Intel has announced its latest data center GPUs in the form of the Data Center GPU Flex Series. Formerly code-named Arc Sound-M, Intel says that these new GPUs are capable of delivering “5x media transcode throughput performance and up to 68 simultaneous cloud gaming streams” and represent the industry’s most open GPU solution for the intelligent visual cloud. The Flex Series are the first GPUs to feature a hardware-based AV1 encoder, according to Intel, enabling “more than 30% bandwidth savings.”
“We are in the midst of a pixel explosion driven by more consumers, more applications and higher resolutions,” said Jeff McVeigh, Intel vice president and general manager of the Super Compute Group. “Today’s data center infrastructure is under intense pressure to compute, encode, decode, move, store and display visual information. Intel Flex Series GPU is a breakthrough design that uniquely solves today’s computing demand while providing flexibility and scalability for the immersive experiences of tomorrow.”
On media processing and delivery:
- The Flex Series GPU media architecture, powered by up to four Xe Media Engines, focuses on the highest stream density demands for the modern data center without compromising quality:
- Delivers five times media transcode throughput, two times decode throughput at half the power (Intel Flex Series 140 GPU compared to Nvidia A10).
- Delivers up to 36 streams 1080p60 transcode throughput per card.
- Delivers eight streams 4K60 transcode throughput per card.
- When scaled to 10 cards in a 4U server configuration, it can support up to 360 streams HEVC-HEVC 1080p60 transcode throughput.
- The Flex Series GPU’s hardware-based, open-source AV1 encoder delivers more than 30% bandwidth savings. The Flex Series GPU’s media engine also supports and improves performance for broad range of industry codecs, including HEVC, AVC and VP9.
- Leveraging Intel Deep Link Hyper Encode feature, the Intel Data Center GPU Flex Series 140 with two devices on a single card can meet the industry’s one-second delay requirement while providing 8K60 real-time transcode7. This capability is available for AV1 and HEVC HDR formats.
On Android cloud gaming:
- A single Flex Series 170 GPU can achieve up to 68 streams of 720p30 while a single Flex Series 140 GPU can achieve up to 46 streams of 720p30 (measured on select game titles).
- When scaled with six Flex Series 140 GPU cards, it can achieve up to 216 streams of 720p30.
Introducing Intel Data Center GPU Flex Series for the Intelligent Visual Cloud
What’s New: The Intel Data Center GPU Flex Series (formerly code-named Arctic Sound-M) helps free customers from the constraints of siloed and proprietary environments and reduces the need for data centers to use separate, discrete solutions. Intel offers customers a single graphics processing (GPU) solution built to flexibly handle a wide range of workloads without compromising on performance or quality. It also helps lower and optimizes the total cost of ownership for diverse cloud workloads like media delivery, cloud gaming, AI, metaverse and other emerging visual cloud use cases.
What They Offer: The Flex Series GPUs address quality, density and latency requirements. With the industry’s first hardware-based AV1 encoder in a data center GPU, the Flex Series GPU provides five times media transcode throughput performance and two times decode throughput performance at half the power of competitive solutions. It delivers more than 30 percent bandwidth improvement for significant total cost of ownership (TCO) savings, in addition to broad support for popular media tools, APIs, frameworks and the latest codecs.
Powered by Intel’s Xe-HPG architecture and backed by an expansive ecosystem of hardware vendors and software developers, these GPUs provide flexible scaling of AI inference workloads from media analytics to smart cities to medical imaging between CPUs and GPUs without locking developers into proprietary software.
Why It’s Important: Media processing, media delivery, AI visual inference, cloud gaming and desktop virtualization are proliferating in data centers. As a result, the data center GPU silicon market segment for visual cloud is projected to grow to reach $15 billion by 2026. That rapid increase comes to an industry largely beset by dependence on proprietary, licensed coding models, like CUDA for GPU programming.
The Flex Series GPU solution stack overcomes these limitations while delivering significant advantages over alternative solutions, improving flexibility, scalability and power consumption. It helps deliver a lower TCO for solution providers by supporting more subscribers with fewer servers.
The software stack is enabled by oneAPI – the productive path for accelerated computing – freeing developers from the economic and technical burdens of proprietary programming models. It’s an open alternative to proprietary language lock-in that enables the full performance of the hardware with a complete, proven set of tools that complement existing languages and parallel models. This allows users to develop open, portable code that will take maximum advantage of various combinations across Intel CPUs and GPUs.
About Performance: From the start, the Intel Data Center GPU Flex Series will offer capabilities for media processing and delivery and Android cloud gaming.
About the Open Software Stack: Developers can access a comprehensive, software stack that combines open-source components and tools to effectively realize the Flex Series GPU capabilities for visual cloud workloads including support for oneAPI and OpenVINO. Intel’s oneAPI tools empower developers to deliver accelerated applications and services, including the Intel oneAPI Video Processing Library (oneVPL), Intel VTune Profiler and others.
About Availability: Systems featuring Flex Series GPUs will be available from global providers including Dell Technologies, HPE, H3C, Inspur, Lenovo and Supermicro. Solutions with the Flex Series GPU will ramp over the coming months, starting with media delivery and Android cloud gaming workloads. These will be followed by Windows cloud gaming, AI and VDI workloads.