Possible Radeon RX Vega 3DMark Time Spy Benchmark Result

It's 25tflops of half precision, 12.5Tflops of single.

 

Because people are stupid.

 

We know it's a Vega 10 ES, that's undeniable. We just don't know what the final product will be, and how many versions there will be. My guess is 4gb and 8gb initially.

Clocks/drivers could be subject to change and boost performance.

 

I will wait for Hilbert's review to see where the AMD Big Vega stands before i make a decision to buy a new video card , Big Vega or wait for Volta ...

 

I recall there being as many as 4 variations of Vega.

 

maybe its the planned anti hype so when they finally unveil it people get excited over 95% same perf as nvidia's offer at 20-30% less money

 

above bench not vega ,, not vega

vega is 1600mhz and 12gig memory ,it is new pro due base on rx480 extremetech.com/wp-content/uploads/2017/04/ProDuoFeature-640x353.jpg

 

Honestly, is anyone expecting anything more than that? I don't see how they could be after the last few generations of cards. That Doom demo basically showed Vega was on par with the GTX 1080. Even if they could squeak out a little more performance, the 1080 just got a memory upgrade that bumped it up a bit, so it seems like the only way they can be competitive is to come in cheap.

 

Do you not know how to read?

Proof, or stop throwing around BS.

Seriously, do you not know how to read?

Now, if you want to say "I believe this is the pro duo based on rx480", that's fine, but that's not what you're saying.

Also, 12gb? where do you get that number?

That number doesn't make sense when it comes to HBM



The only way that statement would make sense, is if each of those stacks are 6GB, which could be true, but i have only heard of HBM2 stacks in 4GB and 8GB, so your statement doesn't make sense.

If you're going to proclaim something, BACK IT UP.

Pro Duo also has 7Gbp/s GDDR5 memory, which would be (i believe unless i am doing it wrong) 1750Mhz, yet this reading states 700Mhz. 700Mhz makes sense for HBM2 memory, but makes absolutely no sense for GDDR5 memory.

And lastly, the Pro Duo has 32GB of memory, which is not 8GB that is being reported here. What do you not understand about this?

 

If it's an engineering sample clocked far below its to-be-released clock, there's nothing at all "disappointing" about it... It may not even be "Vega" at all in any form. Out of curiosity, do people bother to *read* the accompanying text HH takes the trouble to *write* in these rumored, highly speculative articles? Eh? Do you read, or do you find that those ugly squiggles just get in the way of the pretty bar charts? Try reading for fun and profit--it works...

 

If it was "Vega" it was "Vega" pre-engineering, pre-market, compared to a mature 1080. Ergo = no comparison, "Vega" blows it away. The last few generations of cards? What does that mean, I wonder. If you should ever think, per chance, that a benchmark bar chart tells you everything you need to know about a given GPU then you need to stop right there because you've been had--and reeducate yourself because you've missed 90% of the content somewhere along with way. Benchmarks tell but a fraction of what an intelligent consumer needs to know about a GPU before he buys it. You should check out gaming forums globally--they *all* offer mute testament to that fact--especially where laptops are concerned.

Personally, to answer your question, yes...I always expect more from AMD than I do from nvidia--ALWAYS, and that doesn't necessarily encompass frame-rate at all--but many more factors--such as backwards-compatibility with my HUGE library of games, many *SPAM* back 30 years or more. At any rate, he who looks at frame-rate bar-charts and nothing else when considering a 3d card is a chump, pure and simple, as is he who believes he can discern the difference between playing a game at 100fps and the same game at 120fps without the aid of a frame-rate benchmark to tell him the difference... But the world is full of chumps--I suppose--and their money, too, keeps the wheels of commerce churning.

I don't mean to hone in on you, personally--it's more your attitude--"Honestly, is anyone expecting anything more than that?" OF COURSE, they are expecting more than that. Open your eyes and start learning. HH doesn't publish the many-paged reviews he does just so people can ooh and ah at the pretty pictures! Learn how to read, guy...

(OK, enough from me playing GrandPa Lecture Head)>>>

 

That one is clearly Vega, but likely indeed some sort of engineering sample.

That said, I have this suspicion they have major issues with clock speeds and heat there - looking at Polaris - the best they managed was about 1500MHz with terrible power efficiency on a meager shader count.

I don't think Vega is so massively different from Polaris, it is probably a step up kind of thing carrying over Polaris deficiencies, so at best I expect around same clocks if that. Certainly won't be 1200, but pretty sure about 1500 if that.

 

Looking forward to reading the reviews when Vega launches, but I fully expect to be mildly disappointed. I'm on 1440p at the moment, but intend to move to 4K soon(ish), and there's just no way I'm buying a card which doesn't perform within a few hairsbreadths of the 1080 Ti.

 

Did you guys see this post from OnnA:

Spoiler

That looks like BIG Vega and does compete with Titan X Pascal with the same CPU. Not sure if photoshop or leak but that is just freaking amazing if it is BIG Vega.

 

It's not that terrible the power efficiency. It would be most likely roughly the same as 1060 if it had the things they are introducing with Vega that nvidia brought with maxwell. It's ok but not good.

I think the clocks should be around 1500 if the deeplearning card is anything to go by. It should be a bigger step up from polaris/fiji which where closer to each other.

 

That I could get behind of... That would be right next to 1080 ti. And would be more in line what we could expect from 12+ tflop card.

 

Number seems more realistic but until it's a link and not a picture - I'm not convinced it's real. I do think the card will fall around that though.

 

Same here and I am not saying it IS, but HOPING it will be Big Vega... :nerd:

 

You're making a positive judgement about a card that hasn't been reviewed yet, and then you imply I'm the one who's being simplistic in my thinking... :wanker:

Go back and look at the post I was quoting, and then consider my comment in context. AMD over the last few generations, has only been able to offer similar performance for a little less money. They have not managed to offer a card with significantly greater performance-per-dollar when compared against products of the same class from Nvidia. A big part of that seems to be that they're always playing catch-up to the standard that Nvidia has already set, and Nvidia always seems ready to counter within weeks.

Based on that, I think it would be unreasonable for anyone to expect AMD is going to have a product that is capable of significantly out-performing Nvidia's. It seems much more likely they will have cards that offer similar levels of performance at lower prices, and Nvidia will respond by dropping price on their existing line.

We're talking about GPUs. Their market value is measured largely in benchmarks. Any review of a GPU is ultimately there to offer context and additional information, but the numbers tell most of the story. Heat, power draw, price, relative performance, overclockability, and noise - those are the standards by which we judge gaming GPUs. And relative performance is the most important of those factors. Frame rates largely determine how much the manufacturer can charge for their card.

How many games do you have that will run on a modern generation AMD card, but not on a modern generation Nvidia card?

I don't know a single person who looks only at frame rate charts, nor did I suggest anyone should. But frame rate charts are certainly one of the most important factors when considering buying a card. Would you make a purchasing decision without looking at frame rate benchmarks?

I read just fine, and the reviews tell the same story consistently for the last few years - AMD at their best manages to offer cards that are close to the performance of Nvidia's, often for a little bit cheaper price. Nvidia then responds by adjusting price, or releasing an additional product that nullifies any price-to-performance advantage AMD might have had. We saw this with the 290x, we saw it with the Fury and Fury X, and again with the RX 480.