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

[Raevenlord]

New benchmarks of an RX Vega engineering sample video card have surfaced. There have been quite a few benchmarks for this card already, which manifests with the 687F:C1 identifier. The new, GTX 1080 beating benchmark (Gaming X version, so a factory overclocked one) comes courtesy of 3D Mark 11, with the 687F:C1 RX Vega delivering 31,873 points in its latest appearance (versus 27,890 in its first). 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.





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.



View at TechPowerUp Main Site

 

[cdawall]

Sweet it barely edges out a year old gpu. How's it compete with the faster ram model?

 

[RejZoR]

That's actually a very good result considering a 300MHz difference in core...

 

[ratirt]

That's good news right? Improvements 5%per month. Nice I wonder how long they can keep going with this. I'm sure they finally will hit the wall with performance boost over the driver improvement.

BTW. I wonder if there will be any OC potential. If this Vega hits 1800, it might get interesting.

 

[okidna]

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.

I think you misunderstood the Videocardz statement because all of those 687F:C1 benchmark links on the chart dated July 4, 2017.
I think they meant another benchmark score when they wrote "compared to some older benchmark results (~3 months old) ", certainly not one of those 687F:C1 benchmark links listed.

Lowest score :


Highest score :

 

[RejZoR]

That's good news right? Improvements 5%per month. Nice I wonder how long they can keep going with this. I'm sure they finally will hit the wall with performance boost over the driver improvement.

BTW. I wonder if there will be any OC potential. If this Vega hits 1800, it might get interesting.

I don't think 1800MHz is likely. It's a different design than NVIDIA's GPU's, seeing it at 1600MHz is already quite high. Functional tiling might help since it would stop overdrawing same things over and over, making it more power efficient and potentially more keen to overclocking. But again, unlikely imo.

 

[dj-electric]

Here's the deal.

Let's say stock vs stock an RX Vega is 5% faster than a GTX 1080.

14 months late, with a TDP of 275W or higher compered to 180W, and at about the same price... just doesnt do justice.

The only way i could find RX Vega attractive is if it would actually be 100$ cheaper. There's nothing else really redeemable about this product. I don't even wanna imagine OC capabilities between the two, knowing how more than decent the GTX 1080 is

 

[ratirt]

I don't think 1800MHz is likely. It's a different design than NVIDIA's GPU's, seeing it at 1600MHz is already quite high. Functional tiling might help since it would stop overdrawing same things over and over, making it more power efficient and potentially more keen to overclocking. But again, unlikely imo.

Well I was hoping you'd say yeah . Since it is not NV we don't know how far we can push it. Although it's just my wishing for it to run1800Mhz. Chances for this are practically none. Well who knows. Just saying it would be interesting.

Here's the deal.

Let's say stock vs stock an RX Vega is 5% faster than a GTX 1080.

14 months late, with a TDP of 275W or higher compered to 180W, and at about the same price... just doesnt do justice.

The only way i could find RX Vega attractive is if it would actually be 100$ cheaper. There's nothing else really redeemable about this product. I don't even wanna imagine OC capabilities between the two, knowing how more than decent the GTX 1080 is

if Vega will be 100$ cheaper than 1080 I'm going for pre-order If I was sure that price difference would be 100$ between vega and 1080 i'd buy the 4k screen today

 

[bug]

if Vega will be 100$ cheaper than 1080 I'm going for pre-order If I was sure that price difference would be 100$ between vega and 1080 i'd buy the 4k screen today

4k on GTX1080 is spotty at best. 5% won't change that.

 

[FordGT90Concept]

3DMark11? Seriously? Why run anything other than TimeSpy and Firestrike?

 

[Kommanche]

4k on GTX1080 is spotty at best. 5% won't change that.

As someone with a 4K screen and a GTX1080, it's fine. Almost all games run Very High/60FPS.

 

[Aenra]

Sweet it barely edges out a year old gpu. How's it compete with the faster ram model?

It's getting tiring.. you were the number 1 prospective buyer, it was delayed, you got bitter, you got even more bitter.
You bought a Ti, and ever since doing that, you've switched camps and gone full bashing/negativity; for months now. Is it not time to cease?

If i've got your point/emotional response to it all, i'd bet you the oldies here have as well, sooner than i did too.
..We get it. Honestly

 

[Toothless]

As someone with a 4K screen and a GTX1080, it's fine. Almost all games run Very High/60FPS.

Only very high? Not ultra?

 

[bug]

It's getting tiring.. you were the number 1 prospective buyer, it was delayed, you got bitter, you got even more bitter.
You bought a Ti, and ever since doing that, you've switched camps and gone full bashing/negativity; for months now. Is it not time to cease?

If i've got your point/emotional response to it all, i'd bet you the oldies here have as well, sooner than i did too.
..We get it. Honestly

He's just calling a duck a duck.
What would you be posting if Nvidia released this year something 5% faster than RX480?

@Kommanche According to reviews here on TPU, 1080 doesn't push an average of 60 fps (and exactly 60 fps average is not that great since it means half the time you're not getting 60fps). If you say it's better in real life, that's great.

 

[T1beriu]

Classic clickbait title with wrong information that you missunderstood. Again and again and again in the last months when you talk about rumours and leaks.

I'm pretty sure you guys do this on purpose, turning yourself slowly into a tabloid.

 

[R0H1T]

That's good news right? Improvements 5%per month. Nice I wonder how long they can keep going with this. I'm sure they finally will hit the wall with performance boost over the driver improvement.

BTW. I wonder if there will be any OC potential. If this Vega hits 1800, it might get interesting.

The real benefits of Vega will not be seen until after Scorpio & PS5 are released, it's going to be like HD 7970 or R9 290 (i.e. 390) all over again.

Also 1800 boost clocks sound impressive but fake, they'd have to move away a lot from GCN right up until Polaris & I don't see that happening without some major compromise, perhaps in GPGPU?

 

[RejZoR]

It's so funny when people constantly and repeatedly drag out the "but it beats a 1 year old product". Like it matters. If RX Vega was 800€ and it barrely beat GTX 1080 a year later, you'd be wondering that. But clearly, it's no going to cost 800€. So, who gives a damn if it's 1 year after GTX 1080. During RX400 generation, AMD didn't even have ANYTHING in top end segment. And everything was as usual, exept you simply didn't look at AMD if you wanted top end. They were still selling mid and low end just fine. It doesn't matter when you release it, if it performs well and has a reasonable price, you'll always get people who will buy it. Same happened when AMD had bad models of Radeons and same applies to NVIDIA which sold when they had bad models.

 

[john_]

All scores have the same date.
All scores have the same driver.
Top score was archived on an Intel system.
The other scores used a Ryzen 1800X, some scores with turbo at 4.0GHz, some with turbo at 3.7GHz.

 

[ratirt]

The real benefits of Vega will not be seen until after Scorpio & PS5 are released, it's going to be like HD 7970 or R9 290 (i.e. 390) all over again.

Also 1800 boost clocks sound impressive but fake, they'd have to move away a lot from GCN right up until Polaris & I don't see that happening without some major compromise, perhaps in GPGPU?

It was a wishful thinking not information. Stop picking the words and create your own fake understanding. We don't talk about Polaris but Vega and we don't know the potential. Probably it wont hit 1800Mhz. It would be interesting if it did.
Read carefully next time please.

 

[bug]

It's so funny when people constantly and repeatedly drag out the "but it beats a 1 year old product". Like it matters. If RX Vega was 800€ and it barrely beat GTX 1080 a year later, you'd be wondering that. But clearly, it's no going to cost 800€. So, who gives a damn if it's 1 year after GTX 1080. During RX400 generation, AMD didn't even have ANYTHING in top end segment. And everything was as usual, exept you simply didn't look at AMD if you wanted top end. They were still selling mid and low end just fine. It doesn't matter when you release it, if it performs well and has a reasonable price, you'll always get people who will buy it. Same happened when AMD had bad models of Radeons and same applies to NVIDIA which sold when they had bad models.

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.

 

[EarthDog]

That's actually a very good result considering a 300MHz difference in core...

like we could ever compare by core speed directly in the first place???

Yet you already know that...

I don't think 1800MHz is likely. It's a different design than NVIDIA's GPU's, seeing it at 1600MHz is already quite high. Functional tiling might help since it would stop overdrawing same things over and over, making it more power efficient and potentially more keen to overclocking. But again, unlikely imo.

3DMark11? Seriously? Why run anything other than TimeSpy and Firestrike?

+1

In my experience, the gaps between cards grow as you get to 'heavier' benchmarks past 3d11. 3d11 is fairly cpu heavy and cuts off some extremes due to how it tests (lower res, textures, etc)

 

[RejZoR]

Of course it needs more power. It's not yet using tiling. I'm shocked how people still don't get it. Tile based rendering is the essence of efficiency, it's why it's used on mobile devices since forever. It's also why Maxwell/Pascal cards are so efficient.

 

[Filip Georgievski]

At this point im just thinking is this the best VEGA they have?
And if it is priced at 600$ it will be too expensive for my taste, and this is comming from an AMD GPU Fan (not fanboy).
You guys do remember that the R9 FURY X in DX12 is also neck a neck to the 1080 and VEGA barely beats the 1080?

Price wise this would not be a smart move since you can find a FURY X as low as 350$ and it is maybe 5 - 8% slower than the VEGA at almost same TDP ( FURY X has 300W+ TDP)?

I hope AMD release a better VEGA chip since this one is not any better than the FURY X.

 

[Octopuss]

It's so funny when people constantly and repeatedly drag out the "but it beats a 1 year old product". Like it matters. If RX Vega was 800€ and it barrely beat GTX 1080 a year later, you'd be wondering that. But clearly, it's no going to cost 800€. So, who gives a damn if it's 1 year after GTX 1080. During RX400 generation, AMD didn't even have ANYTHING in top end segment. And everything was as usual, exept you simply didn't look at AMD if you wanted top end. They were still selling mid and low end just fine. It doesn't matter when you release it, if it performs well and has a reasonable price, you'll always get people who will buy it. Same happened when AMD had bad models of Radeons and same applies to NVIDIA which sold when they had bad models.

As much as I agree with most of your posts AND you sound like you know what are you talking about, you really sound like a die hard AMD fanboy in the last several weeks

 

[bug]

As much as I agree with most of your posts AND you sound like you know what are you talking about, you really sound like a die hard AMD fanboy in the last several weeks

I don't get this attitude. I mean, I lean towards Nvidia, but that doesn't mean I don't automatically give them the benefit of the doubt on everything. I could, but it's not like they give me a beer in return or something, so what's the point?