-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
AMD R9 390X and R9 380X 20nm Pirate Islands Info Leaked (9x faster than GDDR5)
- Thread starter Green Biker Dude
- Start date
Camp Freddie
Member
Glad to see that AMD are going to have an advantage over Nvidia in the foreseeable future.
The 970 and (probably) 960 are going to be drinking AMD's milkshake until then.
The 970 and (probably) 960 are going to be drinking AMD's milkshake until then.
KillaKalz347
Banned
There is no debate, HBM is ready for production.
MikeE21286
Member
Irobot82
Member
Which is why I believe the R300's is really facing against Nvidias 1000 series or whatever they call it. They'll move Maxwell to 20nm, maybe with the full GM200 chip. Nvidia just pulled a smart move and managed to push out a 10% gain on 28nm that also reduced power. Q1 2015 we'll see the big guns come out for both companies. In the meantime team green will be bringing in the cash with their success of the 900's.
There are rumours that Nvidia won't be going 20nm and instead are looking at 16nm.
chaosblade
Unconfirmed Member
I think you misread my post. If NVidia's new architecture didn't massively outperform AMD's two and a half year old one, would that not be incredibly disappointing?
runningjoke
Member
Ive heard this quite a few times in different areas if technology and have seen it not be the case. Basically, I'll believe it when the OEM announce it in a product.
flyinpiranha
Member
What type of wand do I need to put this magic in my PC?
Irobot82
Member
Correct, I understand both want to move to 16nm finfet as quickly as possible. I read an article yesterday that said ARM is already producing 64 bit chips with TMSC but that it would be some time before the process is mature enough for an something like a GPU chip. I would believe perhaps end of 2015 beginning of 2016 so I don't think Nvidia has a choice but to follow suit into 20nm. I think that this will be 1 generation node. Instead of the two like we saw with 28nm.
thesuperfunk
Member
Pirate Islands is the most awesome name!
Shit yeah.. sorry. I thought you meant Maxwell vs. 390X.
KillaKalz347
Banned
How would a closed loop cooler be hard to install? The 295X2 was extremely simple to install, more so than a CPU cooler.
runningjoke
Member
It's at a price point that someone might walk into a store and buy one, only to find out they can't fit it inside their Dell case or whatever. I'm not AMDs marketing team, they know their target audience and markets better than me so maybe this isn't a problem at all. Maybe it's a gimmick to make the card sell better. Personally I'd pick a smaller card
Irobot82
Member
AMD to go to 14nm from Samsung?
Here's another interesting article. It could just be for AMD's CPU side, but why would they go to TMSC for 16finfet and Samsung for 14finfet.
Here's another interesting article. It could just be for AMD's CPU side, but why would they go to TMSC for 16finfet and Samsung for 14finfet.
I really don't see what you are talking about. This is a memory type. NVlink is an interconnect (and one primarily designed for HPC at that). These are entirely orthogonal.
Dictator93
Member
Orthogonal is one of my favorite english "relation" words. Just throwing that out there.
Nikodemos
Member
Didn't GlobalFoundries recently abandon their 14nm R&D and license Samsung's design?
Crimsonclaw111
Member
Sounds good, though I won't be jumping until 6gb is standard at least.
I'm in.
Was so close to hitting the button on a 970 too.
Delta Assault
Member
This would be nice... if video game devs actually immediately released games in that one year period that took advantage of HBM. But how likely is that to happen?
I mean hell, there are still tons of games nowadays that don't/can't even take advantage of 64 bit CPUs. And how long ago did those come out?
The 970 is a good deal in the US ($400~ CDN). We don't know how much AMD would charge for this either.
Man, just love this shit to death. Within a year we'll supposedly be having 9 times GDDR5 bandwidths and smaller process nodes in combination with new architectures....the jump between all that shit is gonna be massive.
That 200gb/s card is now at the very least 800gb/s....hell even 400gb/s would be crazy to me, that' by itself is still higher than 99% of all cards out on the consumer market to start with
That 200gb/s card is now at the very least 800gb/s....hell even 400gb/s would be crazy to me, that' by itself is still higher than 99% of all cards out on the consumer market to start with
Erm what. If you need a GPU now, there's really no better deal than the 970. A lot can happen in six months but that's a long time to wait. Always something better around the corner anyway.
mugurumakensei
Member
Where does 9x faster come from?
My 970 has 224 GB/s bandwidth. 256 GB/s is not 5x that.
My 970 has 224 GB/s bandwidth. 256 GB/s is not 5x that.
Architecture plays a big part iirc. If we were to only look at that, then we'd be able to assume the 7970 blows the 970 away, but it doesn't.
First generation of HBM is only 4.5x. It's 1GB/s I don't know where you're getting 256 GB/s. 2nd gen is 2 GB/s.
McHuj
Member
It's per device/chip. They're comparing a 7.0Gbps GDDR5 chip (28 GB/sec total) to 1 HMB device that's going at 128-256 GB/sec. A GPU would have several of these allowing it to reach ~1TB/sec in BW.
"Only". Jesus christ dude, isn't this a bit of a big deal concerning actual bandwidths after so many years of inching upward?
Locuza
Member
The 1st Generation of HBM should have 4 GB and 128 GB/s per Stack. (4Hi)
Since you connect four of them, you end up with 512 GB/s.
I think 4 GB are too small, so maybe AMD will made a custom solution based on HBM.
The 2nd Generation of HBM is much more flexible.
The bandwidth per stack doubles and you can scale up to 8 stacks.
Also the capacity goes up.
So 8Hi is maybe 8x256 GB/s = 2048 GB/s and 64 GB.
But the second generation is of course far away.
Delta Assault
Member
1 TB a second? Jesus Christ... that's gotta be enough to achieve photorealism.
mugurumakensei
Member
At which point, it would saturate the PCI express bus.
I'm not sure I buy that this is gonna be used in the 390 and 380 gpus. One, AMD would have announced it by now if they made such a big breakthrough. Two, AMD would probably be focused on integrating this with their HSA APUs as PCI express would be a gigantic bottleneck since transferring the memory from system RAM to VRAM would be drastically slow.
Objectively Bad Opinion
Member
This sits directly on the cards in place of GDDR5 VRAM. The GPU doesn't have to go through PCIe to access these.
So he says, Theres enough processing power on every laptop on the planet today, when AMD's best mobile CPUs are literally as slow as intel's i3 mobile parts.
Absurd.
It will definitely be an interesting future between the two tech giants, but right now Nvidia is gaining a ton in income thanks to the 900 series. That is definitely being fed directly back into their RnD, and I think we may see this success help leverage their future tech by pushing it out far quicker.
McHuj
Member
PCI Express bus is already saturated as compared to a modern high end GPU. It's only 32GB/sec (PCI 3.0, I think).
McHuj
Member
Probably right. If there first HBM's are expected to be 1GB and 128 GB/sec, a very high end GPU with 4 of them would be a good bet for the first attempt. It will probably be expensive as fuck at first, but if the stuff works as advertised, I think prices will plummet.
Prevolition
Member
How does this compare to the P90X?
KillaKalz347
Banned
AMD isn't really betting on better single threaded performance, their apu's are great, plus Mantle and HSA.
dr. apocalipsis
Banned
What nonsense is this. PCIEx isn't used as you think it's used.
Similar threads
- Poll