Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

The Vega RX Thread (Rumors and Info)

Jesus christ. Is this 2016. It doesn't really matter that Vulkan, DX12 or Metal is based on Mantle or not. Nvidia was also involved heavily in developing DX12 and Vulkan. API itself is much more than that. API can also be used in favor of Nvidia given by the fact that some DX12 games are favoring Nvidia's architecture.

AMD's heavy involvement in current low level APIs are only thing keeping them in a competitive picture at the moment, since they are miles behind Nvidia in terms of perf/mm and perf/w department.
 
  • Like
Reactions: tuxon86 and AidenShaw
Kpjoslee said:
Jesus christ. Is this 2016. It doesn't really matter that Vulkan, DX12 or Metal is based on Mantle or not. Nvidia was also involved heavily in developing DX12 and Vulkan. API itself is much more than that. API can also be used in favor of Nvidia given by the fact that some DX12 games are favoring Nvidia's architecture.

AMD's heavy involvement in current low level APIs are only thing keeping them in a competitive picture at the moment, since they are miles behind Nvidia in terms of perf/mm and perf/w department.
Click to expand...
There are a lot of "walls of text" around here, just because some people are butthurt about the failure of Vega at launch.

And you see nonsense about claims of "Vega matching GTX 1080" predictions, when the GP102 (GTX 1080Ti/TitanXp) has been around for nearly a year. Shouldn't the just beginning to ship Vega be judged against Nvidia's currently GP102 offerings, rather than older GP104 cards?

It's pure sophistry to crow that Vega sometimes matches the older Nvidia GP104 - while ignoring that it falls way behind the current GP102 while consuming far more power and generating far more heat. Vega FE and Vega RX are failures.

What will the ATI defenders say when consumer Volta shows up soon?
[doublepost=1501720270][/doublepost]
AidenShaw said:
Are you sure that Nvidia didn't release Pascal drivers so that they could sell some Pascal cards?
Click to expand...
No comment, Koyoot?
 
  • Like
Reactions: fendersrule
Asgorath said:
Get back to me when that happens, sure. AMD is claiming Vega will trade blows with a 1080, not a 1080 Ti or TITAN Xp.
Click to expand...
AMD is claiming that Vega will trade blows in MINIMUMS, with GTX 1080.

According to leaked performance, RX Vega 56 is on the same level of performance as GTX 1080, in some cases(Doom for example, or Battlefield 1).

http://www.tweaktown.com/news/58635...-leaked-benchmarks-gtx-1070-killer/index.html

Battlefield 1: 95.4FPS (GTX 1070: 72.2FPS) Civilization 6: 85.1FPS (GTX 1070: 72.2FPS) DOOM: 101.2FPS (GTX 1070: 84.6FPS) COD:IW: 99.9FPS (GTX 1070: 92.1FPS)

How much faster will "Normal" RX Vega be faster than RX Vega 56? And again, software is still not using all of features of Vega that are important for performance.
AidenShaw said:
There are a lot of "walls of text" around here, just because some people are butthurt about the failure of Vega at launch.

And you see nonsense about claims of "Vega matching GTX 1080" predictions, when the GP102 (GTX 1080Ti/TitanXp) has been around for nearly a year. Shouldn't the just beginning to ship Vega be judged against Nvidia's currently GP102 offerings, rather than older GP104 cards?

It's pure sophistry to crow that Vega sometimes matches the older Nvidia GP104 - while ignoring that it falls way behind the current GP102 while consuming far more power and generating far more heat. Vega FE and Vega RX are failures.

What will the ATI defenders say when consumer Volta shows up soon?
[doublepost=1501720270][/doublepost]
No comment, Koyoot?
Click to expand...
You cannot mask that you understand zero in that "wall of texts" posts, about the architectures.

Vega 56 appears to be matching GTX 1080. How much faster will be Vega 64, with properly optimized software? How much Vega 64 will be faster with proper drivers? Now you know why Nvidia released new drivers for Titan Xp?

Stop posting about hardware you know zero about, and all your interest is to create flame war.

And last bit. Those features, that I am talking about, are making out of Vega Nvidia Volta competitor.

The funniest part will be that for some people gaming will now stop matter, and they will shift to other things.
MisterAndrew said:
I would be interested in an RX Vega 56 if they make a Mac edition and if the drivers are native in MacOS. It looks like it beats out a GTX 1070 in some benchmarks. The TDP for that one isn't as bad as the 64.

http://www.tweaktown.com/news/58635...-leaked-benchmarks-gtx-1070-killer/index.html
Click to expand...
This is GTX 1080 level of performance.

Battlefield 1: 95.4FPS (GTX 1070: 72.2FPS) Civilization 6: 85.1FPS (GTX 1070: 72.2FPS) DOOM: 101.2FPS (GTX 1070: 84.6FPS) COD:IW: 99.9FPS (GTX 1070: 92.1FPS)
bf1_2560_1440.png

civ6_2560_1440.png

doom_2560_1440.png

codiw_2560_1440.png
 
Last edited:
  • Like
Reactions: MisterAndrew
koyoot said:
It appears that this shows your ability to read with comprehension.

AMD is claiming that Vega will trade blows in MINIMUMS, with GTX 1080.

According to leaked performance, RX Vega 56 is on the same level of performance as GTX 1080, in some cases(Doom for example, or Battlefield 1).

http://www.tweaktown.com/news/58635...-leaked-benchmarks-gtx-1070-killer/index.html

Battlefield 1: 95.4FPS (GTX 1070: 72.2FPS) Civilization 6: 85.1FPS (GTX 1070: 72.2FPS) DOOM: 101.2FPS (GTX 1070: 84.6FPS) COD:IW: 99.9FPS (GTX 1070: 92.1FPS)

How much faster will "Normal" RX Vega be faster than RX Vega 56? And again, software is still not using all of features of Vega that are important for performance.

You cannot mask that you understand zero in that "wall of texts" posts, about the architectures.

Vega 56 appears to be matching GTX 1080. How much faster will be Vega 64, with properly optimized software? How much Vega 64 will be faster with proper drivers? Now you know why Nvidia released new drivers for Titan Xp?

Stop posting about hardware you know zero about, and all your interest is to create flame war.

And last bit. Those features, that I am talking about, are making out of Vega Nvidia Volta competitor.

The funniest part will be that for some people gaming will now stop matter, and they will shift to other things.
Click to expand...

I don't expect Vega 64 to be much faster than 56 going by the fact that difference between Fury and Fury X wasn't big.
You keep mentioning about how Vega would do with properly optimized software or proper drivers but since most of the touted Vega features being application dependant, it will also end up being their biggest limitation.

It is possible that Vega might end up being close to 1080ti in 2 years, but before we praise about their "fine wine" improvements on drivers, they better learn to stop serving us plain grape juice during launch. GCN is decent architecture compute wise, but AMD also lost big chunk of their marketshare during their GCN years, due to the fact that they never were able to properly optimize their driver from the beginning.

At least Hawaii and Fiji at launch was able to hang with Nvidia's best, but with Vega they are starting off far behind Nvidia's best that their only hope is to compete on price/performance against Nvidia's 1070-1080 line, which is pathetic considering their much bigger die and "superior" memory.
 
  • Like
Reactions: Flint Ironstag
Kpjoslee said:
I don't expect Vega 64 to be much faster than 56 going by the fact that difference between Fury and Fury X wasn't big.
You keep mentioning about how Vega would do with properly optimized software or proper drivers but since most of the touted Vega features being application dependant, it will also end up being their biggest limitation.

It is possible that Vega might end up being close to 1080ti in 2 years, but before we praise about their "fine wine" improvements on drivers, they better learn to stop serving us plain grape juice during launch. GCN is decent architecture compute wise, but AMD also lost big chunk of their marketshare during their GCN years, due to the fact that they never were able to properly optimize their driver from the beginning.

At least Hawaii and Fiji at launch was able to hang with Nvidia's best, but with Vega they are starting off far behind Nvidia's best that their only hope is to compete on price/performance against Nvidia's 1070-1080 line, which is pathetic considering their much bigger die and "superior" memory.
Click to expand...
RX Vega 64 is ALREADY close to GTX 1080 Ti, if the scores are legit. And I am talking about reference model. Not the liquid cooled one.

Load balancing is working well, if the scores are legit, and thats how they appear to be. RX Vega 64 will be 20% faster than RX Vega 56, just from increased core count, and higher memory bandwidth, and higher core clock on Vega 64. And you have to remember, this is situation, where the games are not optimized for using features that are most meaningful in terms of performance, which should increase performance on both of the GPUs by around 30-40%.
 
koyoot said:
RX Vega 64 is ALREADY close to GTX 1080 Ti, if the scores are legit. And I am talking about reference model. Not the liquid cooled one.

Load balancing is working well, if the scores are legit, and thats how they appear to be. RX Vega 64 will be 20% faster than RX Vega 56, just from increased core count, and higher memory bandwidth, and higher core clock on Vega 64. And you have to remember, this is situation, where the games are not optimized for using features that are most meaningful in terms of performance, which should increase performance on both of the GPUs by around 30-40%.
Click to expand...

I really doubt that is the case by how AMD made their marketing slides for RX. If RX Vega 64 performance was close to GTX 1080ti, it would mean it would convincingly beat 1080, and there is no doubt they would have mentioned in slide, yet they only choose to discuss minimum fps advantage of vega 64 against 980ti(!) and 1080 with AMD favored titles.

Oh, BTW, draw stream binning rasterizer is enabled on drivers.
 
  • Like
Reactions: Flint Ironstag and tuxon86
Kpjoslee said:
I really doubt that is the case by how AMD made their marketing slides for RX. If RX Vega 64 performance was close to GTX 1080ti, it would mean it would convincingly beat 1080, and there is no doubt they would have mentioned in slide, yet they only choose to discuss minimum fps advantage of vega 64 against 980ti(!) and 1080 with AMD favored titles.

Oh, BTW, draw stream binning rasterizer is enabled on drivers.
Click to expand...
In 17.30. 17.20(previous versions) it wasn't.
 
cube said:
Isn't Vega 11 supposed to come later this year?
Click to expand...
Yes, but it is bigger GPU. I don't think we will see initially any Vega GPU in 75W TDP.

But there is already rumored Vega 12... So who knows. My sources are not saying anything about this GPU.
 
koyoot said:
About Low-level APIs you do not have to tell me that ;).

Nvidia released new Titan Pascal drivers because of Vega. They had to.
Click to expand...
Yeah but look at what they enabled. Very cheeky that they sell crippled cards and then only enable performance when there is competition. That's usually only Apple's strategy lol
 
cube said:
What does mindshare have to do with it? They do it for segmentation.
Click to expand...
Both companies are releasing products that are not performing at their 100% potential, just so they can bring higher performance, with new drivers releases, when their competition comes up with new products to either stay competitive, or remain in front of their competition.
 
koyoot said:
Both companies are releasing products that are not performing at their 100% potential, just so they can bring higher performance, with new drivers releases, when their competition comes up with new products to either stay competitive, or remain in front of their competition.
Click to expand...
I don't think that applies to RX4 and RX5.
 
koyoot said:
Both companies are releasing products that are not performing at their 100% potential, just so they can bring higher performance, with new drivers releases, when their competition comes up with new products to either stay competitive, or remain in front of their competition.
Click to expand...

I don't think that's accurate. Professional applications like Maya (i.e. those measured by SPECviewperf) were only optimized in the Quadro line of professional graphics cards for NVIDIA. The TITAN "prosumer" line of cards wasn't originally meant for people who wanted to run Maya. When the Vega VE included optimizations that would traditionally only be found in the equivalent FirePro professional graphics cards from AMD, NVIDIA had two choices:

1) Let AMD win with their "prosumer" Vega FE card, and direct their customers to continue buying Quadro cards to run these applictions.

2) Enable some of their Quadro optimizations on TITAN cards so they could compete with (i.e stomp all over) the Vega FE.

NVIDIA chose the latter. There are still hardware-level differences between the Quadro and TITAN, but enabling some of the software optimizations has allowed apps like Maya to run faster on a TITAN card. So, AMD changed the game for these prosumer cards, and NVIDIA responded as one would expect.

People don't like it when it takes 6-12 months or more for their GPUs to finally run at full speed. People expect speed-of-light performance on the day they purchase the card. If AMD's driver team is incompetent and cannot deliver on that, that's their problem and the market share will continue to reflect that.
 
  • Like
Reactions: tuxon86
Asgorath said:
People don't like it when it takes 6-12 months or more for their GPUs to finally run at full speed. People expect speed-of-light performance on the day they purchase the card. If AMD's driver team is incompetent and cannot deliver on that, that's their problem and the market share will continue to reflect that.
Click to expand...
AMD is not incompetent. They are competing with both NVIDIA and Intel and they have less money.
 
  • Like
Reactions: itdk92
Asgorath said:
People don't like it when it takes 6-12 months or more for their GPUs to finally run at full speed. People expect speed-of-light performance on the day they purchase the card. If AMD's driver team is incompetent and cannot deliver on that, that's their problem and the market share will continue to reflect that.
Click to expand...
Similarly to Nvidia you have two choices. Either you pay 499$ today for GTX 1080 performance level, and wait those 6-12 months for applications to make this GPU 20-30% faster than GTX 1080 Ti, or you are a sucker and pay twice for the same level of performances(499$ for GTX 1080, and another 499$ for GTX 2080).

P.S. Maybe you missed, but GTX 1080 level of performance has RX Vega 56, according to the leak. And this GPU costs 399$, and is cut down version with lower core clocks.

P.S.2 The situation has little, or nothing to do with drivers on Vega. What will make Vega GPUs faster, is adding specific code for Vega architecture into the game codes, that will utilize the Primitive Shaders, and Programmable Geometry Pipeline. It will make the GPU 40% faster per clock than Fiji.

P.S.3. Its actually funny. You pay 1200$ for not full product, and when Nvidia releases new drivers that activate features, that should be there in the first place, its all fine and dandy. When AMD is having problems with releasing properly functioning drivers for brand new architecture - they are incompetent. Laughable.
 
Last edited:
koyoot said:
Similarly to Nvidia you have two choices. Either you pay 499$ today for GTX 1080 performance level, and wait those 6-12 months for applications to make this GPU 20-30% faster than GTX 1080 Ti, or you are a sucker and pay twice for the same level of performances(499$ for GTX 1080, and another 499$ for GTX 2080).

P.S. Maybe you missed, but GTX 1080 level of performance has RX Vega 56, according to the leak. And this GPU costs 399$, and is cut down version with lower core clocks.

P.S.2 The situation has little, or nothing to do with drivers on Vega. What will make Vega GPUs faster, is adding specific code for Vega architecture into the game codes, that will utilize the Primitive Shaders, and Programmable Geometry Pipeline. It will make the GPU 40% faster per clock than Fiji.

P.S.3. Its actually funny. You pay 1200$ for not full product, and when Nvidia releases new drivers that activate features, that should be there in the first place, its all fine and dandy. When AMD is having problems with releasing properly functioning drivers for brand new architecture - they are incompetent. Laughable.
Click to expand...

There's a significant difference between a product segment changing (i.e. now Maya is important for the prosumer cards) versus it taking 6+ months for the drivers to be anywhere near done. But, we get it, you think Vega is the best thing ever and AMD can do no wrong.
 
  • Like
Reactions: tuxon86
Asgorath said:
There's a significant difference between a product segment changing (i.e. now Maya is important for the prosumer cards) versus it taking 6+ months for the drivers to be anywhere near done. But, we get it, you think Vega is the best thing ever and AMD can do no wrong.
Click to expand...
In other words, buying an artificially crippled product is fine but not one which the vendor improves to its best effort.
 
  • Like
Reactions: secretbum, ssgbryan and koyoot
Asgorath said:
There's a significant difference between a product segment changing (i.e. now Maya is important for the prosumer cards) versus it taking 6+ months for the drivers to be anywhere near done. But, we get it, you think Vega is the best thing ever and AMD can do no wrong.
Click to expand...
In terms of hardware features, yes it is game changing. Those are even words of Game Developers. Now we have to wait for adoption of them in software. Contrary to most people who arguing over Nvidia vs AMD, I can put any biases aside(actually that is happening, because you may be astounded that I have been using fanless computer with Core i7 7700T, and GTX 1050 Ti, everyday, because I prefer those two companies) and judge the hardware for its capabilities, because I am hardware enthusiast. I may not know a lot about software, but I too much love hardware. And I always judge hardware for its capabilities, whether the features are used or not.
cube said:
In other words, buying an artificially crippled product is fine but not one which the vendor improves to its best effort.
Click to expand...
There are double standards on this forum.
 
cube said:
In other words, buying an artificially crippled product is fine but not one which the vendor improves to its best effort.
Click to expand...

So you're also upset that AMD cripples their consumer cards so they can charge more for their FirePro workstation cards?
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back