AMD "Vega 10" Bears Core-Config Similarities to "Fiji"

[Nokiron]

False. OCing ram on the 1080 gives framerate increases.

Synthetic bench lololololol

Ofcourse it does, but how much? I'll bet you see a gigantic performance increase by adjusting the core-clocks instead of wasting power constraints and boost clocks by overclocking the RAM.

I have yet to see anything but speculation from your side.

 

[Captain_Tom]

False. OCing ram on the 1080 gives framerate increases.

Synthetic bench lololololol

It never ceases to amaze me how often people think RAM doesn't help lol. If you look at a chart of TFLOPS/Bandwidth over time you would see that ALL modern GPU's are bandwidth starved. In fact the Fury X was the first GPU in a LONG time to not be completely hamstrung by bandwidth.


I have a lot of experience overclocking all types of GPU's: 560 Ti, 6950, 7970, 7950, 6850, 6870, 650 Ti, 390X, Fury, and on and on!!! Every single card gained just as much, and usually more performance if you overclocked its RAM.

Ofcourse it does, but how much? I'll bet you see a gigantic performance increase by adjusting the core-clocks instead of wasting power constraints and boost clocks by overclocking the RAM.

I have yet to see anything but speculation from your side.

How much? My 7970 gained more from overclocking ram than from overclocking its core. That's how much.

 

[Nokiron]

How much? My 7970 gained more from overclocking ram than from overclocking its core. That's how much.

Great for you, my four 7970s didn't. They much preferred the higher core clock.

Again, show me benchmarks that compares it and shows the benefit. I have yet to see one.

Found one from Anandtech, have not found any other yet. But this speaks volumes, a card with quite a small bus width gets minimal gains by overclocking the VRAM. (From 7Ghz to 7.8Ghz)
Memory bandwidth is the least of it's problems.

 

[Vayra86]

You would need a Fury (or Tesla) with gddr5 to test...
And we're going to ignore how fury x perf doesn't tank when you GBs over on VRAM usage, eh?

No, but then the Fury also had just 4 GB to work with, so it HAD to not tank when you want to use more than that. Again, the *added advantage* of using HBM on the Fury is nonexistant, because the core can't go fast enough. So you can whine about Nvidia's tight VRAM budget (which is true, I'm not denying that, you really need to let go of dropping the fanboy-bomb everywhere you go), but you need A to use B. And the fact still is that AMD's A was too weak to saturate B, the Fury X was a card with as much inbalance (overcapacity) on the VRAM as the Nvidia 1080 has overcapacity on the core.

The funny thing about all this, is that *across the board* on a large benchmark suite it is STILL beneficial to have a 'too strong' core than it is to have too much bandwidth. Can you have too much bandwidth? Yes, because high bandwidth translates into an efficiency drop, and efficiency is TDP, and the more of that budget you can reserve for the GPU core, the more performance you can extract from it.

So you can harp on all you want about Nvidia's tight VRAM, as long as the GPU still is king of the road, which it is by a margin of about 50-60% at this point versus AMD's offerings, that is what counts. Not to mention the fact that only a portion of the game engines lean heavily on VRAM, while all game engines want a fast core. Nvidia's simply got a better balance going on than AMD, and this has been the case since Kepler.

Again, the resulting performance is what matters, all the rest is irrelevant, unless you are looking at very specific engines and situations, such as 4K gaming where the Fury X excelled, but still couldn't really beat a 980ti because the latter could get much better OC - a direct result of using a tight VRAM bus.

 

[Casecutter]

If we see 4k much stronger than a GTX 1080, while can spar in close against GTX 1080Ti 11Gb all with a price of $600 it's going to be a good value. If it's smaller, while near perf/watt of GP102 it just get better.

 

[warrior420]

Just because it has a same arrangement of shaders, TMU and ROP units it doesn't mean they are the same. It's also not just Fiji with higher clock and smaller node. Vega has a new scheduler, new pixel and vertex engine, tile based rasterizer, new more flexible compute unit, new memory controller...

That's about the same absurd as saying Volkswagen Up! with 1.0 engine is the same as Mitsubishi Lancer Evo X with a 2.0 Turbo engine, just because they both have 4 cylinders.

Just to be an asshole....

The VW Up! 1.0L was an I3... 3 cylinders.

Your logic is flawed. The point is definitely there....But, flawed nonetheless.

On Topic: This article is trash. Please take it down...

 

[owen10578]

Hahaha i love how every AMD related news ends up with fanboys fighting nonsensical stuff. AMD fanboys are the most obnoxious I swear.

On topic: Although its somewhat interesting I couldn't care less about it. I just want to know the performance. If its got the performance then sign me up for Vega.

 

[mrthanhnguyen]

If Vega trades blow with 1080ti and cost half, consumers win. If its $20-$30 cheaper, I don't think many people will buy it.

 

[owen10578]

If Vega trades blow with 1080ti and cost half, consumers win. If its $20-$30 cheaper, I don't think many people will buy it.

Why the hell would you expect AMD to price it half when they can price it the same or close by?

 

[the54thvoid]

If Vega trades blow with 1080ti and cost half, consumers win. If its $20-$30 cheaper, I don't think many people will buy it.

Are you nuts? The Fury X was the same price as the 980ti. Why would they sell a direct competitor at half price? I'm sure if Ryzen was a 5Ghz clocker it would have been more expensive. But that's a digression, AMD are not in the business of selling top grade, cutting edge technology at bargain basement prices. It'd commercially kill them.

Why the hell would you expect AMD to price it half when they can price it the same or close by?

Beat me to it.....

 

[T4C Fantasy]

IF and IF....

IF and ELSE IF

 

[TheoneandonlyMrK]

Sounds good but why are bears involved , surely its bares and that's from a frequently corrected gramma guy.

 

[_larry]

Hopefully they will figure out how to make it more efficient...
My R9 290 is super power hungry if I even try to overclock it a little.
They need to focus more on the core architecture. Memory bandwidth and speed is not currently an issue even at 4k resolutions for most games.

 

[TheoneandonlyMrK]

Hopefully they will figure out how to make it more efficient...
My R9 290 is super power hungry if I even try to overclock it a little.
They need to focus more on the core architecture. Memory bandwidth and speed is not currently an issue even at 4k resolutions for most games.

They did that with Polaris ,i swapped from one r9 390 to two 480s for less ,or similar power if ocd yet similar performance.
And at 4k with two cards id like more speed and memory bandwidth please.

Shareing similar numbers of inner hardware is both obvious since its their Ip they are building upon and have researched the optimisation of to be able to advance it.
You wouldn't scrap all that went into making any other device and start totally from scratch.
And if a cinfig works well for you why change.

 

[ensabrenoir]

It's a trap... AMD's future won't change if they dont figure out a better pricing model.1080, 1080ti performance for half its price would be stupidity. Even ryzen will hurt them if the don't sell a metric ton of them and thats the key. Got to have massive volume with low prices or your still just bleeding slowly.....

 

[thesmokingman]

If Vega trades blow with 1080ti and cost half, consumers win. If its $20-$30 cheaper, I don't think many people will buy it.

And the 1080ti trades blows with the Titan X for nearly half the cost... so AMD should sell theirs for a quarter the cost? For the win?

 

[TheGuruStud]

Ofcourse it does, but how much? I'll bet you see a gigantic performance increase by adjusting the core-clocks instead of wasting power constraints and boost clocks by overclocking the RAM.

I have yet to see anything but speculation from your side.

We already know what OCing core on 1080 does...and it's not spectacular. Without mem OC you're missing a few percent (about 5 if you have good chips). The only thing wasting boost clocks is nvidia's artificial limitation. OCing mem brings the perf increase up to near parity with the percentage of core increase.

So...guess what? Faster memory is needed or it won't scale.

 

[RejZoR]

If Vega trades blow with 1080ti and cost half, consumers win. If its $20-$30 cheaper, I don't think many people will buy it.

I was waiting for someone to show up with an idea that a perfectly competing product should cost half the price for no logical reason. Dude, this isn't communism where free shit just happens until it runs out of free shit. Business needs to make profits and by making shit free for no reason doesn't make profits. When will you people get it in your heads? They can make product tiny bit cheaper to make it more attractive, but expecting same performance and half the price just because it's an AMD, sorry, that's just an insult to intelligence.

 

[ZoneDymo]

I was waiting for someone to show up with an idea that a perfectly competing product should cost half the price for no logical reason. Dude, this isn't communism where free shit just happens until it runs out of free shit. Business needs to make profits and by making shit free for no reason doesn't make profits. When will you people get it in your heads? They can make product tiny bit cheaper to make it more attractive, but expecting same performance and half the price just because it's an AMD, sorry, that's just an insult to intelligence.

Well you insistence to call half of 700 dollars "free" is also quite the insult to intelligence...
Same for the communism remark.

If 2 products offer the same, you got to do something to make 1 more attractive so people will go for that one instead.
A lower price point would help a lot, now im not saying half would be the smart move as that means they are heavily competing with themselves, but if they are 200 dollars under Nvidia's equivalent, who would ever buy the Nvidia card?

You all seem to think its Ferrari that makes the most money selling 5 cars for 1 million and not Ford selling 5 million cars for 10.000.
Selling tons more at a lower price is much better then selling a few at a higher price.

 

[RejZoR]

Why do you think price is the only factor? Ever heard of FreeSync? It's a major factor recently when people are picking monitors. And guess who (ONLY) supports it? I also totally dig the Wattman idea. I hate using MSI Afterburner or similar tools. I miss the days when NVIDIA also had integrated overclocking capability. And there are bunch of other tiny things that convince people into using Radeon over GeForce. At same price.

 

[Captain_Tom]

Great for you, my four 7970s didn't. They much preferred the higher core clock.

Again, show me benchmarks that compares it and shows the benefit. I have yet to see one.

Found one from Anandtech, have not found any other yet. But this speaks volumes, a card with quite a small bus width gets minimal gains by overclocking the VRAM. (From 7Ghz to 7.8Ghz)
Memory bandwidth is the least of it's problems.

I set up darkcoin mining rigs, and all of mine did.

Frankly, I don't believe you did any experimental testing, and I doubt you even understand what you are talking about.

Your chart suggests memory made a major difference! If you can't see your own chart supports my argument, talking to you is a waste of time.

 

[m1dg3t]

Hahaha i love how every AMD related news ends up with fanboys fighting nonsensical stuff. AMD fanboys are the most obnoxious I swear.

On topic: Although its somewhat interesting I couldn't care less about it. I just want to know the performance. If its got the performance then sign me up for Vega.

Mostly what I see across the web is nVidia fantrolls flaming every. Single. AMD/Vega thread. Doesn't matter what forum I visit. It's hilarious how dumb they look. Obnoxious, HaHaHaHaHaHa Right.

PS: Out of curiosity, would you consider me an AMD 'fanboy'?

 

[okidna]

Sounds good but why are bears involved , surely its bares and that's from a frequently corrected gramma guy.

 

[sweet]

Big Vega was said to have almost the same die size as Fiji, but on 14nm instead of 28nm. AND some are saying it will have the same shader count as Fiji???

Either this is bs or AMD actually comes out with a new pipeline micro architecture (which is very unlikely).

 

[Captain_Tom]

Big Vega was said to have almost the same die size as Fiji, but on 14nm instead of 28nm. AND some are saying it will have the same shader count as Fiji???

Either this is bs or AMD actually comes out with a new pipeline micro architecture (which is very unlikely).

Remember Maxwell? The rumor is that Nvidia switched to "Tiled Rasterization" when they went from Kepler to Maxwell. That's also the first time Nvidia had a real efficiency advantage (A big one too).


Rumors point to Vega being AMD's first arch to use tiled rasterization. All evidence points to it too with a bigger die size per SP, more integrated ROP's, and some pretty crazy geometry and efficiency claims. Fingers crossed that they pull it off!