Graphics Core Next

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Graphics Core Next (GCN) is the codename for a family of graphics processing unit microarchitectures developed by AMD and launched in 2011[1] as the successor to TeraScale.

GCN is used in 28 nm graphics chips in the HD 7000, HD 8000 and Rx 200 series. GCN is also used in AMD Accelerated Processing Units code-named "Temash", "Kabini", "Kaveri", "Beema" and "Mullins", as well as in Liverpool (PlayStation 4) and Durango (Xbox One).

GCN uses a RISC SIMD architecture rather than the VLIW SIMD architecture of TeraScale, which it replaced.[citation needed] GCN implements HyperZ.[2]

GCN 1.0 (Southern Islands, 7000/200 Series)[edit]

  • AMD's Graphics Core Next features a RISC SIMD architecture; it replaces AMD's TeraScale family of microarchitectures (all VLIW SIMD architectures) used since the Radeon HD 2000 Series.[3] The new implementation of SIMD requires considerably more transistors than the old one of VLIW, but offers advantages for GPGPU computation and also leads to better utilization[citation needed]
  • Support for 64-bit addressing (x86-64 address space) with unified address space for CPU and GPU[4]
  • Support for Partially Resident Textures,[6] which enable virtual memory support through DirectX and OpenGL extensions
  • AMD PowerTune support, which dynamically adjusts performance to stay within a specific TDP
  • Support for Mantle (API)

GCN 1.0 combines every 64 shader processor with 4 TMUs and 1 ROP to a compute unit (CU). Each CU has its own Asynchronous Compute Engine (ACE) controlling computation and dispatching for its CU.[7][8]

ZeroCore Power[edit]

AMD introduced ZeroCore Power, a long idle power saving technology with GCN 1.0.[9] AMD ZeroCore Power technology supplements AMD PowerTune.

Unified virtual memory[edit]

In a preview in 2011, AnandTech wrote about the unified virtual memory, supported by Graphics Core Next.[4]
Classical desktop computer architecture with a distinct graphics card over PCI Express. CPU and GPU have their distinct physical memory, with different address spaces. The entire data needs to be copied over the PCIe bus. Note: the diagram shows bandwidths, but not the memory latency. 
GCN supports "unified virtual memory", hence enabling zero-copy, instead of the data, only the pointers are copied, "passed". This is a paramount HSA feature. 
Integrated graphics-solutions (and AMD APUs with TeraScale graphics) suffer under partitioned main memory: a part of the system memory is allocated to the GPU exclusively. Zero-copy is not possible, data has to be copied (over the system memory bus) from one partition to the other. 
AMD APUs with GCN graphics gain from unified main memory conserving scarce bandwidth.[10] 

Heterogeneous System Architecture (HSA)[edit]

GCN includes special purpose function blocks to be used by HSA. Support for these function blocks is available through amdkfd since Linux kernel 3.19.[11]

Some of the specific HSA-features implemented in the hardware need support from the operating system's kernel (its subsystems) and/or from specific device drivers. For example, in July 2014 AMD published a set of 83 patches to be merged into Linux kernel mainline 3.17 for supporting their Graphics Core Next-based Radeon graphics cards. The special driver titled "HSA kernel driver" resides in the directory /drivers/gpu/hsa while the DRM-graphics device drivers reside in /drivers/gpu/drm[12] and augments the already existent DRM driver for Radeon cards.[13] This very first implementation focuses on a single "Kaveri" APU or on a "Berlin" APU and works alongside the existing Radeon kernel graphics driver (kgd).

GCN 1.1 (Sea Islands, R9 290/260 Series)[edit]

Graphics Core Next 1.1 was introduced with Radeon HD 7790 and is also found in Steamroller-based Desktop Kaveri APUs and Mobile Kaveri APUs and in the Puma-based "Beema" and "Mullins" APUs. It has multiple advantages over the original GCN 1.0, including AMD TrueAudio and a revised version of AMD's Powertune technology.

The A10-7850K "Kaveri" contains 8 CUs (compute units) and 8 Asynchronous Compute Engines (ACEs) for independent scheduling and work item dispatching.[14]

At AMD Developer Summit (APU) in November 2013 Michael Mantor presented the Radeon R9-290X.[15]

GCN 1.2 (Volcanic Islands, "R9 285")[edit]

Graphics Core Next 1.2 was introduced with the Radeon R9 285 which has the "Tonga" GPU. It features improved tessellation performance, lossless delta color compression in order to reduce memory bandwidth usage, an updated and more efficient instruction set, a new high quality scaler for video, and a new multimedia engine (video encoder/decoder).

See also[edit]

List of AMD Graphics Processing Units

References[edit]

  1. ^ "AMD Launches World’s Fastest Single-GPU Graphics Card – the AMD Radeon HD 7970" (Press release). AMD. 2011-12-22. Retrieved 2015-01-20. AMD [...] today launched the AMD Radeon HD 7970 graphics card for desktop PCs, introducing the world's fastest single GPU graphics card and the only GPU based on 28nm production technology.[...] With the arrival of the AMD Radeon HD 7970, AMD has unleashed its revolutionary new Graphics Core Next Architecture that enables new levels of gaming and compute capabilities – realizing an improvement of over 150% in performance/sq mm over the prior generation. 
  2. ^ "Feature matrix of the free and open-source "Radeon" graphics device driver". Retrieved 2014-07-09. 
  3. ^ "AMD GCN microarchitecture – whitepaper". AMD. 
  4. ^ a b "t the unified address space that will be used is the x86-64 address space". AnandTech. 2011-12-21. Retrieved 2014-07-11. 
  5. ^ "AMD Radeon HD 7000 Series to be PCI-Express 3.0 Compliant". TechPowerUp. Retrieved July 21, 2011. 
  6. ^ "AMD Details Next Gen. GPU Architecture". Retrieved August 3, 2011. 
  7. ^ Mantor, Michael; Houston, Mike (2011-06-15). "AMD Graphics Core Next" (pdf). AMD. p. 40. Retrieved 2014-07-15. Asynchronous Compute Engine (ACE) 
  8. ^ "AMD's Graphics Core Next Preview: AMD's New GPU, Architected For Compute". AnandTech. 2011-12-21. Retrieved 2014-07-15. AMD's new Asynchronous Compute Engines serve as the command processors for compute operations on GCN. The principal purpose of ACEs will be to accept work and to dispatch it off to the CUs for processing. 
  9. ^ "Managing Idle Power: Introducing ZeroCore Power". AnandTech. 2011-12-22. 
  10. ^ "Kaveri microarchitecture". SemiAccurate. 2014-01-15. 
  11. ^ Dave Airlie (2014-11-26). "Merge AMDKFD". freedesktop.org. Retrieved 2015-01-21. 
  12. ^ "/drivers/gpu/drm". kernel.org. 
  13. ^ "[PATCH 00/83] AMD HSA kernel driver". LKML. 2014-07-10. Retrieved 2014-07-11. 
  14. ^ "AMD's Kaveri A10-7850K tested". AnandTech. 2014-01-14. Retrieved 2014-07-07. 
  15. ^ "AMD Radeon R9-290X". 2013-11-21. 
 
Fixed pipeline
Unified shaders
TeraScale
Unified shaders & memory
GCN
 
Current technologies and software
Audio/Video acceleration
GPU technologies
Software
 
Other brands and products
Workstations & Supercomputers
Consoles
Retrieved from "http://en.wikipedia.org/w/index.php?title=Graphics_Core_Next&oldid=646948993"