

In this video the discussion of it was simply that the chart Nvidia showed of performance improvement did not state where it all came from. We don't know how much of the imporvements in Nvidia's chart are due to the separating of INT+FP units, and how many are related to things like better DX12 (Async?) implementation, improved caches.

As for how he reached his conclusion
1. Used Nvidia's chart they released
2. Looked at how you could relatively reduce the 1080's performance to this card
a. HDR ~10% performance hit on 1080
b. 4K the 1080 gets bandwidth bound
c. using a tech demo
d. use a game which utilizes FP16 (Turing brings this, Pascal does not)
3. looked at the amount of performance these things could explain.

 

[Derangel]

Can you just answer my question, please. Did he take into account the fact that the new CUDA cores are more powerful, or did he not? It's yes or no.

I think my first sentence answered that. How can anyone take that into account with solid information on how much faster they are? Are secretly an Nvidia engineer with accurate access to that kind of info? If not your estimate is just as pointless as anyone else's.

 

[Dayaks]

“It looks to be an architectural masterpiece.”

I agree! God damn impressive what they have done.

I liked that video. Not sure if this guy is super strong technically but I like how he compiles everything.

 

[Gideon]

I believe his estimate for performance will be pretty close to how it turns out, I tend to think 20% or less based on what I have seen so far. Also the fact that Nvidia wants to talk about features and the ray tracing and not talk about performance is telling to me.

 

[Stimpy88]

The thing is, he said not long ago that he thought Turing was 15-20% faster. Now he is saying something else. So which is it? What are we supposed to believe? And why are we trusting his speculation as opposed to nVidia's benchmarks, which tend to be accurate?

The last time I checked, AdoredTV was an extremely questionable source. I refuse to watch their videos or support them in any way.

Some people can’t see the wood for the trees...

AdoredTV has been right about these cards all along, when nobody else in the industry was looking because they don’t want to loose out on shiny new grad cards and marketing money.

 

[SickBeast]

I think my first sentence answered that. How can anyone take that into account with solid information on how much faster they are? Are secretly an Nvidia engineer with accurate access to that kind of info? If not your estimate is just as pointless as anyone else's.

I'm not making an estimate. AdoredTV is making one, and it's BS. nVidia has released benchmarks. And if AdoredTV isn't taking into consideration the fact that the new CUDA cores in Turing are much more powerful, then they are biased as hell, just as I thought.

 

[SickBeast]

Some people can’t see the wood for the trees...

AdoredTV has been right about these cards all along, when nobody else in the industry was looking because they don’t want to loose out on shiny new grad cards and marketing money.

So were they right when they said Turing is 15-20% faster than Pascal?

 

[Derangel]

I'm not making an estimate. AdoredTV is making one, and it's BS. nVidia has released benchmarks. And if AdoredTV isn't taking into consideration the fact that the new CUDA cores in Turing are much more powerful, then they are biased as hell, just as I thought.

Just started watching the video this post is talking about: Adored mentions that he didn't see Nvidia's charts before making the previous video. Which means he had no data on any potential improvements when he made his previous estimate. This new video is going over those charts and talking about the charts.

 

[reaper12]

So were they right when they said Turing is 15-20% faster than Pascal?

How we know? You don't know. He has been on the money about everything so far with Turing. Does that mean he will be right about the performance in the end? I don't know. I will take his performance estimations as speculation, just like I take Nvidia's marketing slides with a massive pinch of salt, the slides you seem to be taking as gospel.

 

[SickBeast]

How we know? You don't know. He has been on the money about everything so far with Turing. Does that mean he will be right about the performance in the end? I don't know. I will take his performance estimations as speculation, just like I take Nvidia's marketing slides with a massive pinch of salt, the slides you seem to be taking as gospel.

Call me crazy I guess for believing benchmarks from a company that has never lied about them instead of someone who is flat out pro-AMD biased with a horrible track record. We might as well link to nVidia viral marketing propaganda at [H] if we are going to link to AdoredTV in the "news" section. This is fake news.

 

[wizzi01]

So were they right when they said Turing is 15-20% faster than Pascal?

Why are you arguing about something that no one has facts for? His video is his opinions and predictions made with the very little information nvidia has released.

 

[Mav451]

AdoredTV is the Stephen A Smith of PC hardware, to put it lightly. More entertainment than analysis.

 

[Derangel]

Call me crazy I guess for believing benchmarks from a company that has never lied about them instead of someone who is flat out pro-AMD biased with a horrible track record. We might as well link to nVidia viral marketing propaganda at [H] if we are going to link to AdoredTV in the "news" section. This is fake news.

The only videos of his I've watched at the RTX stuff, but nothing remotely seems anti-Nvidia or pro-AMD in them. Breaking down a leak, reacting to a conference, and doing a little analysis of the marketing slides. If he was as biased as you claim he'd say Nvidia was lying or claim that AMD would beat them, or whatever. Instead, everything I've seen from the three videos has been completely fair and reasonable. If he's wrong about performance, big deal. Prior to reviews, no one should take ANY numbers too seriously. People discussing the card are making estimates and Nvidia is cherry picking, neither are exactly reliable.

 

[SickBeast]

The only videos of his I've watched at the RTX stuff, but nothing remotely seems anti-Nvidia or pro-AMD in them. Breaking down a leak, reacting to a conference, and doing a little analysis of the marketing slides. If he was as biased as you claim he'd say Nvidia was lying or claim that AMD would beat them, or whatever. Instead, everything I've seen from the three videos has been completely fair and reasonable. If he's wrong about performance, big deal. Prior to reviews, no one should take ANY numbers too seriously. People discussing the card are making estimates and Nvidia is cherry picking, neither are exactly reliable.

He is downplaying Turing. At this point, it is all AMD has left. Try to rain on nVidia's parade. I'm not buying it.

 

[Azphira]

Not just this video, but it all sounds just like the Windows 10 whine, or the whine if someone asks about raid 0, Or the whine when a bundle site that sells most things for $1-3 sells a discography for $100. Or the whine if cdkeys has it on sale.

 

[Derangel]

He is downplaying Turing. At this point, it is all AMD has left. Try to rain on nVidia's parade. I'm not buying it.

I fail to see how examining benchmarks in relation to the, limited, information Nvidia provided on their own charts is downplaying anything. Taking information at face value is incredibly stupid.

 

[Dayaks]

He’s pretty shitty at being biased if in his closing thoughts he says about Turing: “It looks to be an architectural masterpiece.”

Now the overall tone isn’t as optimisitic as I am but he’s somewhat near center IMO.

 

[alamox]

He is downplaying Turing. At this point, it is all AMD has left. Try to rain on nVidia's parade. I'm not buying it.

this is not about amd vs nvidia, it's simply discussing nvidia.
the slide is really deceitful and some ppl really take the slide at face value, then buy the product based on that, because the slides them selves put a very specific scenario that does not apply to even 1% of gamers, but when you put it to the other 99% you suddenly go from 2x to 1.2x , don't you think that's something worth discussing and analyzing ? or does it always have to be AMD vs Nvdia ? AMD has no products coming to desktop for a very long time, and even if they bring a gpu in 2019 it would be a pro card adapted to desktop, meaning still won't be good enough to compete with a Nvidia gpu designed for desktop.

 

[HeadRusch]

"We expect it to be fast......but not as fast as people want it to be".

This guy gets paid?

 

[1Nocturnal101]

Isn't this the same person who thought Turing would only be 15-20% faster than Pascal, even after nVidia released benchmarks showing otherwise? And we are supposed to believe them now why exactly?

He said he lowballed the numbers a bit, but he does like what they did, just not the marketing portion of the bs. He isn't anti NVidia he is just anti NVidia marketing, he does state he loves what they were able to pull off.

 

[1Nocturnal101]

He is downplaying Turing. At this point, it is all AMD has left. Try to rain on nVidia's parade. I'm not buying it.

In his previous video he trashed AMD, Either way even with NVidias chart, I'd seriously wait for a dual reviewers to do benchmarks, anything RT related I wouldn't even bother taking for a grain of salt, until benchmarks and games are built from the ground up using the tech and we have a fair samplebto compare averages and not some games with the tech just tacked on, it's not worth losing your minds over, we have 20-30days to wait, afterwards go post nuclear if deemed, until then everyone needs to chill.

 

[reaper12]

Call me crazy I guess for believing benchmarks from a company that has never lied about them instead of someone who is flat out pro-AMD biased with a horrible track record. We might as well link to nVidia viral marketing propaganda at [H] if we are going to link to AdoredTV in the "news" section. This is fake news.

ok, you are crazy. Nvidia has never lied about them? hahahaha. That's the funniest statement that I have read in a long time. Never believe marketing slides from any company. And I really think you listen to too many Nvidia fanboys. AdoredTV has a pretty good track record, but, just like any other video reviewer, it's just speculation, so they are all going to be wrong from time to time. IF you are going to call AdoredTV crap, you have to call them all crap. And if you watched the videos you would see that he actually likes what Nvidia has done. It's mainly positive.

 

[SickBeast]

his analysis is on point and makes sense, you really think 2080 is going to be 35% to 60% faster than 1080 ? and you switch a setting and you get over 100% n some games ?.... come on
the 2080 would be around 1080Ti 5% give or take, that's the best scenario imo, ppl eyeballing the 50-100% are delusional.
great video as usual, the guy speaks facts, no emotional BS... only when he talks about AMD, he do like them for the same reasons i like them...company politics, but he tries to stay unbias.

I can't wait to quote this post once the reviews come out.

 

[SickBeast]

ok, you are crazy. Nvidia has never lied about them? hahahaha. That's the funniest statement that I have read in a long time. Never believe marketing slides from any company. And I really think you listen to too many Nvidia fanboys. AdoredTV has a pretty good track record, but, just like any other video reviewer, it's just speculation, so they are all going to be wrong from time to time. IF you are going to call AdoredTV crap, you have to call them all crap. And if you watched the videos you would see that he actually likes what Nvidia has done. It's mainly positive.

Did you watch the Pascal reveal? They said that the GTX 1060 would be as fast as a GTX 980, and that a GTX 1070 would be as fast as a GTX 980 Ti. Both true statements. nVidia does not lie during these reveals. They have no need to.