The New FURY X2 Graphic Card!!!
AMD is preparing its fastest graphics card yet, a dual GPU flagship based on the industry’s first HBM GPU code named Fiji. The new premium Radeon graphics card is set for a Q4 launch, in time for the 2015 holiday season. AMD has not as of yet announced the commercial name that will be designated for this particular product, so we’ll refer to it as the R9 Fury X2 in the time being to make it easier for everyone involved.
The R9 Fury X2 will be the fourth and last Fiji based graphics card from AMD this year. The card will feature two full fledged Fiji XT GPUs with 4GB of stacked High Bandwidth Memory each, for a total of 8GB of memory and a combined memory bandwidth of one Terabytes per second. The actual board design is pretty much finalized at this point. AMD engineers are purportedly hard at work optimizing for clock speeds, power and CrossfireX scaling.
They have their work cut out for them for sure as we’re told AMD is gunning for the crown of the world’s fastest graphics card. Which shouldn’t really come as a surprise, looking back at AMD’s previous dual GPU graphics card launches including the R9 295X2, HD 7990, HD 6990 and all the way back to the HD 3870×2 the goal with every one of these launches had always been to claim the title for the world’s fastest graphics card.
This Is What 8GB Of HBM, 1TB/s Of Memory Bandwidth And 17+ Teraflops of Compute Look Like
These are all very impressive figures, but without effective multi-GPU scaling they don’t mean much. And that begs the question, can the upcoming Radeon flagship deliver? AMD had already confirmed during Fiji’s launch event at E3 this past June that the new flagship will be out in the fall of 2015, so it’s not a question of when but how.
Well surprisingly that’s not a difficult question to answer. Because despite having two GPUs on a single board, the graphics card will actually behave just like two separate R9 Fury X cards in a CrossfireX configuration. So all we really have to do to answer that question is look at how well a Fury X – or any Fiji based graphics card – does today in a CrossfireX configuration.
CrossfireX results for the R9 Fury X: At 3840×2160, which is the resolution that we would reasonably expect gamers who’d spend $1300+ on a pair of graphics cards to play games at, two R9 Fury X cards in CrossfireX demonstrated a significant lead in average and/or minimum FPS in 6 out of 7 tested games against two GTX Titan X cards in SLI. Against two GTX 980 Ti graphics cards in SLI, two R9 Fury X cards were faster in all 7 tested titles at 4K.
At 2560×1440 it was a much closer race. The dual R9 Fury X setup comes out ahead in 3 out of 7 games, the dual Titan X setup comes out ahead in 3 other games and both setups come to a dead heat in one out of the seven tested titles. It might seem slightly peculiar at first that Fury X cards in CrossfireX are far more competitive with the Nvidia offerings than the same Fury X cards in single GPU setups. One 980 Ti will often outperform one Fury X at 2560×1440. And while it’s a much closer race between the two cards at 4K, when it comes to multi GPU setups two Fury X cards will often lead two GTX 980 Ti cards in SLI.
The Technology Behind Fury’s Excellent Crossfire Scaling
We’ve seen a similar trend with AMD’s R9 290 series as well and it all comes down to AMD’s CrossfireX DMA engine. It enables faster, lower overhead communication between multiple GPUs, this in turn translates to better CrossfireX scaling. And that’s what we’re seeing take effect here. It also enables GPUs inside multi-GPU setups to inter-communicate through the PCI Express interface which eliminates the need for a CrossfireX bridge.
While Nvidia’s NV-Link technology is targeted towards a different market with very different workloads – mainly for enterprise and HPC applications – the concept itself requires that Nvidia’s upcoming Pascal GPUs are built with the necessary hardware to enable this feature. The same hardware will allow Nvidia to deliver a similar solution to AMD’s XDMA in consumer gaming applications. But until then, AMD’s GCN 1.1 and later GPUs will continue to enjoy an advantage in multi-GPU scaling. An advantage that could allow AMD’s R9 Fury X2 to succeed in securing the frequently sought after “world’s fastest” title.
|WCCFTech||AMD Radeon R9 Fury X2||AMD Radeon R9 Fury X||AMD Radeon R9 Nano||AMD Radeon R9 Fury||AMD Radeon R9 290X|
|GPU||Fiji XT x 2||Fiji XT||Fiji XT||Fiji Pro||Hawaii XT|
|GCN Compute Units||128||64||64||56||44|
|Render Output Units||128||64||64||64||64|
|Texture Mapping Units||512||256||256||224||176|
|GPU Frequency||TBA||Up to 1050Mhz||Up to 1000 MHz||Up to 1000 MHz||1000Mhz|
|Memory||8GB HBM (4 GB Per Chip)||4GB HBM||4GB HBM||4GB HBM||4GB GDDR5|
|Memory Interface||4096-bit x 2||4096bit||4096bit||4096bit||512bit|
|Memory Frequency||500Mhz||500Mhz||500 MHz||500Mhz||1250Mhz|
|Effective Memory Speed||1Gbps||1Gbps||1Gbps||1Gbps||5Gbps|
|Memory Bandwidth||1024 GB/s||512GB/s||512GB/s||512GB/s||320GB/s|
|Cooling||Liquid||Liquid, 120mm Radiator||Air, Single Fan. Custom AIB Solutions Later||Air, Custom AIB Solutions||Air, Single Blower Fan|
|Performance (SPFP)||17.2 TFLOPS||8.6 TFLOPS||8.19 TFLOPS||7.2 TFLOPS||5.6 TFLOPS|
|Power Connectors||Dual 8-Pin||Dual 8-Pin||8-Pin||Dual 8-Pin||6+8 Pin|
|New Prices||TBA||$649||$499||$549||$299 (Retail)|
|Launch Date||Early 2016||24th June 2015||7th September 2015||10th July 2015||24th October 2013|