AMD Radeon R9 Nano Coming Sooner Than You Think?

[EarthDog]

My point exactly Carl. Hence why I feel the Nano, if they said it would beat a 290x, will also beat a 390x or at worst, tie it since they are factory overclocked. It all depends on what they based 'significant' performance increase off of when AMD said that... so who knows.

But since there is a pretty beefy gap (2560x1440) between the 290x/390x and FuryX, that is where I see it ending up. There is a 21% gap between the 290x and Fury and a 13% gap between the 390x and Fury... Right THERE is where it fits in.
http://www.techpowerup.com/reviews/AMD/R9_Fury_X/31.html

Wait...so because a 390x and 290x are not the same cards because one is 10% faster, that means:

• an overclocked 390x really is a 490x because it is about 10% faster
• an overclocked 290x is really a 390x because they have about the same performance
• an overclocked 290x 8GB cards is ???

OMG I am so confused now....

LOLOLOLOL!

 

[Casecutter]

I don't think 390X was out before June 16 so I think they used 290X because it was a known quantity. It'll be interesting to see where Nano lands relative to 390X.

The fact that they said it outperforms a 290x and not a 390x gives us a great hint that the Nano will fall between the 290x and 390x in performance

I'd concur with FordGT90Concept, AMD was not going to compare a next-generation card to a card that they had yet released (390/390X) so all they could/would compare it to was a 290X.

That said, I've been search for the "direct quote" of AMD saying Nano is "significantly faster”, and I'm not finding multiple confirmations or the transcripts of E3 2015 where it is attributed? Every time I search I find such claims come back to either:

Brad Chacos Senior Editor, PCWorld; CEO Lisa Su has said that the six-inch card will offer “significantly more performance than the Radeon R9 290X,”
http://www.pcworld.com/article/2949...ul-radeon-r9-nano-is-launching-in-august.html

Or, Ryan Shrout at PCPerspective; On stage at the AMD E3 2015 press conference, AMD's CEO Lisa Su announced the Radeon R9 Nano, a 6-in PCB small form factor graphics card that will feature "2x the performance per watt of the R9 290X" as well as "significantly" more performance than the R9 290X.
http://www.pcper.com/news/Graphics-Cards/AMD-Announces-Radeon-R9-Nano-6-Graphics-Card

I find it odd that Ryan used the word "significantly" in his passage he "air-quotes" after the information we've known AMD offered. While neither are corresponding quotes?

All I really find actually stated/printed is, "AMD states that the card is faster than the R9 290X and has 2x the performance per watt of the R9 290X."

I would think there would be transcripts or video of that, I'd like to see how that factully transpired... anyone?

 

[EarthDog]

hat said, I've been search for the "direct quote" of AMD saying Nano is "significantly faster”, and I'm not finding multiple confirmations of the transcripts of E3 2015 where tis is attributed to? Every time I search I find such claims come back to either:

Dude posted the video.. 1:50s in she says EXACTLY that...

http://www.techpowerup.com/forums/t...ner-than-you-think.214926/page-2#post-3325796

 

[moproblems99]

Significantly faster means 20% - 30% to me. Especially when we are comparing cards two years apart. I don't see the difference being that great except maybe at 4k which would still technically make her correct. But I don't see a snowballs chance in hell of this being 20% faster or more across the board.

 

[newtekie1]

Wait...so because a 390x and 290x are not the same cards because one is 10% faster, that means:

• an overclocked 390x really is a 490x because it is about 10% faster
• an overclocked 290x is really a 390x because they have about the same performance
• an overclocked 290x 8GB cards is ???

OMG I am so confused now....

Performance wise, in the context of the original question, they are in fact not the same card. And as far as AMD is concerned they are not the same card. They even went as far to rename the internal codename for the GPU, and then hounding W1z to change it in GPU-Z.

So, the point still stands, if the Nano was faster than the 390x, they would have said so. They didn't, so it is pretty safe to assume the Nano isn't faster than the 390x and will fall somewhere between the 290x and 390x.

I think the Nano will beat a 'reference' 390X, but still be short of the Fury.

With AMD's PR team, if that was the case they would have said so. They overstate what their products are capable of as it is, so they wouldn't miss an opportunity like that.

Well that was the vendors like Asus and Gigabyte and not NVidia. Its a reference card coming in that small which makes it appealing to some.

The nVidia reference PCB was just as small.

 

[moproblems99]

So, the point still stands, if the Nano was faster than the 390x, they would have said so. They didn't, so it is pretty safe to assume the Nano isn't faster than the 390x and will fall somewhere between the 290x and 390x.

Let me ask you something. If you were trying to sell 390Xs that aren't even released, would you tell someone that you also had something 1/3 size, twice as power efficient, and faster?*

Probably not. Base it off something that is 2 years old, and your new "different" card might still sell.

*All ethical considerations aside.

EDIT:

Also it is asinine to consider the 390X to be a different card than the 290X on grounds other than the 8GB memory, and even that was available on the 290X. AMD has gone through such great lengths to give the appearance of a different card because they want to sell them.

 

[newtekie1]

Let me ask you something. If you were trying to sell 390Xs that aren't even released, would you tell someone that you also had something 1/3 size, twice as power efficient, and faster?*

Probably not. Base it off something that is 2 years old, and your new "different" card might still sell.

*All ethical considerations aside.

Not an issue, all of that applies to the Fury and Fury-X too. If they had a card that beat the 390x it would cost more too, so saying it was better than a 390x wouldn't be an issue.

 

[EarthDog]

So, the point still stands, if the Nano was faster than the 390x, they would have said so..

LOL, no they wouldn;t... think about it...see moprobs post above...think about it. The 390x MSRP is $429 while the Fury is $550. THere is plenty of room for a card there, no? So Pricing has nothing to do with it... moprobs post stands as far as I am concerned... Its simply the fact that they refuse to say rebrand, so they wouldn't want to step on its toes...

With AMD's PR team, if that was the case they would have said so. They overstate what their products are capable of as it is, so they wouldn't miss an opportunity like that.

Disagree.. see moprobs post... they don't want to kill 390x sales... I have to imagine it will come in around $450-$475 MSRP depending on where it actually lands performance wise. Again, I would be FLOORED if it was between the 290x and 390x.

They didn't, so it is pretty safe to assume the Nano isn't faster than the 390x and will fall somewhere between the 290x and 390x

I guess time will tell, but thinking critically about it, and knowing what a marketing machine AMD is, I have to agree with moprobs logic. I don't find 10% to be 'significantly' faster (maybe AMD does). And there is no way in hell they would break out a card that trumps their 'new' 390x they talked about minutes earlier...

They even went as far to rename the internal codename for the GPU, and then hounding W1z to change it in GPU-Z.

That right there doesn't tell you its the same card with more ram and clockspeeds?

 

[moproblems99]

Not an issue, all of that applies to the Fury and Fury-X too. If they had a card that beat the 390x it would cost more too, so saying it was better than a 390x wouldn't be an issue.

That sounded better typing it out than reading it. +1 to you.

 

[newtekie1]

That right there doesn't tell you its the same card with more ram and clockspeeds?

Again, I never argued it wasn't the same card. I said it wasn't the same PERFORMANCE. Not sure why you are having a hard time understanding that.

They are not the same exact card, the clock speeds have changed, and so has the performance.

When talking about a statement in reference to performance, that is all that matters.

By your logic, the i7-4765T is the exact same processor as the i7-4790. So since the i7-4785T outperforms the i7-4765T, then it must outperform the i7-4790. They are the exact same processor after all. Oh wait, clock speeds make a difference...

 

[EarthDog]

I suppose, as I said earlier, it boils down to what the AMD lady said... "SIGNIFICANTLY beats the 290x". We can run in circles all day trying to pin down something quantitative in regards to what significantly is... and we can do the same trying to figure out if she meant reference 290x or an overclocked 290x with 8GB (390x or TriX 8GB LOL!). Assuming the 290x was lumped together with all of them and not just reference, you can see why I feel my performance point on the Nano is logical, versus what I feel yours is a leap of faith.

Oh well... time will tell where this badboy will land.

 

[newtekie1]

SIGNIFICANTLY beats the 290x

Yeah, and remember, the Fury-X "significantly beats the 980Ti" too...

 

[EarthDog]

Did they actually say that? (being serious, I don't recall that). I recall them saying it was the fastest single graphics card out... LOL

 

[64K]

Did they actually say that? (being serious, I don't recall that). I recall them saying it was the fastest single graphics card out... LOL

Well, AMD did "leak" some benches that showed Fury X to be faster in every game tested against the Titan X at 4k.

 

[EarthDog]

ANd they were in those AMD games at least, right? LOL!

 

[Casecutter]

Dude posted the video.. 1:50s in she says EXACTLY that...

http://www.techpowerup.com/forums/t...ner-than-you-think.214926/page-2#post-3325796

Thank you, I apologize I scrolled over that post from arbiter (great work).

I knew it must be out there, although not until arbiter had I found someone who points it factually transpiring. Heck I've listen to that video when it came out and I never noticed it as projecting the encumbrance its' created. The use of the word "significantly" is a weird, as it hard to quantify, like "noisy". The PR folk that wrote or prepped her, did wrong and should've refrained from that in the talking points (heck she appears to be using a promonter) and caution the use of such non-decisive words, for something that can and would be measured.

 

[FordGT90Concept]

Significantly faster means 20% - 30% to me. Especially when we are comparing cards two years apart. I don't see the difference being that great except maybe at 4k which would still technically make her correct. But I don't see a snowballs chance in hell of this being 20% faster or more across the board.

Clock for clock Nano would be 30% faster than 290X but Nano could be missing 30% of its clocks. It really boils down to what the stock clockspeed is. My guess is 750 MHz which translates to faster than 290X at low resolution and "significantly faster" at high resolution.

 

[moproblems99]

Clock for clock Nano would be 30% faster than 290X but Nano could be missing 30% of its clocks. It really boils down to what the stock clockspeed is. My guess is 750 MHz which translates to faster than 290X at low resolution and "significantly" faster at high resolution.

My scientific number is (290X_Perfomance * 2) * (Nano_POWER/290_POWER). That is what they claim with 2 x Perf/Watt of 290X and they are targeting like 170 or 190 watts. If I wasn't lazy and working I would go pull some numbers from charts and see what we have.

All kidding aside, its a shame they couldn't have pushed all these out with the rest of the rebrands, I mean different cards. They probably would have sold a lot more of them. I would have bought one.

 

[64K]

ANd they were in those AMD games at least, right? LOL!

Yeah, there was some cherry picking. I was wrong about the Titan X though. This was "leaked" slides showing Fury X to be faster than 980 Ti at 4K before Fury X came out

http://videocardz.com/56711/amd-radeon-r9-fury-x-official-benchmarks-leaked

It was a part of the Fury X hype train leading up to the release and caused a few people to be disappointed when finally reviewed at the Fury X performance and overclocking.

 

[arbiter]

I really strongly feel that this product will fall between a 380 and 390, if Fiji is really just based off Tonga, wouldn't it make sense to see a 380-like card in Fury (HBM) form?

^skip to 1:50 mark of the video, AMD's president and CEO claimed it would be then 290x so it wouldn't fall between them when its supposed to be faster.

They do make a difference in performance, which is what we are talking about here.
Like I said, I'm pretty sure AMD would have said it outperforms a 390x if it actually did. The fact that they said it outperforms a 290x and not a 390x gives us a great hint that the Nano will fall between the 290x and 390x in performance.

Problem with that between 290x and 390x is only 10% window at most so being between them tight margin to wall which wouldn't mean its "significantly more performance" wouldn't be true.

I'd concur with FordGT90Concept, AMD was not going to compare a next-generation card to a card that they had yet released (390/390X) so all they could/would compare it to was a 290X.
That said, I've been search for the "direct quote" of AMD saying Nano is "significantly faster”, and I'm not finding multiple confirmations or the transcripts of E3 2015 where it is attributed? Every time I search I find such claims come back to either:

yea they wouldn't say 390x since well it wasn't out yet. everyone knew what 390x really is at this point so no reason to keep that board floating anymore.

Yeah, there was some cherry picking. I was wrong about the Titan X though. This was "leaked" slides showing Fury X to be faster than 980 Ti at 4K before Fury X came out

http://videocardz.com/56711/amd-radeon-r9-fury-x-official-benchmarks-leaked

It was a part of the Fury X hype train leading up to the release and caused a few people to be disappointed when finally reviewed at the Fury X performance and overclocking.

Techreports on their podcast that week following the PR thing, they broke down the settings in question. AMD what they did was turn off anything that wasn't using the shaders on the card like AF which everyone pretty runs 16x or least 8x in most cases. Same for other effects if it wasn't something can use their 4096 shaders to do it got turned off hence where numbers came from. Also hence when independent reviews tested the card they had that stuff turned on which is more closer to what real work players would use it was slower.
That leaked slide end up being one AMD used in their release conf.
edit:

He starts at 41:00 mark but you can skip to around 41:40 where he really dives in to it all.

 

[newtekie1]

Did they actually say that? (being serious, I don't recall that). I recall them saying it was the fastest single graphics card out... LOL

Not exactly in those words. You are right, they said the Fury-X was the fastest GPU in the world, besting the Titan-X and 980Ti. And we know it isn't true, so the PR Team's sense of performance has proven to be very skewed.

 

[EarthDog]

It was in their testing...

 

[haswrong]

That is a fairly compact card..... geeze.....

almost looks like they had to enlarge the card to not let the PCI-E connector stick out!

 

[newconroer]

I'm pretty much set on getting on of these, provided the acoustics are good. Even if performance turns out to be really close to 290X level, at 190W and that footprint it's a winner.

Again, in the end the noise level will make it or break it.

A lot of cards have a fan(s) set to ridiculously low RPMs in their BIOS. Manufacturers /vendors do this to achieve a 'out of the box' quiet and low dB rating for reviews. However anyone who has used a GPU knows that if you want the card to not melt itself, that you change the default fan control.
Given that a lot of GPUs use 75-95mm fans, you end up with a jet fighter inside your case.

If the low power usage results in good temperatures, then maybe they can utilize a low fan RPM.

 

[FordGT90Concept]

It's all about the watts. Your typical 2011 CPU is 140w. This card will easily exceed that. Compare the HSF on a Nano to what you'd put on your average 140w CPU. It is likely inadequate so either the fan has to run very fast or the card has to undervolt/underclock itself.