Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
That's one game, not a trend.
What I told you that one reviewer reviewed actually Vega vs GTX 1070 in games and without MSAA, and in for example GTA 5, Vega scores almost two times higher than GTX 1070? In Project Cars Vega 56 got 50% advantage over GTX 1070. So maybe it is a trend, but depends on how the Game is programmed.


It depends on games, and how they are programmed, tho, and how drivers of this GPU are programmed. But overall, with properly optimized software, this is what you should see with Vega performance. AMD f***** up with designing GPU that is too difficult for them to optimize the drivers, for launch release. I told you that Vega is most advanced architecture on the planet. This is unfortunately double edged sword.

Story of Vega is simple. Raja directed the team to redesign the GCN so it can be more modern in graphics. The problem is that it is impossible for driver team that is starving for resources to optimize the drivers. Simulations suggest that with proper software Vega 64@ 1.2 GHz will be around 20% faster than GTX 1080 Ti. Im sure that one day this will be reality, but, this day is nowhere close, at this point, because of the complexity of features, and flow of the pipeline, and synchronizing all of them. When Lisa Su has heard about the state of drivers, and performance of Vega - she immediately told Raja and his silicon team to push Vega out of its efficiency curve, to be able to compete with GTX 1080, because this is the sweet spot for this design to "make premium". Total manufacturing cost is 120$, including HBM2 chips, so not that bad.

So for those who bought Vega GPUs: you have to wait for drivers. Primitive Shaders are completely not functioning, and High Bandwidth Cache Controller can work with... maybe 5% of its capabilities.

Here is whitepaper for Vega: http://radeon.com/_downloads/vega-whitepaper-11.6.17.pdf
Very interesting read, however, when you compare it to Nvidia flagship chip whitepapers: https://images.nvidia.com/content/volta-architecture/pdf/Volta-Architecture-Whitepaper-v1.0.pdf you will see differences in budget. 15 pages vs 59 pages. Simple sleek, technical paper, vs technical paper that is packed with ginormous amount of marketing slogans.

From the Vega Whitepaper:
The “Vega” 10 GPU includes four geometry engines which would normally be limited to a maximum throughput of four primitives per clock, but this limit increases to more than 17 primitives per clock when primitive shaders are employed.⁷
 
Last edited:
  • Like
Reactions: MrUNIMOG
Again, in the real world AMD is getting beatten to a bloody pulp by NVidia and no spin will change this for the foreseeable future.

AMD had one job to do and they've gone an ****ed it up with vega.
 
Again, in the real world AMD is getting beatten to a bloody pulp by NVidia and no spin will change this for the foreseeable future.

AMD had one job to do and they've gone an ****ed it up with vega.
Explain then why without AA Vega 56 is faster than GTX 1080 Ti in Dirt 4? Explain why in Project Cars without AA Vega 56 is 50% faster than GTX 1070? Explain why Vega 56 is on average 50% faster than GTX 1070 in GTA 5 with AA off?

I see benchmarks that show advantage in performance for Vega, versus their price competitor. Heck there are benchmarks that show that cheaper 399$ is on par with 499$ GTX 1080, and Vega 64 beating GTX 1080 easily. The only place at this point Vega is beaten is performance per watt, because contrary to what you AMD haters tried to spin, based on false claim by Gibbo, on OCUK forum Prices are where they were. And the performance of this GPU is still without proper drivers. What will happen when it will get drivers? Will it mimic the Dirt 4, GTA 5, and Project Cars performance with AA disabled, but this time in proper graphics settings?

Guys, come on. Look without your glasses on stuff. Overall you are correct, that AMD **** up. But its software problem, not hardware.
 
Last edited by a moderator:
  • Like
Reactions: MrUNIMOG
Your comments on volta are lost in those big wall of text containing mostly AMD pr spin.
Do you think you're the only one here with acces to the internet and hardware reviews site and graph?
We all have read about vega and watched the reviews and have come to our own conclusion.
 
  • Like
Reactions: AidenShaw
I will probably buy a 1050 Ti for my Akitio when the 960M in my laptop stops being enough. I am not very keen on spending on a more powerful eGPU box.
 
Your comments on volta are lost in those big wall of text containing mostly AMD pr spin.
Do you think you're the only one here with acces to the internet and hardware reviews site and graph?
We all have read about vega and watched the reviews and have come to our own conclusion.
Well If you are also interested in something more than only brainwashing benchmarks, Im sure there is a lot of valuable stuff about architectures and hardware in the internet.

Interesting to see AMD PR in a post that contained Volta whitepaper.

I will probably buy a 1050 Ti for my Akitio when the 960M in my laptop stops being enough. I am not very keen on spending on a more powerful eGPU box.
Well if you want GTX 1050 Ti you have two choices: GTX 1050 Ti KalmX that is Passively cooled, or Zotac OC edition, which is fastest GTX 1050 Ti, that does not require 6 pin connector, and does not cost an arm and a leg, compared to other OC versions of 1050 Ti.
 
Last edited:
the most disappointing thing to me about vega isn't even the benchmarks, it's the wattage. is that not a hardware problem?
 
  • Like
Reactions: tuxon86
Well if you want GTX 1050 Ti you have two choices: GTX 1050 Ti KalmX that is Passively cooled, or Zotac OC edition, which is fastest GTX 1050 Ti, that does not require 6 pin connector, and does not cost an arm and a leg, compared to other OC versions of 1050 Ti.
I am not sure the KalmX would fit.

I am thinking of the Gigabyte GeForce GTX 1050 Ti OC LP 4G because:

- It is less than 190mm in length
- It is not more than 2 slots thick
- It does not have auxiliary power input
- It has 4GiB VRAM
- The DVI port is not at the bottom
- It has one more port than usual (HDMI)
- It is low profile
 
Last edited:
Well If you are also interested in something more than only brainwashing benchmarks, Im sure there is a lot of valuable stuff about architectures and hardware in the internet.

Interesting to see AMD PR in a post that contained Volta whitepaper.


Well if you want GTX 1050 Ti you have two choices: GTX 1050 Ti KalmX that is Passively cooled, or Zotac OC edition, which is fastest GTX 1050 Ti, that does not require 6 pin connector, and does not cost an arm and a leg, compared to other OC versions of 1050 Ti.

And yet the first thing you've posted was a benchmark of dirt4...
And why are you injecting some nonsense about the gtx 1050 ti in a vega thread?
Stay on topic please.
 
the most disappointing thing to me about vega isn't even the benchmarks, it's the wattage. is that not a hardware problem?
It depends on performance. Yes, it is big let down in current state of Vega performance.
And yet the first thing you've posted was a benchmark of dirt4...
And why are you injecting some nonsense about the gtx 1050 ti in a vega thread?
Stay on topic please.
What does that benchmark show? Context of what? I used it in what context?
 
It depends on performance. Yes, it is big let down in current state of Vega performance.

What does that benchmark show? Context of what? I used it in what context?

Irrelevant really... You dismiss others for citing benchmarks while using them yourself.
 
Irrelevant really... You dismiss others for citing benchmarks while using them yourself.
Where do I dismiss anything? I asked you simple question. What context that benchmark was used? Stop projecting on me your interpretation, and look at things as they are.

In what context was that benchmark used? It is you who have dismissed, not benchmark, but the point of it, and context of it. Who is dismissing what, in the first place?

That benchmark was used to show you what happens in specific situations, where software is not bottlenecking Vega. What will happen when drivers for Vega will be properly optimized? Will we see the same picture of performance, where Vega 56 will be 10% faster than GTX 1080 Ti? Is it that hard for you to accept that this can be true, especially when we have FIRST CLUES that this might be the case?

Because similar situation we have seen in other review that I have posted video of in this thread. In GTA 5 and Project Cars we have seen similar difference in performance between GTX 1070 and RX Vega 56 as in that Benchmark picture. So it was not single case, and tested in HIGHER resolution, that Dirt 4 was.
 
  • Like
Reactions: MrUNIMOG
Where do I dismiss anything? I asked you simple question. What context that benchmark was used? Stop projecting on me your interpretation, and look at things as they are.

In what context was that benchmark used? It is you who have dismissed, not benchmark, but the point of it, and context of it. Who is dismissing what, in the first place?

That benchmark was used to show you what happens in specific situations, where software is not bottlenecking Vega. What will happen when drivers for Vega will be properly optimized? Will we see the same picture of performance, where Vega 56 will be 10% faster than GTX 1080 Ti? Is it that hard for you to accept that this can be true, especially when we have FIRST CLUES that this might be the case?

Because similar situation we have seen in other review that I have posted video of in this thread. In GTA 5 and Project Cars we have seen similar difference in performance between GTX 1070 and RX Vega 56 as in that Benchmark picture. So it was not single case, and tested in HIGHER resolution, that Dirt 4 was.

Spin, spin, spin...
You took one bench that was favorable to your position and extrapolated on the overall supperiority of Vega based on that benchmark. You're dismissing the whole lot of other bench that have been published where it can barely hold it's own against a year old gpu. That's spin my friend.
 
Spin, spin, spin...
You took one bench that was favorable to your position and extrapolated on the overall supperiority of Vega based on that benchmark. You're dismissing the whole lot of other bench that have been published where it can barely hold it's own against a year old gpu. That's spin my friend.
Well, you have proven me correct, that its not me problem here.

This is my actual post:

AMD does not have enough R&D to produce good drivers. Primitive Shaders for gaming and High Bandwidth Cache Controller as a general feature are still "off" in drivers, tho, HBCC can be turned on in Crimson settings, but it is still limited to maximum 64 GB.

In current state of drivers the GPU is massively bottlenecked by drivers. Its unfinished product. For me, it is unfortunately for AMD, not worth a pennie, in current state. Vega 56 can be considered good value, but IMO, AMD did huge mistake with pushing the voltages and frequency out of process and design comfort zone.


Then was benchmark which was supposed to show that in Vega there is promising stuff.

And then I went:
This is with AA off. Look at minimum framerates. Almost at the same level as averages for 1080. Everything tanks when you push AA, and other features that rely on high geometry throughut. This case shows true potential in Vega GPUs. When this situation we will see in other games? I suppose god only knows at this point.

This post was supposed to show that there is a possibility that Vega is bottlenecked by drivers, and general software state, and that its performance can be solved one day.


Whole situation tells me one thing. Stop projecting on me things that you should look for in the mirror in the first place. I have not claimed that Vega is superior, it can be. We have first clues, because this is not one case, because in GTA 5 and Project Cars with AA off we see the same performance difference between GPUs: GTX 1070 and Vega 56.

Who is spinning anything in the first place, huh?
 
  • Like
Reactions: MrUNIMOG
The sooner Apple leaves AMD, the better. Here's to hoping mMP and iMacs/MacBooks come with Volta in the future :D
Even if Apple software will be better optimized for AMD hardware, and AMD hardware will be on par with Volta, in sheer throughput?

What benefit will be there to switch?
 
  • Like
Reactions: MrUNIMOG
Well, you have proven me correct, that its not me problem here.

This is my actual post:

AMD does not have enough R&D to produce good drivers. Primitive Shaders for gaming and High Bandwidth Cache Controller as a general feature are still "off" in drivers, tho, HBCC can be turned on in Crimson settings, but it is still limited to maximum 64 GB.

In current state of drivers the GPU is massively bottlenecked by drivers. Its unfinished product. For me, it is unfortunately for AMD, not worth a pennie, in current state. Vega 56 can be considered good value, but IMO, AMD did huge mistake with pushing the voltages and frequency out of process and design comfort zone.


Then was benchmark which was supposed to show that in Vega there is promising stuff.

And then I went:
This is with AA off. Look at minimum framerates. Almost at the same level as averages for 1080. Everything tanks when you push AA, and other features that rely on high geometry throughut. This case shows true potential in Vega GPUs. When this situation we will see in other games? I suppose god only knows at this point.

This post was supposed to show that there is a possibility that Vega is bottlenecked by drivers, and general software state, and that its performance can be solved one day.


Whole situation tells me one thing. Stop projecting on me things that you should look for in the mirror in the first place. I have not claimed that Vega is superior, it can be. We have first clues, because this is not one case, because in GTA 5 and Project Cars with AA off we see the same performance difference between GPUs: GTX 1070 and Vega 56.

Who is spinning anything in the first place, huh?

That's your opinion, you're entitled to it just as I'm to mine. Now stop bickering please.
[doublepost=1503179449][/doublepost]
Even if Apple software will be better optimized for AMD hardware, and AMD hardware will be on par with Volta, in sheer throughput?

What benefit will be there to switch?

Vega will never be on par to volta. It can barely match pascal.
 
  • Like
Reactions: AidenShaw
The sooner Apple leaves AMD, the better. Here's to hoping mMP and iMacs/MacBooks come with Volta in the future :D

I often wonder if Apple would ever just buy AMD? They would have complete control over their own graphics chips and they could get MacOS running on AMD processors and cut out Intel.
 
I often wonder if Apple would ever just buy AMD? They would have complete control over their own graphics chips and they could get MacOS running on AMD processors and cut out Intel.
AMD cannot be bought because of x86 license. IF AMD will ever be bought they lose the license.
 
  • Like
Reactions: MrUNIMOG
Ahhh OK thanks.....I did not know this! Good reason.
I should tell more about this. I think one of competitive commisions would be extremely interested if anyone would come and try to buy AMD because of this license. One of reasons why we have AMD is x86. Search for Intel and AMD history in the internet. Good read about the history of Silicon valley overall.
 
I don't see what apple would gain by buying AMD's graphics division that they aren't already getting from their current partnership.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.