AMD Vega 10 GPU Based Radeon Vega Graphics Card Pictured – Full Length, Dual Slot Card Featuring 8 GB of HBM2 VRAM

The AMD Vega 10 GPU based Radeon Vega graphics which was demonstrated at multiple events including CES has been pictured in a video published by LinusTechTips. The upcoming graphics card was seen running in a PC alongside the Ryzen CPU and running DOOM at 4K resolution.

AMD Vega 10 GPU Based Radeon Vega Graphics Card Pictured

The Radeon Vega graphics card pictured is still in an engineering state as the product itself is months away from consumer debut. However, there are still many interesting things that Linus and Raja Koduri talk about in the video that can be seen below.

AMD Bids Farewell To GCN Architecture, Ends Driver Support For Radeon 7000, 200, 300 & Fury Series Graphics Cards

First up, we know that AMD Vega 10 graphics cards will be available in consumer, workstation and server variants. The consumer variants will be branded Radeon while the professional cards feature the new FirePro WX and Instinct branding. The engineering sample which was running at CES '17 utilized a longer version of the cooler shroud featured on Radeon RX 480 and it looks really neat for the graphics card itself.

The PC had all its vents covered to prevent the press from sneaking their cameras inside the PC case and taking pictures of the Ryzen and Vega samples. Linus however, managed to get special permission from AMD and took off the side panel to reveal an interesting Vega 10 based Radeon graphics card.

AMD Vega 10 Based Radeon Vega Sample Graphics Board Speculation

Other things to note is that the PCB on the back remains bare. The card utilizes a blower style cooler and has a dual slot body. The Radeon logo on the side has red LEDs which light up when the card is running. The cooling assembly has various parts, the heatsink block is attached above the GPU and held in place through a metal bracket which can be viewed on the back. The cooler shroud is held by several screws that are located on the PCB.

Also, the PCB doesn't has any traces for GDDR5 memory as all the memory is held on the interposer itself since it's HBM2 based. The power connections are still a secret as AMD decided to put tape on the input ports. The reason they did this is because the Vega sample board is not indicative of the final GPU and graphics board. So it's there to avoid any confusion between sample and retail boards.

AMD Launches Radeon Adrenalin 2020 21.6.1 Driver With FSR ‘FidelityFX Super Resolution’ Support In Select Games

AMD also said that they would like their AIB partners to prepare small ITX boards based on Vega so it's likely we may not see an R9 Nano-esque product from AMD themselves in the coming Radeon Vega lineup. Another interesting thing to note is that there's a second PCB at the end of the card which features a USB port. This port will not be included in the consumer version and Raja Koduri gave a pretty good explanation of the purpose of this port.

The port gives engineers at AMD the ability to monitor all the signals, most importantly power and electrical signals coming out of this GPU, remotely and with no overhead. Currently, all graphics cards with the VEGA 10 GPUs running at AMD HQ feature this port so they can determine the quality (performance per watt) of the final product.

AMD Vega 10 GPU Specifications:

AMD's high-end Vega 10 GPU will be available to consumers in the first half of 2017. The chip spans a die size of over 500mm² from early calculations and features two HBM2 stacks, incorporating up to 16 GB of HBM2. The consumer variant which was demonstrated using DOOM and Star Wars: Battlefront featured 8 GB of HBM2 VRAM. The specific device ID for the consumer variant is 687F:C1.

The graphics chip will be utilizing the latest 14nm GFX9 core architecture which is based on the NCU (Next Compute Engine) design. The graphics card will feature 64 Compute Units or 4096 stream processors. AMD plans on increasing the throughput of the chip through increased clock speeds. This will allow AMD to pump numbers better than the Fiji GPU which is based on 28nm GCN 3.0 architecture and comes with the same number of cores, e.g. 4096 SPs.

The server part with the full chip is expected to feature a TDP of 225W with clock speeds around 1526 MHz. A consumer oriented graphics card can feature even higher clock speeds since server parts generally lack the cooling capabilities of consumer level cards which ship with better coolers and PCBs designed to allow overclocking of the GPUs. You can read our Vega GPU architecture preview over here.

AMD Radeon Vega Official Architecture Slides:

AMD Vega Lineup

Graphics Card	Radeon R9 Fury X	Radeon RX 480	Radeon RX Vega Frontier Edition	Radeon Vega Pro	Radeon RX Vega (Gaming)	Radeon RX Vega Pro Duo
GPU	Fiji XT	Polaris 10	Vega 10	Vega 10	Vega 10	2x Vega 10
Process Node	28nm	14nm FinFET	FinFET	FinFET	FinFET	FinFET
Stream Processors	4096	2304	4096	3584	4096 (?)	Up to 8192
Performance	8.6 TFLOPS 8.6 (FP16) TFLOPS	5.8 TFLOPS 5.8 (FP16) TFLOPS	~13 TFLOLPS ~25 (FP16) TFLOPS	11 TFLOLPS 22 (FP16) TFLOPS	>13 TFLOLPS >25 (FP16) TFLOPS	TBA TBA
Memory	4GB HBM	8GB GDDR5	16GB HBM2	TBA	TBA	TBA
Memory Bus	4096-bit	256-bit	2048-bit	2048-bit	2048-bit	4096-bit
Bandwidth	512GB/s	256GB/S	480GB/S	400GB/S	TBA	TBA
TDP	275W	150W	TBA	TBA	TBA	TBA
Launch	2015	2016	June 2017	June 2017	July 2017	TBA

AMD Vega 10 Memory Specifications:

The first generation HBM graphics cards such as the Radeon R9 Fury X was limited to just 4 GB of VRAM and had a bandwidth of 512 GB/s. It had 4-layers per stack (256 MB per layer) and that will continue with the latest Vega GPUs since AMD will have to maximize value on these cards for the gaming audience. In the case of 4-layers, we will be looking at higher densities per layer. The pin speed also increases with HBM2. The new memory standard can clock up to 2 Gb/s compared to 1 Gb/s on HBM1.

The increased clock speed would allow the same memory bandwidth as four HBM1 stacks on just two HBM2 stacks. The increased density also allows AMD to cut down the costs in designing larger interposers. HBM2 itself takes more space compared to HBM1 with a die size around 92mm2 while HBM1 was just 35mm2 in size.

AMD Vega 10 and NVIDIA Pascal GP100 GPU Specs Comparison:

GPU Architecture	NVIDIA Pascal	AMD Vega
Product Market	Tesla (P100.2)	Instinct MI25
GPU Process	16nm FinFET	14nm FinFET
Flagship Chip	GP100 GPU	Vega 10 GPU
GPU Design	SMP (Streaming Multiprocessor Pascal)	NCU (Next Compute Unit)
Maximum Transistors	15.3 Billion	TBD
Maximum Die Size	610mm2	500-540mm2
Maximum Cores	3840 CUDA Cores	4096 Stream Processors
FP16 Compute	~24.0 TFLOPs	25.0 TFLOPs
FP32 Compute	~12.0 TFLOPs	12.5 TFLOPs
FP64 Compute	~6.00 TFLOPs	0.75 TFLOPs
Maximum VRAM	16 GB HBM2	16 GB HBM2 (High Bandwidth Cache and Controller)
Maximum Bandwidth	720 GB/s	512 GB/s
Maximum TDP	300W
Launch Year	Q2 2016	1H 2017

We have seen multiple previews of the Radeon Vega 10 graphics card in Ashes of The Singularity, Star Wars: Battlefront and DOOM. The graphics card is looking to be a killer 4K AAA gaming product and we can't wait for it to launch in Summer 2017.