Thursday, September 27th 2018

NVIDIA's Weakness is AMD's Strength: It's Time to Regain Mid-Range Users

It won't be easy to see an AMD counterpart to NVIDIA GeForce RTX 2080 Ti in the short time. Heck, it will be difficult to see any real competitors to the RTX 2000 Series anytime soon. But maybe AMD doesn't need to compete on that front. Not yet. The reason is sound and clear: RTX prices are NVIDIA's weakness, and that weakness, my fellow readers, could become AMD's biggest strength.

The prices NVIDIA and its partners are asking for RTX 2080 and RTX 2080 Ti have been a clear discussion topic since we learnt about them. Most users have criticized those price points, even after NVIDIA explained the reasoning behind them. Those chips are bigger, more complex and more powerful than ever, so yes, costs have increased and that has to be taken into account.
None of that matters. It even doesn't matter what ray-tracing can bring to the game, and it doesn't matter if those Tensor Cores will provide benefits beyond DLSS. I'm not even counting on the fact that DLSS is a proprietary technology that will lock us (and developers, for that matter) a little bit more in another walled garden. Even if you realize that the market perception is clear: who has the fastest graphics card is perceived as the tech/market leader.

There's certainly a chance that RTX takes off and NVIDIA sets again the bar in this segment: the reception of the new cards hasn't been overwhelming, but developers could begin to take advantage of all the benefits Turing brings. If they do, we will have a different discussion, one in which future cards such as RTX 2070/2060 and its derivatives could bring a bright future for NVIDIA... and a dimmer one for AMD.

But the thing that matters now for a lot of users is pricing, and AMD could leverage that crucial element of the equation. In fact, the company could do that very soon. Some indications reveal that AMD could launch a new Polaris revision in the near future. This new silicon is allegedly being built on TSMC's 12 nm process, something AMD did successfully with its Ryzen 2000 Series of CPUs.
AMD must have learnt a good lesson there: its CPU portfolio is easily the best in its history, and the latest woes at Intel are helping and causing forecast revisions that estimate a 30% market share globally for AMD in Q4 2018. See? Intel's weakness is AMD's strength on this front.

2018 started with the worst news for Intel -Spectre and Meltdown- and it hasn't gone much better later on: the jump to the 10 nm process never seems to come, and Intel's messages about those delays have not helped to reinforce confidence in this manufacturer. The company would be three generations further than they are now without those big problems, and the situation for AMD would be quite different too.

Everything seems to make sense here: customers are upset with that RTX 2000 Series for the elite, and that Polaris revision could arrive at the right time and the right place. With a smaller node AMD could gain higher yields, decrease cost, increase clock frequencies and provide that 15% performance increase some publications are pointing to. Those are a lot of "coulds", and in fact there's no reason to believe that Polaris 30 is more than just a die shrink, so we would have the same unit counts and higher clocks.
That won't probably be enough to make the hypothetical RX 680 catch up with a GTX 1070: performance of the latter is +34% the one we found in the RX 580 on average according to our tests, so even with that refresh we will have a more competitive Radeon RX family that could win the price/performance battle, and that is no small feat.

The new cards would also not target just existing GTX 7/9 Series users, but also those older AMD Radeon users that were expecting a nice upgrade on performance without having to sell their souls. And for the undecided users, the ones that are thinking about getting a GTX 1050/Ti or a GTX 1060, AMD's offer could be quite attractive if price/performance ratio hits NVIDIA where it hurts more.

That would put that new family of graphic cards (Radeon RX 600?) on a pretty good position to compete with GeForce GTX 1000. NVIDIA presumably could still be king in the power consumption area, but besides that, AMD could position itself on that $300-$500 range (and even below that) with a really compelling suite of products.

So yes, AMD could have a winning hand here. Your move, AMD.
Add your own comment

110 Comments on NVIDIA's Weakness is AMD's Strength: It's Time to Regain Mid-Range Users

#76
fynxer
For reals, you think AMD even still has a chance in low and mid. Can tell you that train has left the station.

nVidia will just dump a 12nm shrink of GP104 on Polaris 30, call it GTX 2030 and be done with it.

Lisa Su has no clue how to run Radeon Group, she hires Raja Koduri to head it up and then just runs him over by crippling VEGA development among other things.

I understand why he just gave Su the finger and moved on to Intel, Intel knows Raja's true worth and just gave him a blank check and said do your thing.
Posted on Reply
#77
Vayra86
fynxer
For reals, you think AMD even still has a chance in low and mid. Can tell you that train has left the station.

nVidia will just dump a 12nm shrink of GP104 on Polaris 30, call it GTX 2030 and be done with it.

Lisa Su has no clue how to run Radeon Group, she hires Raja Koduri to head it up and then just runs him over by crippling VEGA development among other things.

I understand why he just gave Su the finger and moved on to Intel, Intel knows Raja's true worth and just gave him a blank check and said do your thing.
This. Although I'm not sure about the Raja part. And also not about GP104 performance in the x30 bracket :D more like an x60(ti).
Posted on Reply
#78
jabbadap
del42sa
"This new silicon is allegedly being built on TSMC's 12 nm process, something AMD did successfully with its Ryzen 2000 Series of CPUs. "

AFAIK Ryzen is build on GloFo 12nm ....
Not to mention in which TSMC 12nm process, I'm quite sure that 12 FFN is for nvidia only. Sure there is 12nm FFC but that is for mobile applications.
Posted on Reply
#79
iO
If a "12nm shrink" of Polaris was financially viable they would have done it long ago...
Posted on Reply
#80
B-Real
RX 580 8GB already won against the GTX 1060 6GB in the mid section theoretically. It is a bit faster, has more VRAM and its cheaper. Yes, it consumes more power. Most people play maximum of ~3 hours a day. It nearly doesn't affect your electricity bill. What is more important out of the 4? But you have the brainwashed people there saying "I buy Intel and NV" or the sellers only saying customers to buy Intel and NV. Those guys have to be informed about the truth.

You just have to have well optimized games like Forza Horizon 4 and you will see that AMD is way better than it shows in (earlier) NV sponsored titles, where sometimes they are on par with a lower tier NV card. Newer titles like AC: Origins or Shadow of the Tomb Raider show that it may not be this way anymore, as a Vega 56 is on par with a 1070 in NV sponsored games.

gamerman
...
What a comment. :D Saying the 2080 is an excellent product just shows how green you are. People DID NOT expect to get a 70-80% jump again as with Pascal, but if you get a little less performance increase than the 700-900 change, where the 900 cards were 50$ cheaper, and now you get the new cards 100-150, and 500-600$ MORE EXPENSIVE, saying these are excellent products shows how biased you are. You are also forgetting that maybe around 2% of the gamers buy the high end GPU of the current gen from NV, which is marginal. Most of the cards come from the 1050Ti and 1060 performance level. And you even had to mention that Intel is the king in CPU. You just forget (again) that most of the gamers use mid range GPUs, from which you don't get a single fps boost with Intel compared to AMD, even not with mid-high 1070... And I can assure you that most 1080Ti owners do not play CS: GO with 720p minimum settings, but use it on Ultrawide or UHD monitors, where the that maximum 10% CPU difference totally disappears. I just laugh at you so loud.
Posted on Reply
#81
londiste
Power is not about electricity bill with GPUs. It is about heat and noise.
Posted on Reply
#82
B-Real
londiste
Power is not about electricity bill with GPUs. It is about heat and noise.
Check reviews, AIB modell heat difference was less than 10 degrees, and noise wasn't much more either.

Techpowerup MSI Gaming X:
GTX 1060 6GB 67 degrees with 28 DBa
RX 480 8GB 73 degrees with 31 DBa

Guru3D MSI Gaming X:
GTX 1060 6GB 65 degrees with 37 DBa
RX 480 8GB 73 degrees with 38 DBa
Posted on Reply
#83
londiste
You do realise that these do not exist in a vacuum, right?
Given the same cooler, GPU with more power consumed will emit more heat. This will either translate into more heat and/or faster fan speed on the cooler. More emitted heat will also heat up surrounding air.

And while most readers of TPU likely do have proper air movement in the chassis, many people do not. I have seen many many people who have come to complain about performance issues. When looking at the computer, they slapped a GPU into a random machine in mATX tower than has one fan lazily blowing air in. After about 20 minutes of gaming, temperature inside the case is 50+C. And that gives interesting results.

In general, less heat, less problems.
Posted on Reply
#84
jabbadap
iO
If a "12nm shrink" of Polaris was financially viable they would have done it long ago...
Are they still on making? They already could have moved to GF 12nm process for Polaris, that would not even be a that big change. I seriously don't think that Zen+ would take all the manufacturing capacity from that node.
Posted on Reply
#85
B-Real
londiste
You do realise that these do not exist in a vacuum, right?
Given the same cooler, GPU with more power consumed will emit more heat. This will either translate into more heat and/or faster fan speed on the cooler. More emitted heat will also heat up surrounding air.

And while most readers of TPU likely do have proper air movement in the chassis, many people do not. I have seen many many people who have come to complain about performance issues. When looking at the computer, they slapped a GPU into a random machine in mATX tower than has one fan lazily blowing air in. After about 20 minutes of gaming, temperature inside the case is 50+C. And that gives interesting results.

In general, less heat, less problems.
I don't think I said less heat is not better, I just linked the results that show there are marginal differences. We are not speaking of 390X numbers, where an MSI lightning got 82 degrees with 40 DBa (Techpowerup also). That's what stuffy people should accept. Times are changing, and you shouldn't rate a new product even from the numbers of the previous generation.
So you say most people buy a 1060-1070 with a 10$ chassis? :D
Posted on Reply
#86
londiste
B-Real
So you say most people buy a 1060-1070 with a 10$ chassis? :D
Oh god no. But I have seen enough people who did so a lot of people do and I shudder at the thought. :twitch::D

B-Real
Check reviews, AIB modell heat difference was less than 10 degrees, and noise wasn't much more either.

Techpowerup MSI Gaming X:
GTX 1060 6GB 67 degrees with 28 DBa
RX 480 8GB 73 degrees with 31 DBa

Guru3D MSI Gaming X:
GTX 1060 6GB 65 degrees with 37 DBa
RX 480 8GB 73 degrees with 38 DBa
These are high end of the selection for these cards. Good cards sporting very good coolers.
I do not know what was up in TPU's RX480 review but 196W for RX480 sounds just wrong. Maybe a bad card.

Guru3D's figures:
Power consumption 168W vs 136W (23% more).
Ambient is probably 22C, lets say 20C. Delta temps 53 vs 45 (17% more).
1 DBa probably means a slightly slower fan speed so that is the rest of the difference.

Btw, noise is a logarithmic scale. 3 DBa louder is about 30% more perceived loudness.
Posted on Reply
#87
Casecutter
looniam
at launch how reviewers observed the power via pci-e slot running a little out of spec.
And such launch issues are always going to get by and out, even by even for the biggest companies selling premium products. Does that really keep people on the fence, not that many. In AMD's power issue, sure if you had some OE box with a underwhelming mobo like most, then with some overrate PSU, some of those saw that rear it's ugly head.
https://www.techpowerup.com/247985/nvidia-fixes-rtx-2080-ti-rtx-2080-power-consumption-tested-better-but-not-good-enough

looniam
personally, i'm convinced i will be holding on to my 980ti until it blows up.
Casecutter
People buy: A) On price; B) Monitor resolution now and near future; C) If they see or find games that tax the experience/immersion of play.
So, right your 1440p monitor and games you play... you feel are good holding on... While I can't expect you to ante up perhaps another $500 over what 2-1/2 years. Although, there are gamer who are on 280X that are leaning to 1440p and if a $230 card came up they might see the value to purchase.

Camm
Steam survey shows the 1060 outpacing it around 5-1. Which is a shame, the 580 is a better card than the 1060 most of the time
So you're saying AMD sold 580's in mass, but those who bought them (miners) weren't showing up on Steam playing games... you don't say? And I'm surprised that for every say 100 - 1060's (all versions ) AMD had 20 folks showing up with a 580, even though all 1060's together and had been on the market over almost a year before a 580 showed and still best's it. Or, does that include the 480, and the 470/570's... Cause if it's just a 580's wow how unfair; it still shows AMD was actually kicking-ass, even while the phenomenal pricing got to from mining... AMD still had 1in5 showing up with a 580, while be cognizant that such prices drove folks to other... How are you exactly extrapolating the data?
Posted on Reply
#88
DeathtoGnomes
fynxer
For reals, you think AMD even still has a chance in low and mid. Can tell you that train has left the station.

nVidia will just dump a 12nm shrink of GP104 on Polaris 30, call it GTX 2030 and be done with it.

Lisa Su has no clue how to run Radeon Group, she hires Raja Koduri to head it up and then just runs him over by crippling VEGA development among other things.

I understand why he just gave Su the finger and moved on to Intel, Intel knows Raja's true worth and just gave him a blank check and said do your thing.
another Nvidia shill without a clue. The sky is falling!
Posted on Reply
#89
notb
B-Real
Check reviews, AIB modell heat difference was less than 10 degrees, and noise wasn't much more either.

Techpowerup MSI Gaming X:
GTX 1060 6GB 67 degrees with 28 DBa
RX 480 8GB 73 degrees with 31 DBa

Guru3D MSI Gaming X:
GTX 1060 6GB 65 degrees with 37 DBa
RX 480 8GB 73 degrees with 38 DBa
Comparing values like that is just totally pointless.

Sound level measured in dB is logarithmic and relative (dB are used to compare, not to give absolute values).
31 dB vs 28 dB means that RX480 makes 40% more noise (measured in how you feel it) and wastes twice as much power on noise.

Temperature is even worse, because you should be thinking about the distance to ambient level. And when you think about actual impact on your comfort, you should be thinking about heat...
That's why review sites that do this properly report thermal rise, not temperature (and often give very extensive information on how the test setup looks).

The simplest explanation I can give is this: 73*C vs 65*C most likely means something like 50*C vs 42*C of temperature rise. And since the coolers are more or less the same (they move similar quantities of air) it means the RX480 generates A LOT more heat.
Posted on Reply
#90
efikkan
ArbitraryAffection
On the note of Memory Bandwidth: Why would AMD opt for a quad-stack, 4096-bit interface on Vega 20 (This is confirmed as we have seen the chip being displayed), with potentially over 1TB/s of raw memory bandwidth if it wasn't at least somewhat limited by memory bandwidth? Or is that purely due to memory capacity reasons? Honestly almost everyone I talk to about GCN says it is crying out for more bandwidth. It's also worth pointing out that NVIDIA's Delta-Colour Compression is significantly better than AMD's in Vega: 1080 Ti almost certainly has quite a bit more effective bandwidth than Vega when that is factored in.
Well, I was talking about gaming performance, where Vega 64 is outperformed by GTX 1080 which has much less memory bandwidth and computational power. So to make it clear; Vega is not bottlenecked in gaming by memory bandwidth or computational performance.
There are certainly professional workloads where even more memory bandwidth could be useful, which is what Vega 20 is targeting.

ArbitraryAffection
So resource utilisation is a major issue for Vega, then. Do you think there is any chance they could have 'fixed' any of this for Vega 20? I won't lie, I've been kinda hoping for some Magical Secret Sauce for Vega 10, perhaps NGG Fast Path or the fabled Primitive Shaders. -shrug- even if it doesn't happen, I am satisfied with my Vega 56 as it is, I am only playing at 1080p, 60 Hz so it is plenty fast enough.
I seriously doubt there will be any major changes in Vega 20, it's mostly a node shrink with full fp64 support and potentially some other hardware for professional use. AMD would need a new architecture to fix GCN, Vega 20 wouldn't do that, and probably Navi will just be a number of tweaks.
Posted on Reply
#91
looniam
Casecutter
So, right your 1440p monitor and games you play... you feel are good holding on... While I can't expect you to ante up perhaps another $500 over what 2-1/2 years. Although, there are gamer who are on 280X that are leaning to 1440p and if a $230 card came up they might see the value to purchase.
please don't read too much into a few words i wrote. besides that i do not feel good hanging on, in a 2.5 years period before i bought the 980ti i went from a gtx 570 to a 770, 780ti and then the 980ti. i don't base my purchases on other people but what i want and what i can reasonably pay. and honestly, only one purchase was gaming related.
Posted on Reply
#92
notb
efikkan
probably Navi will just be a number of tweaks.
+ an Nvidia Tensor competitor, likely.
Posted on Reply
#93
Jism
efikkan
Well, I was talking about gaming performance, where Vega 64 is outperformed by GTX 1080 which has much less memory bandwidth and computational power. So to make it clear; Vega is not bottlenecked in gaming by memory bandwidth or computational performance.
There are certainly professional workloads where even more memory bandwidth could be useful, which is what Vega 20 is targeting.


I seriously doubt there will be any major changes in Vega 20, it's mostly a node shrink with full fp64 support and potentially some other hardware for professional use. AMD would need a new architecture to fix GCN, Vega 20 wouldn't do that, and probably Navi will just be a number of tweaks.
The professional market does require the quad stacked HBM2 and as much as possible memory. There are cards (professional ones) that allow for 1TB of SSD storage to be accessed as it was video memory. https://www.extremetech.com/extreme/232416-amd-announces-new-ssg-a-gpu-with-1tb-of-ssd-storage-attached

The point with AMD is, it creates card primarily for Pro market, and downsizes this for the consumer market (gaming). Nvidia does the very same. You cant flash Geforce's these days into Quaddro's anymore without performance being crippled compared to a Pro card. And due to this vega is behind. And it took a few driver revisions to get it on par with the 1080 and not 1080Ti.

Vega 56 would be worth better for your money, esp the flashing to 64 bios, overclocking and undervolting. These seem to have very good results as AMD was pretty much rushing those GPU's out without any proper testing about power consumption. The Vega arch on this procede is already maxed out. Anything above 1650Mhz and a full load applied is running towards 350 to 400W terrority. Almost twice as a 1080 and proberly not even 1/3rd performance more.

The refresh on a smaller node is good > it allows AMD to lower power consumption, push for higher clocks and hopefully produce cheaper chips. The smaller you make them the more fit on a silicon wafer. RTX is so damn expensive because those are frankly big dies and big dies take up alot of space on a wafer.

The Polaris was a good mid-range card, and still is. Pubg does excellent at 75Hz/FPS lock at WQHD. In my opinion people dont need 144fps on a 60hz screen. Cap that and you can half your power bill easily. :) Something you dont hear people saying either.
Posted on Reply
#94
notb
Jism
The professional market does require the quad stacked HBM2 and as much as possible memory. There are cards (professional ones) that allow for 1TB of SSD storage to be accessed as it was video memory. https://www.extremetech.com/extreme/232416-amd-announces-new-ssg-a-gpu-with-1tb-of-ssd-storage-attached
That SSG was certainly an interesting product, but was it useful?
I've yet to see an actual use of this card. AFAIK it's not even being offered by OEMs (but I'd love to be surprised).
There's also one other argument for it being pointless - Nvidia ignored the idea. It's not that hard to put an SSD into a Tesla or something.
The point with AMD is, it creates card primarily for Pro market, and downsizes this for the consumer market (gaming).
Yeah... the Pro market disagrees.
AMD may be thinking they're making GPUs for pros, but actually they're still just making powerful chips. A pro product has to offer way more than just performance.
To be honest, I don't understand why this is happening. AMD is just way too big to make such weird mistakes.
I doubt AMD share in enterprise GPU market is larger than in CPU one...

Radeon does make very good custom chips though. So when someone orders a particular setup, large potential of GCN can finally explored. Consoles are great. The Radeon Pro GPUs inside MacBooks are excellent.
Also, the Radeon Pro made for Apple is beautifully efficient unlikely desktop parts (or even other mobile ones).
It shows that the power consumption / heat issues of GCN are a result of either really bad tuning or the quality is really bad (i.e. Apple gets all the GCN that gets near Nvidia's quality).

There's not a single AMD GPGPU accelerated machine on Top500 as well. There's nothing based on EPYC either.
Posted on Reply
#95
Readlight
Another product or tech who will be worthless after some time. especially mobiles when they break, there is no moneyback.
Posted on Reply
#96
Casecutter
As this has gone off the rails as to AMD having any new mid-range offerings, I'll just ask.

Is there a new product coming... something, or AMD just goes idle for 12-18mo's maintaining with Polaris products as they are?
Posted on Reply
#97
hat
Enthusiast
Jism
The Polaris was a good mid-range card, and still is. Pubg does excellent at 75Hz/FPS lock at WQHD. In my opinion people dont need 144fps on a 60hz screen. Cap that and you can half your power bill easily. :) Something you dont hear people saying either.
Just... no... you cannot halve your power bill by tweaking settings or buying more efficient hardware, even if you flushed your PC down the toilet (and hence it draws no power at all).
Posted on Reply
#98
W1zzard
Rockarola
Welcome to the hornets nest, you wrote about red/green...now you have to live with a plethora of fanboys, trolls and a few shills.
(good piece, but anything mentioning red/green in a positive fashion WILL get trolled into oblivion by those who subscribe their version of the truth)
That's pretty much exactly what I told him when he wanted to write the article :)
Posted on Reply
#99
Jism
hat
Just... no... you cannot halve your power bill by tweaking settings or buying more efficient hardware, even if you flushed your PC down the toilet (and hence it draws no power at all).
Yeah, it's called Frame rate target control. It works really well. And it almost halves the power consumption if you dont need 144Hz on a 75Hz screen. The heat output is considerable lower.
Posted on Reply
#100
Caring1
Steevo
I think you fail to understand the majority of users don't buy high end cards, but mid-range or low end cards. Which is why Intel is the leader in GPU sales.
Intel? :confused:
Posted on Reply
Add your own comment