Separate names with a comma.
Discussion in 'AMD Flavor' started by _mockingbird, Aug 3, 2017.
This thread is for rumors about Navi
Lol! No you didn't. Year and half of back and forth coming.
I got one. #poorVolta lol
Only rumor I got is that Navi has taped out, just waiting on the EUV machines to be installed for mass production which I am guessing will take 6 months to 8 months.
I blame this thread on c3k.
This is going to be epic
It's got genuine Nexgen Memory (GDDR5X?) and it'll be out early 2018.
Some fun is cool, but let's not turn this into some bashing thread, either for AMD or people who like their stuff.
Navi will so kick Volta butt! Because. Discuss, but if you don't agree, you're wrong.
Navi will reach peak efficiency faster than all other efficient things born before or after. It will efficiently heat not just entire rooms but whole condos. It will efficiently play 4-7 year old games with great driver optimization 4-7 years after it comes out. If you where to try to quantify its efficiency level on a scale of efficient to efficient it would surpass all expectations of efficiency. But, the best part is it will achieve equal or better, or trade blows with at a higher level of efficiency with volta for only 12 billion more resistors.
I just hope one rumor is true. That Navi will truly be a scalable design in the way Ryzen is scalable. And they find a way to make multi gpus work in tandem as a single gpu with some enhanced version of infinity fabric.
Any source to this? I have never heard of it. I know originally it was early 2018. But now I am hearing late 2018. But AMD has said they have design teams working in parallel.
While I know its wtftech, it is simply restating info from an investors conference so I would say the tape out is at least confirmed.
its got a wide foot, its going to take both balls at once!
Tape out is 2nd half, expect that to be end of 2017 though, that is pretty much how AMD has been going thus far. Vega was to be 1st half of 2017 tape out, and it ended up just after that. Zen launch 2nd half of 2016 launch, that was end of 2016 launch. Polaris 2nd half of 2016 launch and it was end of 2016 launch.
Same with Vega launch too 1st half of 2017, and it came with FE end of 1st half.
Wouldn't that mean Navi would be getting released within 6 months of Volta?
It took 1 year after tape out for Polaris and Vega to come out. If they stay true to that, they end of 2018.
I think Vega could have been here sooner. Driver might be problem since they still didn't feel comfortable activating DSBR. But they did show this shit working a while ago. I think they were truly screwed by their choice to use HBM2. Would it be too much to say it could actually launch by end of July 2018 a year after vega? If there are no delays in components?
Probably fair but their claim is it will use some sort of next gen memory, so don't get your hope sup.
It will launch with drivers that hide any type of FPS being shown by any software , because FPS won't matter.
But it will show you buttery smooth minimum FPS.
Possible, just going by AMD's history for now. They didn't have those problems with Polaris, and they expected to have a huge lead over Pascal for launching too, so there is some food for thought.
They will probably not have HDMI 2.1 because AMD.
Just to stir things a little... images may break.
The thing is there is no looming memory crisis so few games even top out at 4gb of vram.
That opportunity slide proves that AMD loves their margins like any other company. It also says a lot about how AMD sees the profitability of the high end/HPC segment contrary to some people saying that it's in the mid-range (Polaris) is where all the money is at.
I find that picture pretty funny, actually, right down to the "professional graphics".
Seems like AMD/RTG has a lot of things to do before Navi.
"The product operates with an I/O data rate of 16Gbps(Gigabits per second) per pin, which is the industry’s fastest. With a high-end graphics card, this DRAM processes up to 768GB(Gigabytes) of graphics data per second. SK Hynix has been planning to mass produce the product for a client to release high-end graphics card by early 2018 equipped with high performance GDDR6 DRAMs."
"A client", lol.
"First introduced in June 2016, the HBM2 boasts a 256GB/s data transmission bandwidth, offering more than an eight-fold increase over a 32GB/s GDDR5 DRAM chip. With capacity double that of 4GB HBM2, the 8GB solution contributes greatly to improving system performance and energy efficiency, offering ideal upgrades to data-intensive, high-end computing applications that deal with machine learning, parallel computing and graphics rendering.
In meeting increasing market demand, Samsung anticipates that its volume production of the 8GB HBM2 will cover more than 50 percent of its HBM2 production by the first half of next year."
"GLOBALFOUNDRIES today announced the availability of its 7nm Leading-Performance (7LP) FinFET semiconductor technology, delivering a 40 percent generational performance boost to meet the needs of applications such as premium mobile processors, cloud servers and networking infrastructure. Design kits are available now, and the first customer products based on 7LP are expected to launch in the first half of 2018, with volume production ramping in the second half of 2018."
“We are very pleased with the leading-edge technology that GF is bringing with its advanced 7nm process technology. Our collaborative work with GF is focused on creating high-performance products that will drive more immersive and instinctive computing experiences.”
Mark Papermaster, CTO and senior vice president of technology and engineering, AMD.
Somewhere else I was mentioning that by Q3 2018 things will be different, and this is the sort of thing that leads me to believe that.
Is it way off base to think of an available 7nm HBM2 product by the end of 2018?
BY MY POWERS OF PROPHECY...
I WILL PREDICT THE LAUNCH OF NAVI!!!
1. IT WILL BE LATE
Navi will not hit when AMD first hints it to launch. In fact, it will probably be 2-3 quarters later than what even the most pessimistic of roadmaps predict. It will also be teased, announced and launched multiple times before anyone can actually get their hands on it.
2. IT WILL BE INEFFICIENT COMPARED TO NVIDIA'S THEN-CURRENT CARD
It won't hit the energy efficiency of Nvidia's cards. In fact, it may not even match Pascal at the high end. It's die-size will be bigger than similarly priced Nvidia cards, and it will not be as easy to OC.
3. IT WILL NOT BE ALL THAT IMPRESSIVE
Not to say it will be bad, rather, it won't break new ground, and it won't disrupt the market much at all. Dispite AMD's suggested MSRPs, the retail cards will rise in line to match simillarly performant Nvidia cards. It will not take the performance crown. It may have more VRAM per dollar than Nvidia's products, but that is about it.
4. IT WILL NOT HAVE ANY DISRUPTIVE TECHNOLOGY, REGARDLESS OF WHAT AMD CLAIMS
Regardless of what RTG says, the cards' memory architecture and core improvements won't count for much. It's most lauded features will really just result in 'being faster than Vega'. If anything, these new tech bits will only provide more leeway for AMD to make "drivers aren't ready" excuses.
HERE SPEAK'TH THE [H]ARD PROPHET.
MAY HIS WORDS BE SET IN STONE
Please don't be another Vega or Fury, PLEASE, get Navi right...
The interesting thing here is Vega has an interesting memory controller that could definitely be something that is adopted acros the board by even NVIDIA. The problem as usual goes to who is actually going to optimize their games for it. We know Bethesda for some odd reason has gone full on AMD support, but not really anyone else.
The problem is release date for AMD. Even if they can magically get it on par with Volta, it'll probably come out after Volta had been out for over a year lol.
The thing is if AMD doesn't make good cards (GPU) and can't get market share, you think Bethesda is going to care to stick with AMD optimizations? Is MS and Sony going to stay with AMD as well? For consoles its all about saving money, but if AMD can't get power consumption in control with the same or higher performance profile, all of these things will be in jeopardy. Power usage in consoles is important too, if they can cut down 5 bucks in cooling needs then they can spend that in GPU or SOC needs.
CPU wise they no longer need to stick with x86 they can go to ARM. Granted it will kill backward compatibility but that has always been the reason to buy new consoles cause new games will only come out on the new consoles. This generational console thing, I think Sony and MS are going to find out, its not going to help them sell more consoles so market share is going to be stagnant, its going to slow down the sales of new consoles. How ever little they are making from the new consoles and R&D for them, is all going to go to waste.
Memory crisis is only to reach exaflop compute levels.
Isn't that part of what steers product development, or do I misunderstand? From what I gather some of those projects are strong considerations for Nvidia and more recently AMD since they got grant money and agreed to invest in it with the US DOE, among other ... customers.
it is, but with gaming cards, the needs for certain tech just isn't there yet. HBM is definitely not needed right now and its too expensive to really put on a graphics card, there are better options.
Things like HPC and DL hell ya they need that bandwidth but those cards or compute comps can sustain the increased cost of memory and manufacturing of HBM type memories.
I've been thinking that HBM didn't perform as well as they thought it would. Is that off base?
I've also been thinking that improvements to HBM in a number of ways could change the performance of AMD/RTG graphics cards in favorable ways. Is it fair to say that the memory is more the problem than the chip?
AMD/RTG would have you believe that there is a lot more performance down this road and they're betting the farm on it.
After looking at product teardowns of a lot of graphics cards its looking like the rest of the card is somewhat standard now.
hmm not exactly, it performs the way it should just that AMD needs HBM and the like for their GPU's, their GPU's need more bandwidth to sustain the performance they have. Pascal needs less bandwidth receptively.
For AMD to use GDDR5x their on die bus size will make the die bigger. As much as probably 5% to 10% bigger. Not much but in the grand scheme of things nV has quite bit of advantage with their current uarch.
What ever memory is used its about latency and bandwidth, how to feed the GPU for the needed applications. HPC and DL markets, the amount of calculations being done they need more bandwidth and the faster it gets to the GPU (less latency) the better. Games on the other hand, use more localized data to create their effects, every single pixel doesn't need to be calculated for all the time. Not only does it not need to be calculated, it can be extrapolated for outlier pixels too at times. And this is where games just don't have the same needs.
AMD's approach although sound prior to GDDR5x kinda fell apart. They didn't factor in the first delay of HBM and Fiji (relatively speaking), and after that they got locked into an exclusivity contract, which sounded great cause yeah it would limit their competitors but at the end the delays helped their competitors cause now GDDR5x was available and they were also able to get HBM 2 from Samsung.
It was a bit of a logistical nightmare for AMD more than anything else. Hard to predict cause well they not are manufacturing the memory and the manufacturer of course will say everything is looking good for their end, since the exclusivity contract is in place, there is nothing AMD could do about it.
That's the assumption I've understood so far.
Wouldn't the current level of HBM2 memory performance limit the performance potential of RTG gaming products?
Would GDDR6 enable performance that AMD/RTG will not be able to match without using it?
About HBM2 yeah it might be done don't know yet till tests come out
Well GDDR6 will come out early next year, so at this point, don't know any AMD cards that will really need that till Navi.