AMD Vega (FE and RX) Benchmarks [Updated Aug 10 - RX Vega 64 Unboxing]

[Cloudfire777]

So... seems that the product itself might not be ready at all...

Yeah they have no belief in their product otherwise they would have made a proper launch with lots of noise.

Their whole "Vega RX feels smoother than Geforce bla bla bla" rather than show FPS comparisons was a dead giveaway.
Or "Vega RX will be cheaper than GTX 1080.....if you buy a freesync monitor"

They are a joke. Atleast the CPU division is doing great.
Time to fire Raja and start over again because that dude and Radeon have performed piss poor compared to the old way they did GPUs.

 

[dark zero]

Seems that is the flip coin of the CPU division, heck, I feel that even AMD K12 will surprise even better than the GPU division...

 

[Paratus]

Ok just to make this clear to everyone whose going, "Ha I know what those subjective test mean! It means it's slower"

Congratulations that's exactly what it means. Quite frankly... we all know that.

What AMD is saying is that despite being slower when coupled with a free sync display you won't be able to tell the difference.

What [H] did was perform a tough but fair test , giving AMD the benefit of the doubt by using. DOOM. If that test didn't support what AMD was saying then they were completely full of it.

It seems they are not. Whether that continues to be the case will depend on the relative performance and price of Radeon + free sync vs GTX + G sync.

 

[railven]

Ok just to make this clear to everyone whose going, "Ha I know what those subjective test mean! It means it's slower"

Congratulations that's exactly what it means. Quite frankly... we all know that.

What AMD is saying is that despite being slower when coupled with a free sync display you won't be able to tell the difference.

What [H] did was perform a tough but fair test , giving AMD the benefit of the doubt by using. DOOM. If that test didn't support what AMD was saying then they were completely full of it.

It seems they are not. Whether that continues to be the case will depend on the relative performance and price of Radeon + free sync vs GTX + G sync.

Yeah, but say have a Fury and a FreeSync monitor, what is AMD showing you that is making you consider upgrading?

Or if you had Polaris even Hawaii/Grenada XT and a 1080 FreeSync monitor, or even a 1440p FreeSync monitor? The two games used are already well optimized for the AMD hardware. If I were in these boats AMD isn't showing me a reason to upgrade. If I were outside the AMD ecosystem, and looking from the NV side, unless I were in the game to buy both a FreeSync monitor and a GPU, I'd see hardly any reason to not buy a GTX 1080 especially if they are the same price.

AMD hasn't moved the bar forward very far for their own users. They are doing absolutely nothing to take users from the NV camp. At this point from what I'm reading on Reddit and NeoGaf, people are jumping ship after waiting for so long especially with rumors of high cost and low availability.

This is one weird launch for the history books.

 

[guachi]

I think if Vega come in slower than a 1080 by more than 5% it will be a very tough sell.

The 480 was slower by 5% or so but cheaper and a fantastic deal when combined with a Freesync monitor. Frankly, the 1060 was a terrible deal for someone upgrading. You could get a 1080 Freesync monitor cheap.

Now, obviously, mining had made them scarce.

But needing 2 8 pin connectors (at least I think that's what it is) and a more expensive monitor naked it less appealing.

Because, let's face it, if you don't have a Freesync monitor you shouldn't get Vega. I do. I have a 2160 Freesync monitor that was $300 less than the GSYNC equivalent. So I'm interested in Vega.

But this card requires so many "if" to be true the market can't be that large.

 

[stahlhart]

Everyone was already warned in the Vega/Navi rumor thread:

https://forums.anandtech.com/threads/vega-navi-rumors-updated.2486940/page-189#post-38991727

Personal attacks directed at employees of the companies discussed in this forum are not welcome in this forum. Stop it.

-- stahlhart

 

[railven]

I think if Vega come in slower than a 1080 by more than 5% it will be a very tough sell.

The 480 was slower by 5% or so but cheaper and a fantastic deal when combined with a Freesync monitor. Frankly, the 1060 was a terrible deal for someone upgrading. You could get a 1080 Freesync monitor cheap.

Now, obviously, mining had made them scarce.

But needing 2 8 pin connectors (at least I think that's what it is) and a more expensive monitor naked it less appealing.

Because, let's face it, if you don't have a Freesync monitor you shouldn't get Vega. I do. I have a 2160 Freesync monitor that was $300 less than the GSYNC equivalent. So I'm interested in Vega.

But this card requires so many "if" to be true the market can't be that large.

At this point basically anyone buying a new monitor after 2016 is most likely buying a FreeSync monitor. My wife has a FreeSync monitor only because it was a 21:9 model on sale. But with so many FreeSync monitors entering the userbase, I'd still see AMD struggling to sell their cards to gamers regardless the FreeSync advantage.

I'm already looking to replace my wife's monitor with a G-Sync model. The cost premium is definitely higher but it's a one time cost that will be negated if I stay in the NV camp and she continues to get my old card.

 

[JDG1980]

I'm looking forward to seeing TechReport's excellent frame-time benchmarks for RX Vega when the embargo drops. This should reveal whether AMD marketing's claims of smoothness and a good gaming experience have any empirical basis, or if it's just an inferior card all-around. If something in Vega (HBCC?) optimizes for less frame-time variation and fewer lag spikes, then there might actually be a rational argument for going with it for some people, though the power consumption is still going to be hard to stomach.

 

[ozzy702]

At this point basically anyone buying a new monitor after 2016 is most likely buying a FreeSync monitor. My wife has a FreeSync monitor only because it was a 21:9 model on sale. But with so many FreeSync monitors entering the userbase, I'd still see AMD struggling to sell their cards to gamers regardless the FreeSync advantage.

I'm already looking to replace my wife's monitor with a G-Sync model. The cost premium is definitely higher but it's a one time cost that will be negated if I stay in the NV camp and she continues to get my old card.

I bought a 1440p 144hz g-sync monitor a year ago. Why? Because AMD didn't have a high end card available and to be honest, I'm perfectly fine with being locked into NVIDIA for the next few years because they simply deliver. Hopefully AMD can get their act together by the time I'm looking at 4k 144hz monitors. This coming from a guy that rocked 4850, 4870, 4890, 6870, 7950, 7970 and 390 AMD gpus...

 

[Malogeek]

I'm looking forward to seeing TechReport's excellent frame-time benchmarks for RX Vega when the embargo drops. This should reveal whether AMD marketing's claims of smoothness and a good gaming experience have any empirical basis, or if it's just an inferior card all-around. If something in Vega (HBCC?) optimizes for less frame-time variation and fewer lag spikes, then there might actually be a rational argument for going with it for some people, though the power consumption is still going to be hard to stomach.

Keep in mind that it is in combination with a Freesync monitor for the latest demonstrations and blind tests. That is a factor.

 

[Guru]

It won't be a factor if they can't deliver high end performance. No one would go for a freesync monitor if the RX Vega is slower than the 1080.

Hopefully its faster, but my optimism has gone away completely the more they've been using these fake marketing moves and adding alternative costs and crap like that, it leads me to believe it really is an extremely inferior product.

One year late and its got more power consumption, slower than the 1080 while having more transistors than the 1080ti and being bigger in die size, that is a HUGE failure if you ask me. I'd fire Raja right now and everyone who worked on this card.

I think their biggest hope now is to upscale Polaris, its shown to be an amazing card for DX12 and Vulkan, heck add in 200 more shaders and something like a RX 585 can beat the GTX 1070 in Doom Vulkan!

The RX 580 8GB now easily wins over the 1060 in DX12 and Vulkan, it's not even a competition in most games, where the 580 has up to 12fps advantage in some cases and scenes, on average good 6-7fps advantage!

Upscale the RX 580, double the shaders with a clock of 1500MHz its likely not going to use more power than Vega, we are looking at around 350W. They can add in tighter binning, better clock gating and can shave off 20-40W easily!

 

[3DVagabond]

Keep in mind that it is in combination with a Freesync monitor for the latest demonstrations and blind tests. That is a factor.

But it seems that people who have claimed that Gsync is superior since Freesync released are now claiming that it's Freesync that's making an inferior GPU look as good or better. So, is Freesync now better?

 

[Anarchist Mae]

One year late and its got more power consumption, slower than the 1080 while having more transistors than the 1080ti and being bigger in die size, that is a HUGE failure if you ask me. I'd fire Raja right now and everyone who worked on this card.

Based on the negligible amount of verifiable information we have, it's way over reaching to demand someone be fired.

Also, this is on the same page of this thread:

Everyone was already warned in the Vega/Navi rumor thread:

https://forums.anandtech.com/threads/vega-navi-rumors-updated.2486940/page-189#post-38991727

Personal attacks directed at employees of the companies discussed in this forum are not welcome in this forum. Stop it.

-- stahlhart

 

[wilds]

But it seems that people who have claimed that Gsync is superior since Freesync released are now claiming that it's Freesync that's making an inferior GPU look as good or better. So, is Freesync now better?

I can't tell the difference between them.

 

[3DVagabond]

I can't tell the difference between them.

Agreed.

 

[Cloudfire777]

Ok just to make this clear to everyone whose going, "Ha I know what those subjective test mean! It means it's slower"

Congratulations that's exactly what it means. Quite frankly... we all know that.

What AMD is saying is that despite being slower when coupled with a free sync display you won't be able to tell the difference.

What [H] did was perform a tough but fair test , giving AMD the benefit of the doubt by using. DOOM. If that test didn't support what AMD was saying then they were completely full of it.

It seems they are not. Whether that continues to be the case will depend on the relative performance and price of Radeon + free sync vs GTX + G sync.

They were not full of it?

AMD had two setups. One with 100Hz G-sync at 3440x1440p and one with a Freesync 100Hz. Both running Vulcan.

For one AMD chose Doom and Vulcan because they have an advantage here. Secondly and most importantly, even Fury X can do that resolution on 100FPS+. So whats the point? GTX 1080Ti is obviously framed capped and the chose a game and setting they could easily do, even with the old Fiji card. Its like running a 600HP mustang against a 60HP Lada on a road with 50 mph speed limit and saying "Look, the Lada could do 50 mph too".....

No, this was just another sneaky attempt at marketing the GPU while hiding the real numbers because they know it sucks compared to Pascal.

 

[R0H1T]

They were not full of it?

AMD had two setups. One with 100Hz G-sync at 3440x1440p and one with a Freesync 100Hz. Both running Vulcan.

For one AMD chose Doom and Vulcan because they have an advantage here. Secondly and most importantly, even Fury X can do that resolution on 100FPS+. So whats the point? GTX 1080Ti is obviously framed capped and the chose a game and setting they could easily do, even with the old Fiji card. Its like running a 600HP mustang against a 60HP Lada on a road with 50 mph speed limit and saying "Look, the Lada could do 50 mph too".....

No, this was just another sneaky attempt at marketing the GPU while hiding the real numbers because they know it sucks compared to Pascal.

A better analogy would be racing the Koenigsegg One against a(ny) Ferrari & not going past the fourth gear. Anyway it's pretty clear that the Vega is not a match for 1080Ti & at best will be close to aftermarket 1080's which is a real shame considering it's been in the works for so long!

 

[Zstream]

They were not full of it?

AMD had two setups. One with 100Hz G-sync at 3440x1440p and one with a Freesync 100Hz. Both running Vulcan.

For one AMD chose Doom and Vulcan because they have an advantage here. Secondly and most importantly, even Fury X can do that resolution on 100FPS+. So whats the point? GTX 1080Ti is obviously framed capped and the chose a game and setting they could easily do, even with the old Fiji card. Its like running a 600HP mustang against a 60HP Lada on a road with 50 mph speed limit and saying "Look, the Lada could do 50 mph too".....

No, this was just another sneaky attempt at marketing the GPU while hiding the real numbers because they know it sucks compared to Pascal.

Gasp! A marketing team wanting to make their product look good? How dare they!

 

[OatisCampbell]

It won't be a factor if they can't deliver high end performance. No one would go for a freesync monitor if the RX Vega is slower than the 1080.

Hopefully its faster, but my optimism has gone away completely the more they've been using these fake marketing moves and adding alternative costs and crap like that, it leads me to believe it really is an extremely inferior product.

One year late and its got more power consumption, slower than the 1080 while having more transistors than the 1080ti and being bigger in die size, that is a HUGE failure if you ask me. I'd fire Raja right now and everyone who worked on this card.

I think their biggest hope now is to upscale Polaris, its shown to be an amazing card for DX12 and Vulkan, heck add in 200 more shaders and something like a RX 585 can beat the GTX 1070 in Doom Vulkan!

The RX 580 8GB now easily wins over the 1060 in DX12 and Vulkan, it's not even a competition in most games, where the 580 has up to 12fps advantage in some cases and scenes, on average good 6-7fps advantage!

Upscale the RX 580, double the shaders with a clock of 1500MHz its likely not going to use more power than Vega, we are looking at around 350W. They can add in tighter binning, better clock gating and can shave off 20-40W easily!

I bet you don't own a business then.

You seem to be missing the facts that:

1. Only one team can have the best idea/design.

2. You can't tell a team, "Go make the best design!" They either have the best invention or they don't, and you find out long after choices have been made for the primary design worked on.

3. Money plays a role in the development as you can hire the best design talent if you have more cash.

There are 32 teams in the NFL, the 31 who don't win the Super Bowl don't clean house on staffing. Companies make the moves they with the money they have.

Designing CPUs and GPUs is not stocking shelves.

If a mom and pop store has two stockers and one puts 20% more boxes on the shelf, mom or pop might say, "Sorry Willy, Jeff is doing 20% more work than you consistently, we're going to hire a new stocker.". There's not a lot of risk.

At Raja's level there aren't a lot of candidates and the ones that exist already work for your richer competitors.

 

[Cloudfire777]

I bet you don't own a business then.

You seem to be missing the facts that:

1. Only one team can have the best idea/design.

2. You can't tell a team, "Go make the best design!" They either have the best invention or they don't, and you find out long after choices have been made for the primary design worked on.

3. Money plays a role in the development as you can hire the best design talent if you have more cash.

There are 32 teams in the NFL, the 31 who don't win the Super Bowl don't clean house on staffing. Companies make the moves they with the money they have.

Designing CPUs and GPUs is not stocking shelves.

If a mom and pop store has two stockers and one puts 20% more boxes on the shelf, mom or pop might say, "Sorry Willy, Jeff is doing 20% more work than you consistently, we're going to hire a new stocker.". There's not a lot of risk.

At Raja's level there aren't a lot of candidates and the ones that exist already work for your richer competitors.

Dont come here and tell me that the whole road from engineering to launch have been full of errors and mistakes.

Let me mention a few keywords:
HBM2
Capsaicin
Delays
Capsaicin & Cream
"We announce today that we will announce something soon"
"Today we announce the names!!!"
Delay after delay but lets release professional cards first
Worst efficiency in years
Almost no IPC gain from years and years of engineering and design

Radeon couldnt have executed this any worse. Rubbish.

 

[beginner99]

They are a joke. Atleast the CPU division is doing great.
Time to fire Raja and start over again because that dude and Radeon have performed piss poor compared to the old way they did GPUs.

It's a classic. if you think experts are expensive wait and see what amateurs will cost you. eg. fire expensive engineers and hire cheap ones and that is what you get.

 

[Malogeek]

But it seems that people who have claimed that Gsync is superior since Freesync released are now claiming that it's Freesync that's making an inferior GPU look as good or better. So, is Freesync now better?

Yeah I don't know. Technically G-Sync is a superior product and the panels are better so maybe there is something to Vega that provides a better experience. We'll know.... one day.

 

[tviceman]

This blind testing goes in the face of their never settle campaign a few years ago. They are completely and undoubtedly asking you to settle on inferior performance this time around with Vega.

 

[Cloudfire777]

It's a classic. if you think experts are expensive wait and see what amateurs will cost you. eg. fire expensive engineers and hire cheap ones and that is what you get.

I want to post picture of their Vega team, those who engineered this fantastic Vega architecture, but Im afraid I will come across as a racist. Which Im not.

But feel free to search around for it.

 

[biostud]

I bet you don't own a business then.

You seem to be missing the facts that:

1. Only one team can have the best idea/design.

2. You can't tell a team, "Go make the best design!" They either have the best invention or they don't, and you find out long after choices have been made for the primary design worked on.

3. Money plays a role in the development as you can hire the best design talent if you have more cash.

There are 32 teams in the NFL, the 31 who don't win the Super Bowl don't clean house on staffing. Companies make the moves they with the money they have.

Designing CPUs and GPUs is not stocking shelves.

If a mom and pop store has two stockers and one puts 20% more boxes on the shelf, mom or pop might say, "Sorry Willy, Jeff is doing 20% more work than you consistently, we're going to hire a new stocker.". There's not a lot of risk.

At Raja's level there aren't a lot of candidates and the ones that exist already work for your richer competitors.

One videcard could be fastest at 1080p the other at 4k.

One videocard could deliver 95% of the other, at half the cost.

Not saying it is the case here though.