Wednesday, June 1st 2016
AMD Radeon RX 480 "Polaris" Launched at $199
AMD made a bold move in launching its first "Polaris" architecture based performance-segment GPU, the Radeon RX 480 at a starting price of US $199. The company claims that it will perform on-par with $500 graphics cards from the previous generation, directly hinting at performance being on par with the Radeon R9 Fury and R9 Nano. Although it's not in the league of the GTX 1070 or the GTX 1080, this level of performance at $199 could certainly disrupt things for NVIDIA, as it presents an attractive option for people still gaming on 1440p and 1080p resolutions (the overwhelming majority). The R9 Fury can handle any game at 1440p.
The Radeon RX 480 is based on the 14 nm "Ellesmere" silicon, fabbed by GlobalFoundries. It's publicly known that GloFlo has a 14 nm fab in Malta (upstate New York), USA. The RX 480 is based on AMD's 4th generation Graphics CoreNext architecture, codenamed "Polaris." It features 2,304 stream processors, spread across 36 compute units (CUs). Its single-precision floating point performance is rated by AMD to be "greater than 5 TFLOP/s." The chip features a 256-bit wide GDDR5 memory interface, with memory clocked at 8 Gbps, yielding memory bandwidth of 256 GB/s. There will be two variants of this card, 4 GB and 8 GB. It's the 4 GB variant that starts at $199, the 8 GB variant is expected to be priced at $229. AMD confirmed that the GPU will support DisplayPort 1.4 although it's certified up to DisplayPort 1.3. The typical board power is rated at 150W. The card could be available from 29th June.
The Radeon RX 480 is based on the 14 nm "Ellesmere" silicon, fabbed by GlobalFoundries. It's publicly known that GloFlo has a 14 nm fab in Malta (upstate New York), USA. The RX 480 is based on AMD's 4th generation Graphics CoreNext architecture, codenamed "Polaris." It features 2,304 stream processors, spread across 36 compute units (CUs). Its single-precision floating point performance is rated by AMD to be "greater than 5 TFLOP/s." The chip features a 256-bit wide GDDR5 memory interface, with memory clocked at 8 Gbps, yielding memory bandwidth of 256 GB/s. There will be two variants of this card, 4 GB and 8 GB. It's the 4 GB variant that starts at $199, the 8 GB variant is expected to be priced at $229. AMD confirmed that the GPU will support DisplayPort 1.4 although it's certified up to DisplayPort 1.3. The typical board power is rated at 150W. The card could be available from 29th June.
104 Comments on AMD Radeon RX 480 "Polaris" Launched at $199
Well done Amd... mow even if performance is about 10% less (standard hype tax) than what you claimed your still in a sweet spot....well played. that means the 1060 wll have its work cut out for it.
RX 480 has (it seems) GDDR5 (non X), so it's not even collecting the lower power benefits of new memory (tiny bit from 5X or bigger from HBM).
R9 Nano (w/ HBM 1) had an improvement to around 175W already (and was still full 28nm), which made it perform usually slightly better than the normal Fury(with the full Fiji's 275W). I bet they learned a lot from this experiment/concept.
Vega will have brand new developed HBM2 and also 14nm. Will it have 150W like the 1070, or 180W like the 1080? We don't know, but I'm betting it won't be that far.
Would anyone buy the 950 for $100-120; a 960 for $140-160; or a 970 for $220-240. moving forward! My one son was all hot to buy a friend EVGA 970 SC for $230 (he paid $330 -AR Sept '15 he got ripped). I said don't that's not going to look like any deal in 3 weeks. He's already said thanks for the advice... that form a 18 year old!
While it makes sense for AMD to target a chip at this price/perf, as nvidia currently does not have an answer for it, the larger picture is more opaque but certainly of grander importance in the grand scheme of products we'll see over the next year or two.
For the former group (myself included), it will be interesting to see how well these will clock within 150w (or more loosely will be allowed to clock via bios/power draw etc). One might assume they could clock to ~1500mhz+/- (given leakage graphs of 14nmLPE which have the first raise in relative power/clock around 1300mhz, where RX480 is stock clocked, and the next around 1500mhz), but that's anything but a sure thing given AMD's typical arch design. If that WERE the case though, it would be a very interesting product....not only because it could start getting close to typical compute of 1070, but also actually see realistic perf/playability gains in 1440p/VR/potentially even 4k in some cases. IOW, it could actually be a cost-efficient replacement for Nano/Fury...and that would be a big deal.
On the flip side of that is the inevitable 150-225w part, which truly is most exciting IOHO, I think. The one we can all imagine that carries GDDR5x, perhaps be 2560sp (and/or have a higher clock...1500-1800mhz?), and compete with 1070 on true performance. Again, while clocking is ANYTHING but a sure thing, as is realistic power consumption and/or cooling of a chip/card that may be much smaller than 1070/1080, THAT is the card that would truly be disruptive AF if it's still within the 'mainstream' or 'performance' pricing structure. It could do what 1070 is doing (replacing the last-gen high-end), perhaps be less performant but not tangibly (since neither are truly 4k cards), and have lower perf/w than GP104 but still <225w, all while potentially being priced lower.
It's all pretty much predicated on how the design will clock though, and how heat is dissipated, especially as we reach levels greater than ~1500-1600mhz. There's the potential there for similar clockspeeds to Pascal (given 14LPE scaled to around 1800mhz in the same way 16FF+ scales to ~2000-2100mhz, and 14LPP should be slightly better), which could simply mean AMD for all intents and purposes would have a great all-around 1440p/VR chip at low cost (even if high-power) without needlessly trying to push it into fringe 4k territory (as a single card, at least) as nvidia did with 1080 but rather leaving that for Vega/Greenland...
Hence we wait for news from the field on how great the potential of this chip truly is.
The 150W maximum output is'nt really true. Standard 12V wires are up to 12A a piece maximum usage, so 3 wires still gives you a 36A = 432W maximum power.
It's the VRM's that is supposed to convert this power to a voltage the GPU and memory needs. It will be the VRM proberly that's being the 'limit' first when you OC it.
A 200$ card offering 5Gflops of performance is a very very sweet deal.
WSJ (partially paywalled) Once again, 6 pin connection is DESIGNED to be 75w.
8 pin - 150w.
PCI-e can feed you another 75w.
One can try to pull more, but that would be against specification.
en.wikipedia.org/wiki/PCI_Express#Power
It states nothing about 12A, only voltage and total power.
The PCI-express power from your PSU is not limited to either 75W or 150W, as advertised, but can deliver way more then that.
It comes down to the cards VRM which should convert all that power for the GPU and memory. To add more wires, would simply stabilize your power feed some little more, but would not restrict from going above 150W as advertised.
Molex and 12V PCI-express wire, are able to sustain over 12A on 12V, so this in theory would give you up to 432W of power without burning things up.
Citation needed.
Even 380 comes with 2 power connectors.
On 390x it's two 8 pins as well, which gives you 75w + 150w + 150w = 375w.
On 295x2 it was four (yep, 4) 8 pin connectors, which give you: 75w + 150w*4 = 675w.
`
What certain card/PSU CAN deliver is completely irrelevant, AMD can NOT be designing cards that break the specs.
What it came down to was, oc'ers would have serious burning marks on their motherboard ATX power-connector, esp when going with AMD cards. It looked like AMD cards did'nt follow PCI-express specification 'up to 75w' but would exceed that easily causing to pull way more power from the motherboards PCI-express then allowed. This would leave burned ATX power connectors on the motherboard which took them a while to figure out what was going on.
I am saying is that the specification on AMD cards, sometimes seem to exceed way beyond when going for mad clocks. And a 6 pin wire, that is supposed to deliver up to 75W of power, is'nt limited to 75W alone. You have these power boosters that are feed straight into your PCI-express slot providing a more stable current. It works when oc'ing and for example your pci-express bus is shared among other devices as well.
I just dont understand why a 6 wire should be a problem, while most of these 6 pins are rated for 12A per yellow wire. It comes down to the VRM at the end. If that VRM is basicly garbage, and unsuitable to manage more power being pulled then the standard 75 / 75W combo, then it's a shitty design anyway.
Here was a limited
Radeon HD 7870 XT = Tahiti LE
7950/7970/280/280X = Tahiti Pro/XT
So are you pointing to the 7870 XT? I would see that easily a Polaris 11 kind of place.
While Polaris 11 is the RE 460 will be the lowest discrete and/or C-F APU, while all the rest is in Apple pads, Laptops, and consoles (IDK).