Radeon RX Vega Benchmark Shows 3DMark 11 Performance just above GTX 1080

In order for Vega to be like Ryzen it would have to have comparable performance to a 1080 but at half or third the price. Simply put, it's not going to happen (large die + HBM2 makes it impossible).

As for mining, it all depends on the hash rate and power draw. If it consumes more power than a 1080 then I doubt miners would touch it. Also, Nvidia cards hold the advantage in ETH mining over time, which might work in favor of Vega for gamers (here's hoping).

 

Wouldn't the obvious solution be to sell two lines of GPU's, one gaming and mining compatible and one only gaming compatible?

 

Unfortunately that's not how mining works. It's easier to make cards for 'mining-only' (e.g. remove video out) than to do it the other way around. The crypto coins (in their multitude) have largely been developed to take advantage of the technology and resources that can be used for gaming/compute. There's no extras added to make a 'mining card'.

 

Although ASICs specifically developed for mining are much more efficient than GPUs. FPGAs have been used in the past as well. The only way to make people stop buying GPUs for mining is to make something that mines better.

 

I'm really most interested in the price. Performance wise, this has been expected for a long time now. However, considering the production cost on this is quite high, and to add the mining craze atop of that, I don't see it having "best bang for buck" price as many have dreamed of for like 1 year now. Better late than never I guess...

 

ASICs are not productive to mine Ethereum. Also Ethereum has no such limits on total currency amount as BTC have.
The only real thing that could end GPU mining rush now is either Ethereum moving to proof of stake or cryptocurrency value dropping below electricity running costs.. Probably neither of which will happen at least for another 3 to 6 month period I think.

To me Vega looks like mirage now AMD Mirage Its not yet released, HBM is still scarce (or is it not?) and considering miners bought RX series even for double the price whos to say they are not eyeing Vega now. At least the early reports of Vega ETH hashrate center at around 30-35 Mh/s. Its funny when for the first time a card can be to powerful and thus not reach gamers Still its early and even such hashrate is like 10x profitable (depending on electricity costs) compared to Vega TDP.

Will miners wait in line at AMD factory door like people for new iPhone or will gamer guy finally be able to upgrade his GTX 1050 without selling his car to afford a better card? We will see in the next episode

 

I get that and it makes sense but could something be added to the cards to make them mining incompatible or maybe just cripple mining, bios block or anything?

 

Yeah, why one earth would any business sell a product that you keep having to buy the next iteration in a short interval... Terrible for business..... /sarcasm

 

If the people who do GPU mining on a large scale take a shine to this new AMD card , i suppose us gamers will end up paying a inflated price should you want one , and that's if there any stock available .

 

I'll wait and see real reviews with real world benchmarks instead of guessing how it works from random 3dmark benchmarks with no basis to say how it actually was tested.

 

I want to see how these things perform on the workloads that the GPUs were designed for like deep learning and neural network training.

 

And what would be the benefit for AMD/nVidia? None, only losses.

 

Easy to simulate, just run some older games at minimum settings and you'll have hundreds of frames per second to look for screentearing with.

How is it ever better to have someone say "you'll have A LOT of screentearing, I couldn't even play and bought a new monitor with freesync just so I could have my eyes not die" - which we know is the default answer from most people. Or "I didn't notice any screentearing" ... what does any of those subjective answers mean in this? I mean, I have a freesync monitor and I don't think I can even tell the difference with it on or off... except that on seems to cause flickering in some titles.

 

GURU3D 1080 3DMARK11 SCORES 24867 http://www.guru3d.com/articles_pages/nvidia_geforce_gtx_1080_review,27.html
RX VEGA SCORES 31875

1080TI + 7700K SCORES 31172 http://www.3dmark.com/hall-of-fame-2/3dmark+11+3dmark+score+performance+preset/version+1.0.132/1+gpu
THIS IS HALL OF FAME, THE BEST FROM THE BEST SCORES, IT"S NOT TOP 10000 USERS WHO CLAIMED 3DMARK RESULTS, THIS IS TOP 100 WORLDWIDE.

24867 + 28% = 31829 = RX VEGA = 1080 TI + 7700K

 

Are you comparing Pscore with GPU score? The 1080 TI from the HOF link you gave has almost 45000 gpu score.

 

yeah, what I was going to say. HH clearly referring to 'GPU score' in benches but the guy is linking to a P score bench chart for 1080.

 

Just wake up ? We are talking about GPU score , you are talking about P score ... Here mine 1987/1389 33k 3DMARK SCORE

So yeah Vega is behind a 1080 gtx atm VEGA SCORE . And a 1080ti = ~45k GPU score you can't compare with 3DM SCORE .

 

Please do your research gurus. Let's not turn this into the youtube comment sections.

 

Oddly enough, I'm still extremely intrigued about finally getting the real lowdown on the final RX Vega cards. Despite the Vega FE debacle, RX Vega is still a real enigma. Some extremely important questions remain.

1.) What hardware tweaks and / tiers of Vega cards will we see from the RX versions? Different HBM2 clocks / configs, different cooling configs, GPU clock/throttling differences, etc.

2.)Drivers. Drivers. Drivers. Many naysayers say theres no way you can gain more than 10-20% performance with driver revisions. The reality is that completely depends on the status of the driver you are comparing it to. Proper drivers can easily achieve massive performance gains over poorly optimized ones. Remember the difference between using Glide or DirectX for the old Voodoo2 cards? Glide was 100% faster in many cases.

Vega FE seems like a rushed money grab / attempt at keeping a CEO's promise, putting out a nice hardware product before the software is ready. Yes, it functions-- its functional, but is it at full potential for gaming APIs? Not according to AMD, and I believe them-- because if it was, there would be no point in delaying the RX launch. They apparently believe they are leaving some real gains on the table because of unoptimized drivers, and are scrambling to get them written. The big question is, what kind of gains are we looking at- 5-10%, 15-20%, or 30-45%? This is, according to AMD, a brand new uArch, and the largest and most complex GPU they have built to date. Optimizing drivers for such hardware may be more complicated than many believe it to be.

3.) Price points. This is the big kicker and depends entirely on 1 and 2 above. If its competitive with the 1080 at $450 or less, that's going to be big. If they have multiple versions, will they take the place of the top Polaris cards or slot right above?

Its all quite intriguing.

 

This is what I could dig up on pricing: RX Vega Core ($399), RX Vega Eclipse ($499) and RX Vega Nova ($599). So its meant to compete with Nvidia GTX 1070, 1080 and 1080ti respectively.