AMD Vega, Nvidia Volta

I think that Vega is dead on arrival. AMD waited too long. Nvidia doesn't have to do much to beat it into the dirt at this point.

I agree performance wise. Im sure gaming drivers will bring it up a little. It will come down to cost. Might be a hard call with HBM2 going up against cheaper memory.

If AMD had been able to release earlier, they would have. Vega FE was released so they wouldn't get sued by investors, they promised to release something in H1 2017. And what they released looks very much unfinished, nothing of the new tech seems to be working when people test it. I think they simply didn't have enough resources to compete with Nvidia. I don't think RX Vega will be much different, something will probably be released in August, but I'm not holding my breath. What resources they have are probably going into fulfilling their contract with Apple. The Mac Pro with Vega is supposed to launch in December right?

GTX Volta will probably release sometime next year, I expect the usual perf increase over the previous gen.

1 Like

I just want Freesync 2.0. If NVIDIA supported it, I would already have at least a 1080.

I personally like what amd is doing with their graphics card architecture. As it looks like they are trying to add an on board graphics card accelerator. I know that is not what it is called. But I think that amd trying to reinvent the wheel that is the graphics card might have interesting prospects with asynchronous computation on ryzen if it pans out right for amd. But I like that amd is doing this as it means that hardware on graphics cards can be improved upon, rather than just throwing ten billion more transistors at it.

We dont really know anything about Vega or Volta yet in terms of performance figures.
So its hard to say anything about it.
Atleast from a gaming perspective that is.

But Nvidia holds the performance crown in the highend gaming segment for a while.
And i personally dont think that Vega is going to shock the world in that.

You wanted predictions? Well strap in for a rambling coffee fuelled tirade brought to you by stupidly little sleep. Please excuse poor spelling, punctuation and grammar. I'm bad at the best of time, and this is not a good time!

Now, Vega RX cards, within 10% of the GTX1080, will it impress your pants off? Hell no, it's going to be a worse Fermi, hot, power hungry and slower than the competition. Fermi was at least faster than ATI/AMD back then. Hopefully AMD will survive and move upwards and onwards to Navi. Navi is probably going to be disruptive, but less disruptive than I thought 3 days ago. Now, in an AMA over on Reddit, Raja (y'know the lead engineer guy for Vega), said that Vega is the first InfinityFabric GPU.... What could that mean for Navi? Well, there's one big hint from the Capsaicin conference AMD held back in March this year. Scalability. What was one of AMD's big things when they've been talking about Ryzen and Epyc? Scalability. Vega, first step on the road to a scalable GPU in line with Ryzen. InfinityFabric on the GPU, that is how AMD intends to leapfrog back into the high end GPU market.

Now, I mentioned that I thought Navi would be more disruptive 3 days ago. That is because it just came to my attention that Nvidia has released a whitepaper on MCM-GPUs, or Multi-Chip-Module, and to be frank(no I'm not Frank, bugger off) it looks a lot like InfinityFabric. Not to say that Nvidia didn't cook this up on their own, but the timing is... Suggestive.

So I guess it comes down to this, who can release MCM-GPUs to market first, AMD already has the technology on the CPU, does Nvidia have it in a lab? Will InfinityFabric adapted to the GPU scale as well as IF does on the CPU(Near perfect scaling!). My prediction at this moment in time is, 2017 is definitely coming on strong as the year AMD takes back some share in the CPU market, will 2018 look the same for the GPU market? (Roadmap from Capsaicin conference shows Navi release first half of 2018) <- Roadmap from Capsaicin conference March 2017 <- Raja comments from AMA thread <- "Leaked" benchmark showing Vega RX slightly ahead of 1080 (there's more reason to belive this is accurate, but I'm detecting blood in my coffee circulation so the brain is throttling) <- MCM-GPUs whitepaper <- InfinityFabric on Epyc (not the best image, but it gets the point across)

FE and RX are the same silicone with different memory. If RX is clock for clock significantly faster in games (which might happen) it will be a driver thing and that will trickle down into the FE driver as well. There is no difference in optimization between RX and FE in gaming mode. The statement "Vega FE is not a gaming card" is factually correct but irrelevant.

Won't be seen before 2018. Therefore irrelevant for now.

I'm reading this over and over, but I never really found any source (never really cared and/or looked for it)... are these actually the same chips? I mean, they probably come from the same wavers, but so do 1080, Quadro and Tesla chips, still they are different in performance and features. Soooo... I wouldn't say the "same" per se.

... which are also the same GPUs.

That was the point, they still perform differently though.

Sorry that I have to do this to you....

1 Like

Gotta watch that at home, thanks in advance tho :slight_smile:

For me it is price to performance. Give me the best for what I can afford. Vega is only a failure if it is priced wrong.

All blower cards are hot, period. I had 2x GTX 480s OC, never went above 45c in furmark on water. That is one of the reasons why I water cool. I do not care about heat or power, just performance.

it is looking more and more each day my new build will be Threadripper and 2x Vega FE now that EK is going to make waterblocks for them.

1 Like

Volta is obviously going to perform better than Vega unless somehow they messed up real bad, even Pascal I think is (for a die-shrink, Pascal was a monumental improvement over Maxwell), seems like AMD's focus on Ryzen took a toll on their competition on GPU. I mean I am happy CPUs are getting better but the GPUs not looking so hot as of late.

All that said, I wish we could have multiple VMs running on 1 GPU, that would be great, too bad AMD didn't put that on Vega, hopefully Navi has it. But it's too bad since I was going for a Raven Ridge APU.

I think that may be actually AMD's plan.

I thought it didn't. I thought they ditched it last second.

Some improvements for Vega FE when on water:

"The primary data point from our testing of the AMD Radeon Vega Frontier Edition liquid-cooled card is that the performance delta between it and the air-cooled card were significantly wider than I expected. We measured improved average frame rates that were 10%+ higher with the card in its 350-watt setting, which I expect most consumers to use, the vast majority of the time. At 2560x1440 there were a handful of instances that show only single digit increases, but that was infrequent. As a result, the Vega FE liquid card is much more capable of competing with the GeForce GTX 1080 in our gaming tests, coming within a few percent in most tests (taking away GTA V) and taking the lead in Dirt Rally. Considering my mostly negative outlook on the Vega product family after reviewing the air-cooled card, the liquid cooled version shifts the landscape again.

Which leads to the real question: what does this mean for the upcoming RX Vega product? While I know that testing the Frontier Edition in ONLY gaming is a bit of a faux pas, much of our interest was in using this product to predict what AMD is going bring to gamers later this month. It is apparent now that if the clocks are in the 1600 MHz range consistently, rather than dipping into 1400 MHz and below states as we found with the air cooler at stock, Vega in its current state can be competitive with the GeForce GTX 1080. That’s an upgrade over where it stood before – much closer to GTX 1070 performance.

Before doing this testing, I had dismissed the liquid-cooled version of the Vega Frontier Edition completely, but the performance and improvement in clock rate/temperature we saw today gives me hope that RX Vega will be able to turn in a better-than-expected performance when it hits the scene. There are still plenty of ways that AMD could drop the ball, but if they can launch RX Vega at similar performance to the GTX 1080, as we saw here with the card in its 350-watt state, and prices it correctly, there will be a line up for buyers. And that is despite the obvious power efficiency disadvantage that AMD will have to accept.

The sad thing is, they used GPU profits to fund Zen and now they will use Zen profits to fund the GPUs...
There should be independent GPU and CPU division as far as I know. Now Zen will fall back, because the money will go to help the radeon catch up and by that time radeon may have to save the cpus again...
This is a situation I don't like. I am exclusively using amd because I consider them technologically amazing and honestly superior to the competition. But that doesn't seem to be working out for them very well...

1 Like

I'm pretty sure the real distraction for the Radeon group engineering is/has been Raven Ridge APU's and it is the Raven Ridge APU's that will justify further funding and make money back to the Radeon Group.

1 Like