AMD Vega 10 “RX 490” Specs leaked, 4096 GCN 5.0 Cores & 16GB HBM2
Leaks concerning AMD’s Polaris graphics cards and their specifications have been consistently coming in over the past several weeks, signalling that we’re getting ever closer to the immanent Radeon 400 series launch this summer. AMD has developed three different 14nm FinFET GPUs, vega 10, Polaris 10 & Polaris 11. The specs of two of which have recently been leaked & today we’re adding the third to complete the list.
A listing of one AMD Polaris 11 graphics card SKU that’s been spotted on the Compubench database via videocardz. This database has played an instrumental role over the past several years in uncovering incredibly accurate information about next generation AMD graphics cards.
AMD’s entire Fury, 300 and 200 series were leaked on this very same database and all the specifications that we’ve learned from those listings about AMD GPUs over the past couple of years checked out. With that in mind, we have little doubt that today’s leaked Polaris 11 specifications are accurate.
Despite 11 being a larger number than 10, Polaris 10 is actually a larger, higher end GPU than Polaris 11. Chief Architect & head of AMD’s Radeon Technologies Group confirmed that the naming scheme is time-based & Polaris 10 had simply been designed before Polaris 11, hence the smaller numerical designation.
Polaris 10 is AMD’s mid-range 14nm FinFET graphics chip, Polaris 11 is the smallest of the family and Vega 10 is the HBM2 powered high-end flagship 14nm GPU launching in 2017 to succeed the Fury lineup. It will feature AMD’s next generation Vega architecture and will be the company’s first GPU to feature second generation High Bandwidth Memory.
Vega 10 To Succeed AMD’s Flagship Fiji GPU & The R9 Fury X
We’ve learned that AMD is preparing a next generation high-end GPU based brand new GCN architecture meant to succeed the company’s previous flagship R9 Fury X. Vega 10’s launch will come after the company’s roll-out of its mainstream Polaris RX 400 series graphics cards this year and will feature a brand new graphics architecture according to to the LinkedIn profile of a leading AMD design engineer.
I should point out that the “RX 490” moniker is by no means official and is simply a place holder until the card’s announced. AMD could opt to give the card an RX 500 series brand i.e. RX 590 or perhaps give it the Fury brand name i.e. RX Fury Vega.
Vega 10 Specs
|GPU||Polaris 10||Vega 10||GP102||GP104||GP102|
|Graphics Card||RX 480||TBA||GTX TItan X Pascal||GTX 1080||GTX 1080 Ti|
|Transistors In Billions||5.7||TBA||12||7.2||12|
|Performance||5.8 TFLOPS||12 TFLOPS||11 TFLOPS||9 TFLOPS||10 TFLOPS|
|Memory||8GB GDDR5||16GB HBM2||12GB GDDR5X||8GB GDDR5X||12GB GDDR5X|
|Bandwidth||256 GB/s||512 GB/s||480 GB/s||320 GB/s||480 GB/s|
The GPU will feature second generation stacked High Bandwidth Memory with up to 1 terabyte/s of memory bandwidth, more than three times that of the GTX 1080. The GPU is configured with a whopping 4096 GCN 5.0 stream processors. In addition to the near 80% increase in raw compute horsepower compared to the RX 480 & Polaris 10, Vega 10 will deliver a huge jump in energy effciency according to the company’s latest roadmap.
The Cut Back Version of AMD’s Polaris 11 GPU Has 16 Compute Units – 1024 GCN 4.0 Cores & A 128bit Memory Interface
So let’s get straight to it. There are six different Polaris 11 SKUs, likely three for desktops and three for notebooks. The specifications that have leaked are for an SKU with the device ID 67FF which is the most cut back, lowest end, Polaris 11 SKU there is. This means that the full Polaris 11 GPU will have more than 1024 GCN cores.
This indicates that the full chip will likely have 1280 GCN cores and will be a replacement for AMD’s Pitcairn GPU – 370, 270 & 7800 series cards – in terms of pricing and market positioning. So it’s very likely we’re looking at the R9 470 here. The SKU in question also features a 128bit memory interface, however the memory capacity data is missing. But 4GB and 2GB of GDDR5/X memory are two likely configurations for this GPU.
This is the third AMD 14nm FinFET GPU spec leak, prior to that we’ve seen the specifications for Polaris 10 & Vega 10 also leak online.
AMD Polaris Radeon 400 Series "Preliminary" Specs:
|WCCF||Radeon R9 480 Series||Radeon R9 480 Series||Radeon R9 380 Series||Radeon R9 470 Series||Radeon R9 470 Series||Radeon R7 370 Series|
|GPU||Polaris 10||Polaris 10||Tonga||Polaris 11||Polaris 11||Pitcairn|
|Fabrication Process||14nm FinFET||14nm FinFET||28nm||14nm FinFET||14nm FinFET||28nm|
|GCN Generation||GCN 4.0||GCN 4.0||GCN 3.0||GCN 4.0||GCN 4.0||GCN 1.0|
|Stream Processors||2560 SPs||2304 SPs||2048 SPs||1280 SPs||1024 SPs||1024 SPs|
|Memory||8GB GDDR5||8GB GDDR5||4 GB GDDR5||4GB GDDR5||4GB GDDR5||4GB GDDR5|
Polaris 11 Is The Same GPU That AMD Demoed Against The GTX 950 In December & Again At CES In January
Polaris 11 was the very first FinFET graphics chip in the world from either company, Nvidia or AMD, to be demoed publicly. AMD first demoed the chip to the press, including WCCF, back in December ahead of CES. AMD did it once again in January at CES and that was the very first public demo of the Polaris or in fact any high performance FinFET GPU.
The demo consisted of two identical systems one configured with Polaris 11 , R9 470 , and Nvidia’s GTX 950 running Star Wars Battlefront at 10 frames per second. The Polaris 11 system in its entirety consumed 54W less than the GTX 950. Considering that the GTX 950 on its own pulls ~95W on average under gaming workloads, Polaris 11 is essentially more than twice as power efficient. Delivering the same performance at roughly 40 watts. Which is less than half the power consumed by the GTX 950.
This is very early silicon, by the way. We have much more performance optimization to do in the coming months. But even in this early silicon, we’re seeing numbers versus the best class on the competition running at a heavy workload, like Star Wars—The competing system consumes 140 watts. This is 86 watts. We believe we’re several months ahead of this transition, especially for the notebook and the mainstream market. The competition is talking about chips for cars and stuff, but not the mainstream market.
AMD Polaris Graphics Cards Maintain Similar Core Counts To Their Predecessors With Significant Per Core Performance Gains
With all these leaks coming out we’re beginning to get a much clearer picture of what AMD is doing with its next generation Polaris based graphics cards. All the leaks that we’ve seen so far for Vega 10, Polaris 10 & Polaris 11 indicate that AMD is configuring its next generation GPUs with similar Stream Processor counts as their Fiji, Hawaii & Pitcairn predecessors, the Fury, R9 390 and R7 370 series.
The majority of the IP blocks in the graphics engine are new. Including a memory controller, for upcoming memory technologies, in addition to more capable and more power efficient compute units and geometry engine. The level 2 cache and the command processor have also been updated. Beyond what AMD detailed above about the key features and the display/multimedia updates the company did not go into much detail about what exactly went into these new IP blocks. We were promised more information will be revealed as we get closer to the release of Polaris graphics products.
The previous leak for the 2304 GCN core Polaris 10 – R9 490 – showed the card outperforming the GTX Titan Black and the GTX 980, despite a reported 800mhz clock speed. Since 14nm FinFET pushes frequencies way beyond what we’ve seen with 28nm we’re likely looking at 1200Mhz+ clock speeds out of production Polaris chips. Based on the performance we’re seeing at 800mhz, that puts the cut down Polaris 10 chip that we’ve seen, a likely R9 490 candidate, ahead of the R9 Fury X and GTX 980 Ti.
AMD has also stated on the few occasions that it talked about Polaris that 4th generation GCN core are significantly more powerful than their predecessors, hence the tame stream processor counts we’re saying.
Raja Koduri, Senior Vice President and Chief Architect, Radeon Technologies Group, speaking with Venturebeat.com :
We’ve redesigned many blocks in our cores. We’ve redesigned the main processor, a new geometry processor, a completely new fourth-generation Graphics Core Next with a very high increase in performance.
AMD Launching Its Flagship “Vega” GPU Late 2016
It’s true that we will have to wait until next year for HBM2 GPUs from either Nvidia or AMD. Nvidia’s GP104 powered GTX 1080 and GTX 1070 cards – not official names – are reportedly configured for GDDR5X. And there’s virtually no chance of neither Nvidia or AMD risking production of large GPUs like GP100 or Vega 10 this early in the nodes’ lifecycles. Neither 16nm or 14nm will be mature enough for large GPU launches this year. This is all assuming HBM2 production has already sufficiently ramped up, which hasn’t yet.
We’ve seen this highlighted at Nvidia’s GTC 2016 keynote during the GP100 powered TESLA P100 announcement. The company revealed that it doesn’t expect to see products based on its P100 accelerator shipping before Q1 2017.
Despite that, what we’re getting this year is still going to deliver a very significant jump in performance, performance per watt and performance per dollar. This is thanks to the much needed arrival of FinFET manufacturing technology as well as GDDR5X.
It’s not every day that we get GPUs with 2.5x better perf/watt, in fact not ever. This is the biggest jump from one generation to the next that we have ever seen. But it’s not all about efficiency, Raja is also promising the “most revolutionary jump in performance so far”.
Raja Koduri revealed at Capsaicin that AMD’s focus with Polaris is all about giving gamers a “sweet” performance per dollar deal. And this is exactly what Polaris 10 and Polaris 11 will do. AMD announced that Polaris graphics cards are going to be released mid 2016, this summer. AMD’s CEO Lisa Su later elaborated further, saying that Polaris graphics cards will be available on shelves and in notebooks before the back to school season this year.