Monday, April 24th 2017

AMD Announces New Radeon Pro Duo - Polaris x2

Today AMD (NASDAQ: AMD) announced the world's first dual-GPU graphics card designed for professionals: the Polaris-architecture-based Radeon Pro Duo. Built on the capabilities of the Radeon Pro WX 7100, the Radeon Pro Duo professional graphics card is designed to excel at media and entertainment, broadcast, and design and manufacturing workflows, delivering outstanding performance and superior flexibility that today's creative professionals demand.

The Radeon Pro Duo is equipped with 32GB of ultra-fast GDDR5 memory to handle larger data sets, more intricate 3D models, higher resolution videos, and complex assemblies with ease. Operating at a max power of 250W, the Radeon Pro Duo harnesses a total of 72 compute units (4608 stream processors) for a combined performance of up to 11.45 TFLOPS of single-precision compute performance on one board, and twice the geometry throughput of the Radeon Pro WX 7100. The Radeon Pro Duo enables professionals to work up to four 4K monitors at 60Hz, drive the latest 8K single monitor display at 30Hz using a single cable, or drive an 8K display at 60Hz using a dual cable solution.
"Today's professional workflows continue to increase in complexity, often demanding that creators switch between a wide variety of applications to progress their work, pausing efforts in one application while computing resources are focused on another. We designed the Radeon Pro Duo to eliminate those constraints, empowering professionals to multi-task without compromise, dedicating GPU resources where and how they need them. It's a continuation of our promise for Radeon Pro: to provide greater choice in how professionals practice their craft, enabling superior multi-tasking, accelerated applications, and powerful solutions for advanced workloads like VR," said Ogi Brkic, general manager, professional graphics, Radeon Technologies Group, AMD.

Divide
The Radeon Pro Duo's distinct dual-GPU design gives professionals the flexibility to divide and conquer their workloads, enabling smooth multi-tasking between applications by committing GPU resources to each. Professionals can maintain their creative momentum and get more done faster, allowing for a greater number of design iterations in the same time.

"I was very impressed with the power of the Radeon Pro Duo, particularly in Nuke. The flexibility of being able to divide GPUs between tasks is phenomenal and represents the ultimate in multitasking: compositing a complex shot while jumping into a 3D application to create assets, exporting back to Nuke to keep compositing then switching to Photoshop or Mari and paint a projection, to load it back into Nuke and continue. The Radeon Pro Duo handles the general and varied tasks without missing a beat with excellent 3D performance. For the kind of projects I undertake as a generalist, the Radeon Pro Duo is a no-brainer. It does it all," said Kynan Stephenson, freelance artist.

Accelerate

The Radeon Pro Duo dramatically increases performance in today's most popular professional applications through support of multiple GPUs or plug-ins, speeding through tasks. On select professional applications, the Radeon Pro Duo delivers up to 2 times faster performance compared with the Radeon Pro WX 71003 and up to 2 times faster performance than the closest competing professional graphics card.

Create
The Radeon Pro Duo's potent combination of performance and dual-GPU flexibility makes it the ideal solution for today's advanced workloads, including professional VR content creation. VR represents a significant inflection point for the media and entertainment industry and is becoming more commonplace in today's studios. Radeon Pro Duo professional graphics card leverages the power of two GPUs to render out separate images for each eye, increasing VR performance over single GPU solutions by up to 50% in the SteamVR test.4 AMD's LiquidVR technologies are also supported by the industry's leading real-time engines, including Unity and Unreal, to help ensure smooth, comfortable, and responsive VR experiences on Radeon Pro Duo.

"In developing 4K 360 VR content, the biggest hurdle is the tech, because as an artist, I just want to create and not worry about limitations of the hardware. Faced with raw, un-optimized content, VR creators need a lot more horsepower than VR consumers. With the new Radeon Pro Duo, I have performance in spades. I immediately saw a speed difference of up to 2X, allowing me to push the boundaries of my projects without having to compromise on creativity or productivity," said Jonathan Winbush, Founder & Creative Director, Winbush.tv.

The Radeon Pro Duo is designed to meet the stringent requirements of workstation form factors, comes with 24/7 support5, and is bolstered by professional-grade software certified across leading applications for media, entertainment, CAD and engineering. Radeon Pro Software offers users unprecedented driver stability, quality and reliability, with quarterly updates for features, performance and stability.

The Radeon Pro Duo's planned availability is the end of May at an expected SEP of US$999.
Add your own comment

59 Comments on AMD Announces New Radeon Pro Duo - Polaris x2

#26
jabbadap
RejZoRIs it? Capitalizing would be selling GTX 1080 for same price for longer. Not releasing Ti for no real reason and ANOTHER Titan. Again, for no real reason. They are just tossing out money, making their expensive cards redundant. They could go reactive, if AMD released Vega and it proved to be unusually faster, they could respond with Ti or new Titan and take the crown from there in a moment.

The Vega has been planned for 1H 2017 for ages, so there is no way they were not prepared and they just blindly released those because they already had them in the pipeline.
Ohh there's plenty of reasons to release Ti and new titan. a) Being in market first, gives them time to sell cards without any competition. b) gtx1080 has been market for long time, I bet nvidia has some kind of threshold of quantities still selling which dictates there's time to new product. c) Being bigger chip, it has been take more time to get yields in order and getting healthy stock. d) Vega is late, they got bored...
Posted on Reply
#27
Xzibit
PatriotThis isn't hate, this is speculation based on unusual silence. I doubt anyone here wants AMD to do poorly but we don't understand why they keep demoing it and hyping without releasing it. They are likely working on a fix for something but some transparency would be nice.

I want vega to do well, but what's the holdup? Supposedly it will launch in may but we shall see.
When the thread spirals off into speculation with, OMG they are releasing this because something is wrong with VEGA. Well you kind of look silly since this NEWS was about a PRO card in the first place and the VEGA release window hasn't changed.

Same sillyness from people as with Ryzen launch where people were upset they weren't being feed all the info at their convenience.
Posted on Reply
#28
owen10578
PatriotThis isn't hate, this is speculation based on unusual silence. I doubt anyone here wants AMD to do poorly but we don't understand why they keep demoing it and hyping without releasing it. They are likely working on a fix for something but some transparency would be nice.

Even without beating the 1080ti it would still be a compute monster as they allow packed math on consumer cards giving you double half precision performance. Nvidia only does this on the p100 so vega WILL be faster at all deep learning applications other than the p100. The cache controller allows it to use memory in a unique way that should allow for interesting implementations, perhaps sdk around that is what is slowing launch.

I want vega to do well, but what's the holdup? Supposedly it will launch in may but we shall see.
Yea I get it its unusually quiet but making baseless speculation doesn't help at all. You would just look like a helpless AMD hater to me.

Im hoping Vega to do well too because we do need competition.
Posted on Reply
#29
Patriot
XzibitWhen the thread spirals off into speculation with, OMG they are releasing this because something is wrong with VEGA. Well you kind of look silly since this NEWS was about a PRO card in the first place and the VEGA release window hasn't changed.

Same sillyness from people as with Ryzen launch where people were upset they weren't being feed all the info at their convenience.
It's not a pro card... They simply branded it such because it doesn't compete on the gaming realm. It sits between gaming and pro, a prosumer card at best. There was no dual fury apart from the "pro" card and no dual rx480 apart from the "pro" card. Don't just swallow all the bs you hear because it is what you want to hear.
owen10578Yea I get it its unusually quiet but making baseless speculation doesn't help at all. You would just look like a helpless AMD hater to me.

Im hoping Vega to do well too because we do need competition.
It is not baseless, they have yet to release anything faster than the first Pro duo in a single card form. They have historically released a dual card when they don't have a single card to do the trick.
Posted on Reply
#30
TheoneandonlyMrK
XzibitWhen the thread spirals off into speculation with, OMG they are releasing this because something is wrong with VEGA. Well you kind of look silly since this NEWS was about a PRO card in the first place and the VEGA release window hasn't changed.

Same sillyness from people as with Ryzen launch where people were upset they weren't being feed all the info at their convenience.
Exactly and no one read through it to realise the use cases that can't be done without two GPUs ie final render on one GPU while allocating the other for live use sorting the next model or whatever ,all appealing in a new way for Professional use ,they didn't even mention gaming on it or crossfire though obviously it would be capable and at a grand only a Muppet would buy one for crysis or csgo.
Posted on Reply
#31
Frick
Fishfaced Nincompoop
Speculation on Vega aside, that is a very attractive card.
Posted on Reply
#32
r9
owen10578Actually im easily beating a 1080 with dual 480s and even the 1080Ti when overclocked...well in benches. Games I don't really know so it might very well be 1070 speeds in non optimized ones. A dual 480/580 gaming card would be pretty cool though. But im not sure it would even sell considering there are just as fast if not faster nvidia single cards.



Why all the hate? It's not even out yet ffs. I don't really see how AMD should be shocked by the Ti or Xp considering the original XP was out for a long time already...also the same way HBM increase costs adding traces for GDDR memory on the pcb also adds cost so not sure that HBM2 "cost way too much". Not saying Vega will be great, if anything im a bit skeptical considering how quiet AMD is but we can do without the bashing and baseless speculation.
Posted on Reply
#33
TheoneandonlyMrK
r9
I kinda thought this given at 1430 my max is 7236 in timespy ,, better then 980sli apparently but not quite 1080 levels.
I'm loosing a bit because of the platform though, not too much,, yet.
Posted on Reply
#34
HD64G
r9
At 4K they are practically equal though on average. So, in some games better and in others worse. And at 1440P very close also. Always SLI and CF work best at higher resolutions.

Posted on Reply
#35
snakefist
This is a card for professionals. It is not supposed to be a card for gamers (none of which would profit from 32Gb of fast memory). It's supposed to be supported in specialised software, where CF (on a single board) will be utilised, so gaming benchmarks aren't very relevant... A niche product.
Posted on Reply
#36
FordGT90Concept
"I go fast!1!11!1!"
PatriotYeah.... this is not giving me high hopes about vega...
Except that Vega has more TFlops (12-12.5) than this (11.45) with fewer stream processors and can reasonably have two on one card like this pumping a crazy 24-25 TFlops of data. Titan Xp is 12 TFlops...

I think producing RX 480, RTG discovered the design philosophy for 14nm GloFo didn't carry over from 28nm TSMC. Once they realized that, all of their efforts were poured into Vega. In the interim, all they managed to do was do some process tweaking to produce RX 580.

RTG has everything riding on Vega. This launch is to stop professionals from jumping ship: Titan Xp-like performance with significantly more memory and $200 cheaper. This card has deep learning written all over it.
Posted on Reply
#37
OneCool
So they put on two slower gpu's and it better? o_O
Posted on Reply
#38
FordGT90Concept
"I go fast!1!11!1!"
If you have an application that can use dual GPUs and requires >12 GiB of VRAM, yes, it's better. You could always sell your first born ($10,000+) for a Tesla P100 (11.7 TFlop) but why when you could have 10 Polaris Duo cards (114.5 TFlop) for the same price?
Posted on Reply
#39
TheGuruStud
FordGT90ConceptIf you have an application that can use dual GPUs and requires >12 GiB of VRAM, yes, it's better. You could always sell your first born ($10,000+) for a Tesla P100 (11.7 TFlop) but why when you could have 10 Polaris Duo cards (114.5 TFlop) for the same price?
Get out of here with your logic! You can only bash AMD, even when they release a great product at a fantastic price.
Posted on Reply
#40
Aenra
I thought it best to come here and add my own constructive criticism.

I don't like the color.

(ergo) AMD sucks.

"One must above all be scientific my dear"

edit: on a serious note, has anyone used something like this? Am curious about their temps, what with just that fan on them.
Posted on Reply
#41
Foobario
OneCoolSo they put on two slower gpu's and it better? o_O
Maybe it's the drastically reduced TDP combined with a lower cost to manufacture as two Polaris die on 14nm is much cheaper to produce than two Fury die on 28nm.

This also moves more production to GLOFO (Polaris) from TMSC (Fury) which reduces penalty payments owed to GLOFO for using another foundry under their new wafer supply agreement.

The slight drop in TFLOPs doesn't change where this falls in professional landscape it is targeted for. It is now cheaper and more efficient for those that purchase in this segment.
Posted on Reply
#42
mroofie
FrickSpeculation on Vega aside, that is a very attractive card.
$300 more for small performance boost(over Ti obviously) and 21GB vram :ohwell:

Meh.. :pimp:
Posted on Reply
#43
mroofie
FoobarioMaybe it's the drastically reduced TDP combined with a lower cost to manufacture as two Polaris die on 14nm is much cheaper to produce than two Fury die on 28nm.

This also moves more production to GLOFO (Polaris) from TMSC (Fury) which reduces penalty payments owed to GLOFO for using another foundry under their new wafer supply agreement.

The slight drop in TFLOPs doesn't change where this falls in professional landscape it is targeted for. It is now cheaper and more efficient for those that purchase in this segment.
Till it gets marketed as a gaming card then my post above applies :)
Posted on Reply
#44
EarthDog
Its not a gaming card... who cares forngamong gents. :)
RejZoRAMD is being super silent on Vega intentionally. Just look how scared NVIDIA is, releasing faster cards for no reason lol. A Ti card and two Titans. It's interesting they keep leaks so tightly secured for this one...
scared... lol, lemmings walking off the cliff following raevenlord in his news post i see...lololol!!!
Posted on Reply
#45
craigo
Did anyone else think this when they saw the big R on the fan of the dual GPU card?

Posted on Reply
#46
nem..
with nvidia to an similar price you got an 1060 5.3tflops (quadro P4000)
Posted on Reply
#47
efikkan
RejZoRIt's strange they haven't made a RX580 X2 with RX480 clocks and 16GB of VRAM for the price point of a bit under GTX 1080. That could easily eat into the GTX 1080 segment. But strangely, they haven't done that.
No, it's not strange at all. Gaming and professional compute workloads scale completely differently. While rendering is a pipelined synchronized workload, many kinds of professional workloads can easily be divided into multiple independent chunks of work, and provided you have a fast enough CPU, enough PCIe lanes, etc. you can probably scale "perfectly" with two or more GPUs. Demanding users may run racks with 8/16 Titans or Teslas, sometimes even consumer grade GPUs. If the workload is completely independent, it also means the GPU memory is independent.

Gaming on the other hand scales very badly, so no informed person would choose to buy 2× RX 480 over GTX 1080.
Posted on Reply
#48
owen10578
PatriotIt's not a pro card... They simply branded it such because it doesn't compete on the gaming realm. It sits between gaming and pro, a prosumer card at best. There was no dual fury apart from the "pro" card and no dual rx480 apart from the "pro" card. Don't just swallow all the bs you hear because it is what you want to hear.



It is not baseless, they have yet to release anything faster than the first Pro duo in a single card form. They have historically released a dual card when they don't have a single card to do the trick.
Did you even read the article of the point of dual gpu cards? There is simply workloads that need 2 GPUs. Now if they released a dual GPU gaming card your theory would be more convincing.
r9
Alright I see in games it doesn't exactly beat the 1080 at stock then but in benches it does and I'm running mine overclocked anyways.
Posted on Reply
#49
TheinsanegamerN
RejZoRIs it? Capitalizing would be selling GTX 1080 for same price for longer. Not releasing Ti for no real reason and ANOTHER Titan. Again, for no real reason. They are just tossing out money, making their expensive cards redundant. They could go reactive, if AMD released Vega and it proved to be unusually faster, they could respond with Ti or new Titan and take the crown from there in a moment.

The Vega has been planned for 1H 2017 for ages, so there is no way they were not prepared and they just blindly released those because they already had them in the pipeline.
Or, perhaps, nvidia saturated the 1080 market since AMD didnt bother competing at all, and nvidia decided to lower the price to entice the rest of the market that hadnt upgraded and corner AMD into a tiny market in the process. If VEGA is 1080 level, the market will be full of 1080 users who upgraded a year ago, and will have no incentive to buy vega, or people will expect vega to be substantially cheaper due to higher heat output, and AMD will make no money. .

Same as GOTY edition games. lower price with DLC, but not targeting the same gamers that buy games brand new at launch.
xkm1948VR has been great. It is AMD has promised great software/driver support VR in advertising with NOTHING to back it up.

I have been working on a small VR demo. Getting any help from AMD is impossible. Their tech department never respond to any emails. Their tech support line knows nothing about their own LiquidVR.
Sounds like AMD's support for game devs. Developers always claimed AMD support was poor to non existent.
Posted on Reply
#50
warrior420
Lol, wow... Of course Vega got brought up in a Professional card's news post.

You are all retarded if you think AMD is going to be releasing ANY information on Vega before/after they JUST dropped brand new cards on us. Do you think they are that dumb? This is marketing at it's finest. They new a refresh was going to happen. You NEVER disclose anything about you're newest development before or even right after you're about to release an inferior product. Lol that's suicide for the newly released product. Then the customer (that would actually consider buying a 580X) will say, why should I buy now when waiting will bring a better value, I've already waited this long...

Plan on AMD waiting a while before the real Vega stuff leaks. It's the smart thing to do. Deal with it. Speculation at this point is just pointless.
Posted on Reply
Add your own comment
Nov 23rd, 2024 13:53 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts