AMD RX Vega Reportedly Beats GTX 1080; 5% Performance Improvement per Month

[EarthDog]

I can't read squat off that image xkm...

 

[MrGenius]

1730mhz clock speed it will touch stock GTX 1080ti FE

Where did you find this?

 

[xkm1948]

I can't read squat off that image xkm...

I guess that was the intention of the results original maker. Original run added as well

 

[Crap Daddy]

Just gonna leave this here to fuel the fire.

http://www.3dmark.com/fs/13051960

If that's for Vega it is on par with a factory overclocked 1080 which can be had since about June 2016.

 

[notb]

I don't think you know what you are talking about. It's easier to stir into the skid like you do with your opinion right now getting Ryzen into the picture. We are talking about Vega here not Ryzen, This competition is between AMD and NVIDIA not AMD and Intel. What's your point here? You put NV and Intel into one basket? or what is it?

No, this is not about AMD vs NVIDIA. This is about @RejZoR's confusing comments.

With this logic you been praising Nvidia for several weeks as well. So you did with Intel. and??

I'm consistent in my opinion about the way companies do business and improve their products.
Some people are consistent in their support toward a company, regardless of what they do.

 

[ratirt]

No, this is not about AMD vs NVIDIA. This is about @RejZoR's confusing comments.


I'm consistent in my opinion about the way companies do business and improve their products.
Some people are consistent in their support toward a company, regardless of what they do.

You can say that over and over. There's always a way for you to give a reason for what you have wrote and defend yourself.
Don't use Intel in this thread and Ryzen. It's pointless. Just a suggestion. I see nothing confusing with what RejZoR said. Maybe you just can't see his point of view or you simply don't understand. We learn all the time so don't be alarmed you'll catch up. Consistent? What makes you think Rej... is not consistent in what he says? Because he says good stuff about one company but also about the other if he sees it is worth to mention? It's an opinion and everybody can have one. and stop pointing that out it's just stupid.
People can always change their mind cause they saw something they haven't seen earlier and that made them change the decision or point of view or approach.

Do business? I think you know nothing about it. You think you know but that's different. That's my opinion and you can always disagree.

 

[RejZoR]

Its funny how people quickly brand you an AMD fanboy even though you have an Intel CPU and NVIDIA GPU in the system as soon as you say any kind of praise towards Radeon or Ryzen. So bizarre. If I were really an AMD fanboy, why in hell would I acknowledge smart decision by NVIDIA to go TBR on 28nm and avoid power issues almost entirely. Or the supremacy of Pascal. I never ever said Pascal was crap. It clearly isn't. It's incredibly fast. But RX Vega just sparks more interest in me. I like exotic stuff and RX Vega is filled with all the exotic goodness. Most of my purchases are out of curiosity and not out of necessity. It's one of reasons why I regret not going with R9 Fury X back then and why I skipped Pascal because it was essentially more of the same Maxwell 2 that I have now. But RX Vega, it has so much really fiddly exotic stuff planned, that alone makes me want to have that card even if it's slightly slower than Titan Xp. It's gonna be cheaper in that case so I'm fine with that. And being almost maxed out D3D12_1 level, HBC with memory extending, makes it rather future proof, which means it might turn out rather well long term. Unless something really exciting comes from NVIDIA which would shift my cravings for exotic stuff to the green side again.

 

[arbiter]

Since the clock speed of the 687F:C1 RX Vega has remained the same throughout this benchmark history, I think it's fair to say these improvements have come out purely at the behest of driver and/or firmware level performance improvements.

Looking at the chart i think says it remained the same clock but why does it say 1630mhz+? What does the + at the end mean? I would bet that isn't same clocks from first test to last.

 

[cdawall]

Just gonna leave this here to fuel the fire.

http://www.3dmark.com/fs/13051960

http://www.3dmark.com/fs/12667341

So a 1080Ti is what 50% faster?

 

[efikkan]

The folks at Videocardz have put together an interesting chart detailing the 687F:C1 RX Vega's score history since benchmarks of it first started appearing, around three months ago. This chart shows an impressive performance improvement over time, with AMD's high-performance GPU contender showing an improvement of roughly 15% since it was first benchmarked. That averages out at around a 5% improvement per month, which bodes well for the graphics card... At least in the long term. We have to keep in mind that this video card brings with it some pretty extensive differences from existing GPU architectures in the market, with the implementation of HBC (High Bandwidth Cache) and HBCC (High Bandwidth Cache Controller). These architectural differences naturally require large amounts of additional driver work to enable them to function to their full potential - full potential that we aren't guaranteed RX Vega GPUs will be able to deliver come launch time.

So, using linear regression we can expect this to beat GV100 next year?

We never know how much a product will improve after release, but usually it's in the range of 5-15% for new architectures. But considering that RX Vega is a little delayed, we can expect the drivers to be pretty mature on release.

 

[notb]

You can say that over and over. There's always a way for you to give a reason for what you have wrote and defend yourself.

That's called being consistent.

I see nothing confusing with what RejZoR said. Maybe you just can't see his point of view or you simply don't understand.

Or maybe you don't understand? Did you consider that?

Consistent? What makes you think Rej... is not consistent in what he says?

Because he's been bashing Intel for a strategy he now finds acceptable for an AMD product. I think I've written that clear.

People can always change their mind cause they saw something they haven't seen earlier and that made them change the decision or point of view or approach.

So what changed his mind about small incremental updates? The fact that AMD also does it?

Do business? I think you know nothing about it. You think you know but that's different.

Where did that come from? :-o How is this connected to the discussion? And how do you know?
And what do you mean by "do business"? It sounds like owning a sex shop.

That's my opinion and you can always disagree.

That's the American way. You are from US, aren't you?
You say I can disagree, but you've just criticized me for not agreeing with someone else. What's happening here?

 

[MrGenius]

http://www.3dmark.com/fs/12667341

So a 1080Ti is what 50% faster?

~36.92% faster.

 

[cdawall]

~36.92% faster.

What's sad is that's a slow result (mine)

 

[xkm1948]

What's sad is that's a slow result (mine)

And that Vega was clocked to explosion

 

[cdawall]

And that Vega was clocked to explosion

To be fair we could see a better clocked card drop as yields hit. I keep hearing rumors of 1800mhz, but that will not fix a 40+ percent gap.

 

[okidna]

Looking at the chart i think says it remained the same clock but why does it say 1630mhz+? What does the + at the end mean? I would bet that isn't same clocks from first test to last.

I think they meant that those with "+" sign is possibly an overclocked results, 3DMark11 doesn't do well in detecting overclocking in unrecognized cards.
In my opinion, the #7, #8, #9 result are stock result, and the rest is overclocked results.

 

[TheHunter]

Where did you find this?

Well here is a 1080TI OC 2075MHZ and it has 44K by gpu score,. That 37K must be cpu bottleneck.


P30 752 with NVIDIA GeForce GTX 1080 Ti(1x) and Intel Core i7-5820K Processor
Graphics Score 44 550
Physics Score 16 351
Combined Score 15 364

 

[RejZoR]

To be fair we could see a better clocked card drop as yields hit. I keep hearing rumors of 1800mhz, but that will not fix a 40+ percent gap.

A functional TBR rasterizer and clock increase can. Why do you think Maxwell 2 "out of nowhere" dropped massive performance gains at impressive efficiency?

I mean, we've seen TBR doesn't work on Vega FE, it's not a secret. By not being focused on gaming and with immature drivers it was still reaching GTX 1070/1080 speeds. People really don't seem to realize how much of a difference fully functional drivers with all new features enabled and actually working TBR rasterizer makes. In terms of power and performance. I mean, am I the only one who remembers Kyro II from what, decade and a half ago which was the first TBR powered desktop card ever that really worked? It looked totally unimpressive spec wise and yet it was taking on GeForce 2 GTS/Ultra (which was on paper superior in every aspect), sometimes literally humiliating it entirely and taking even on GeForce 3?

Fast forward to 2017 and you have a chip that is on paper a brute as it is in terms of clock, shader count and memory bandwidth/latency. And then you enable TBR on it properly to render games. What do you people expect? 5-10fps boost only? C'mon, get real. Maxwell 2 wasn't just magically faster, it was because of TBR along with few other similar power/performance technologies (like framebuffer compression).

 

[RejZoR]

Eh, that's only because it does matter. It means AMD simply cannot match Nvidia. If they would completely forego high-end and offer something in the mid-range that would wipe the floor with Nvidia, yes, then you could say it doesn't matter what they do about high-end. But since they're playing the same game, standings are relevant.
And don't forget Vega FE: slightly faster than Titan Xp, still needs more power to pull it off. Much cheaper, though.

Oh my god, this is why I get rage episodes. We know Vega FE drivers are totally immature, we also know Tile-based Rendering doesn't work. AMD released Vega FE with drivers in such state because it was naturally assumed it would be used in productivity tools, not benched in freaking games. And yet, here we are with people making gaming performance assumptions on a card with totally immature drivers which was meant for productivity first and gaming second. AMD was simpy hoping to release Vega FE as is and fix the gaming stuff later since it's not exactly meant for that. And people are going batshit crazy over it. I watched PCPer stream and I otherwise respect the guys, they generally know their shit, but most of the stream and also their later recap, I was like, guys, what the hell are you doing... It's like they threw all logic, common sense and entire existing knowledge about graphic cards, drivers and gfx card releases out the window the moment they got Vega FE in their hands.

That's about as insane as whining over Lamborghini Aventador not being all that great and harvesting wheat. But Lamborghini also makes tractors! It should perform great at that too!

 

[ratirt]

That's called being consistent.

Or maybe you don't understand? Did you consider that?

Because he's been bashing Intel for a strategy he now finds acceptable for an AMD product. I think I've written that clear.

So what changed his mind about small incremental updates? The fact that AMD also does it?

Where did that come from? :-o How is this connected to the discussion? And how do you know?
And what do you mean by "do business"? It sounds like owning a sex shop.

That's the American way. You are from US, aren't you?
You say I can disagree, but you've just criticized me for not agreeing with someone else. What's happening here?

Well that's your point of view. Somebody has his own points of view. With what you wrote it would seem that you don't go along with people who disagree. Maybe you don't see the bigger picture. I don't know. American, Polish or UK does it really matter to You? I know why American since you pointed once my bad English or common words usage or other crap right? that wasn't OK, right Dude? Don't need to bring that one up.
I disagree with attacking others cause they don't vouch for one company but point goodness in each one and technology improvements. That's what it is with technology mostly, improvements.
AMD does the same? Well if you compare excavator the last one released before Ryzen then you will understand. So I disagree here. Even though it took them long time for the new CPU's.
Consistent. This again. You keep saying that over why? From my perspective you are not consistent you just to see other stuff.
Well nevermind.

Well here is a 1080TI OC 2075MHZ and it has 44K by gpu score,. That 37K must be cpu bottleneck.


P30 752 with NVIDIA GeForce GTX 1080 Ti(1x) and Intel Core i7-5820K Processor
Graphics Score 44 550
Physics Score 16 351
Combined Score 15 364

are you sure it's a CPU bottleneck? Do we know what was the OC of the 1080TI on the slide? Anyway wonder if this is true about Vega. If it is then that's a great news Maybe the benchmark was made in a different resolution than 1080p? 2k or 4k maybe? Would that make the scores right?

To be fair we could see a better clocked card drop as yields hit. I keep hearing rumors of 1800mhz, but that will not fix a 40+ percent gap.

Where did you hear that rumor? if Vega can hit 1730Mhz that's what the graph shows that's great. Hope that is true.

 

[cdawall]

A functional TBR rasterizer and clock increase can. Why do you think Maxwell 2 "out of nowhere" dropped massive performance gains at impressive efficiency?

It is somewhat strange that Draw Stream Binning Rasterization has not been fixed in Vega considering in January it was mentioned by amd that it was a big part of Vega working correctly.

I mean, we've seen TBR doesn't work on Vega FE, it's not a secret. By not being focused on gaming and with immature drivers it was still reaching GTX 1070/1080 speeds. People really don't seem to realize how much of a difference fully functional drivers with all new features enabled and actually working TBR rasterizer makes. In terms of power and performance. I mean, am I the only one who remembers Kyro II from what, decade and a half ago which was the first TBR powered desktop card ever that really worked? It looked totally unimpressive spec wise and yet it was taking on GeForce 2 GTS/Ultra (which was on paper superior in every aspect), sometimes literally humiliating it entirely and taking even on GeForce 3?

You probably are. I find it unfathomable that we saw working prototypes running games in what December of last year, yet 7 months later we don't have a driver on release that has functions bragged about by AMD in January that actually work. Functions that under a different name and slightly different approach have been working in nvidia cards for two full generations now.

Fast forward to 2017 and you have a chip that is on paper a brute as it is in terms of clock, shader count and memory bandwidth/latency. And then you enable TBR on it properly to render games. What do you people expect? 5-10fps boost only? C'mon, get real. Maxwell 2 wasn't just magically faster, it was because of TBR along with few other similar power/performance technologies (like framebuffer compression).

Fast forward to 2017 and we have a GPU that still isn't released, on paper offers performance similar to 1080Ti yet what we do have out is trading blows with the 1070. Somewhat similar to the situation we have with the RX480/RX580. On paper they are not far off of the 1070 in performance, yet here we are years later looking at a card that still trades blows with the 1060.

On paper AMD releases cards that can destroy and in compute heavy tasks they do exactly that. There is a reason the RX480 is so popular in mining that is a task they happily compete with the 1070 (and 1080 for that matter) in. The Vega cards (fury 2) once they have drivers and miners that take advantage will do exactly the same. It is a quite wide GPU with tons of potential. Yet again however the disappointing driver writers at AMD disappoint, not to mention the gaming models STILL AREN'T OUT. I mean seriously is this some kind of mean joke? It is July, the competition has been out so long that they have released cards with better memory just because.

I hope and pray this card does what I hypothesized in December when I wanted to buy one of them. Trade blows in DX12/Vulkan with the 1080Ti and then trade blows with the 1080 in DX11/OGL. I would happily buy into that as the multi-GPU support on AMD is better as of late and I hate myself enough to always run multi-GPU setups.

 

[bug]

Oh my god, this is why I get rage episodes. We know Vega FE drivers are totally immature, we also know Tile-based Rendering doesn't work. AMD released Vega FE with drivers in such state because it was naturally assumed it would be used in productivity tools, not benched in freaking games. And yet, here we are with people making gaming performance assumptions on a card with totally immature drivers which was meant for productivity first and gaming second. AMD was simpy hoping to release Vega FE as is and fix the gaming stuff later since it's not exactly meant for that. And people are going batshit crazy over it. I watched PCPer stream and I otherwise respect the guys, they generally know their shit, but most of the stream and also their later recap, I was like, guys, what the hell are you doing... It's like they threw all logic, common sense and entire existing knowledge about graphic cards, drivers and gfx card releases out the window the moment they got Vega FE in their hands.

That's about as insane as whining over Lamborghini Aventador not being all that great and harvesting wheat. But Lamborghini also makes tractors! It should perform great at that too!

Seriously, mister "no baseless assumptions", how much faster do you think the consumer Vega will be? Because I don't have a crystal ball, I'm just going by the available numbers.

 

[deu]

So, AMD's latest and greatest GPU with the big d and lots of power and heat beats NVIDIA's year old GPU (Founders Edition of course) with the little die by a whopping 5% and is significantly slower than the 1080 Ti with the big die.

Wow, take my money AMD!!

Now, again, before all you AMD apologists start foaming at the mouth at me for being "anti-AMD" and an "NVIDIA fanboy", you should instead be annoyed with AMD for continuing to put out disappointing products, not the guy (me) pointing out their failings.

For the record, I would have loved Vega to leapfrog the performance of the 1080 Ti and make NVIDIA play catch up for a change. That's real competition and results in better products for us at lower prices.

Stop trolling. Anyone that have the slightest idea of how a GPU works would not claim that RX VEGA is out or not understanding that there is such a thing called drivers bios etc. The physical RX VEGA is done, but the end specs, the drivers, the bios, the specific gameoptimizations is not. If so AMD would be the first in history to pull that of 1 month before launch. So in other words; You assume alot (negatively) about VEGA that there is not rational argument for. To me you're combining troll/fanboism and poor understanding into a classic shitpost. Some might attack you because of fanboism but alot of people just cant take people saying stupid shit with any proof/information or an educated argument. RejZoR is pretty much spot on in his argument or prediction to the level you should get at. VEGA is a step in AMDs longterm strategy; not a "ok we'll just bind and boost" Says the dude that have had NVIDIA for 10 years but is tired of shitposters that does not add arguments or insight to the discussion.

 

[bug]

Stop trolling. Anyone that have the slightest idea of how a GPU works would not claim that RX VEGA is out or not understanding that there is such a thing called drivers bios etc. The physical RX VEGA is done, but the end specs, the drivers, the bios, the specific gameoptimizations is not. If so AMD would be the first in history to pull that of 1 month before launch. So in other words; You assume alot (negatively) about VEGA that there is not rational argument for. To me you're combining troll/fanboism and poor understanding into a classic shitpost. Some might attack you because of fanboism but alot of people just cant take people saying stupid shit with any proof/information or an educated argument. RejZoR is pretty much spot on in his argument or prediction to the level you should get at. VEGA is a step in AMDs longterm strategy; not a "ok we'll just bind and boost" Says the dude that have had NVIDIA for 10 years but is tired of shitposters that does not add arguments or insight to the discussion.

See my question above. How much faster do you think RX Vega will get between now and launch date? How much faster do you think it will get over its lifetime?
I think 25% over its lifetime is a very optimistic expectation and even that doesn't make it faster than 1080Ti.

 

[ratirt]

See my question above. How much faster do you think RX Vega will get between now and launch date? How much faster do you think it will get over its lifetime?
I think 25% over its lifetime is a very optimistic expectation and even that doesn't make it faster than 1080Ti.

Ok 1080 TI is and will always be the fastest. Happy?
Nobody here is telling how much faster Vega will be over 1080 TI. Because nobody knows how fast it is now since it's not here yet. People here just share thoughts and suggest something what they have noticed or came up with. Please go to NV threads.