Heterogeneous System Architecture

From Wikipedia, the free encyclopedia
  (Redirected from Heterogenous System Architecture)
Jump to navigation Jump to search

Heterogeneous System Architecture (HSA) is a cross-vendor set of specifications that allow for the integration of central processing units and graphics processors on the same bus, with shared memory and tasks.[1] The HSA is being developed by the HSA Foundation, which includes (among many others) AMD and ARM. The platform's stated aim is to reduce communication latency between CPUs, GPUs and other compute devices, and make these various devices more compatible from a programmer's perspective,[2]:3[3] relieving the programmer of the task of planning the moving of data between devices' disjoint memories (as must currently be done with OpenCL or CUDA).[4]

CUDA and OpenCL as well as most other fairly advanced programming languages can use HSA to increase their execution performance.[5] Heterogeneous computing is widely used in system-on-chip devices such as tablets, smartphones, other mobile devices, and video game consoles.[6] HSA allows programs to use the graphics processor for floating point calculations without separate memory or scheduling.[7]



The rationale behind HSA is to ease the burden on programmers when offloading calculations to the GPU. Originally driven solely by AMD and called the FSA, the idea was extended to encompass processing units other than GPUs, such as other manufacturers' DSPs, as well.

Modern GPUs are very well suited to perform single instruction, multiple data (SIMD) and single instruction, multiple threads (SIMT), while modern CPUs are still being optimized for branching. etc.


Originally introduced by embedded systems such as the Cell Broadband Engine, sharing system memory directly between multiple system actors makes heterogeneous computing more mainstream. Heterogeneous computing itself refers to systems that contain multiple processing units – central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), or any type of application-specific integrated circuits (ASICs). The system architecture allows any accelerator, for instance a graphics processor, to operate at the same processing level as the system's CPU.

Among its main features, HSA defines a unified virtual address space for compute devices: where GPUs traditionally have their own memory, separate from the main (CPU) memory, HSA requires these devices to share page tables so that devices can exchange data by sharing pointers. This is to be supported by custom memory management units.[2]:6–7 To render interoperability possible and also to ease various aspects of programming, HSA is intended to be ISA-agnostic for both CPUs and accelerators, and to support high-level programming languages.

So far, the HSA specifications cover:

HSA Intermediate Layer[edit]

HSA Intermediate Layer (HSAIL), a virtual instruction set for parallel programs

HSA memory model[edit]

  • compatible with C++11, OpenCL, Java and .NET memory models
  • relaxed consistency
  • designed to support both managed languages (e.g. Java) and unmanaged languages (e.g. C)
  • will make it much easier to develop 3rd-party compilers for a wide range of heterogeneous products programmed in Fortran, C++, C++ AMP, Java, et al.

HSA dispatcher and run-time[edit]

  • designed to enable heterogeneous task queueing: a work queue per core, distribution of work into queues, load balancing by work stealing
  • any core can schedule work for any other, including itself
  • significant reduction of overhead of scheduling work for a core

Mobile devices are one of the HSA's application areas, in which it yields improved power efficiency.[6]

Block diagrams[edit]

The block diagrams below provide high-level illustrations of how HSA operates and how it compares to traditional architectures.

Software support[edit]

AMD GPUs contain certain additional functional units intended to be used as part of HSA. In Linux, kernel driver amdkfd provides required support.[9][10]

Some of the HSA-specific features implemented in the hardware need to be supported by the operating system kernel and specific device drivers. For example, support for AMD Radeon and AMD FirePro graphics cards, and APUs based on Graphics Core Next (GCN), was merged into version 3.19 of the Linux kernel mainline, released on 8 February 2015.[10] Programs do not interact directly with amdkfd, but queue their jobs utilizing the HSA runtime.[11] This very first implementation, known as amdkfd, focuses on "Kaveri" or "Berlin" APUs and works alongside the existing Radeon kernel graphics driver.

Additionally, amdkfd supports heterogeneous queuing (HQ), which aims to simplify the distribution of computational jobs among multiple CPUs and GPUs from the programmer's perspective. Support for heterogeneous memory management (HMM), suited only for graphics hardware featuring version 2 of the AMD's IOMMU, was accepted into the Linux kernel mainline version 4.14.[12]

Integrated support for HSA platforms has been announced for the "Sumatra" release of OpenJDK, due in 2015.[13]

AMD APP SDK is AMD's proprietary software development kit targeting parallel computing, available for Microsoft Windows and Linux. Bolt is a C++ template library optimized for heterogeneous computing.[14]

GPUOpen comprehends a couple of other software tools related to HSA. CodeXL version 2.0 includes an HSA profiler.[15]

Hardware support[edit]


As of February 2015, only AMD's "Kaveri" A-series APUs (cf. "Kaveri" desktop processors and "Kaveri" mobile processors) and Sony's PlayStation 4 allowed the integrated GPU to access memory via version 2 of the AMD's IOMMU. Earlier APUs (Trinity and Richland) included the version 2 IOMMU functionality, but only for use by an external GPU connected via PCI Express.[citation needed]

Post-2015 Carrizo and Bristol Ridge APUs also include the version 2 IOMMU functionality for the integrated GPU.[citation needed]

The following table shows features of AMD's APUs (see also: List of AMD accelerated processing units).

Codename Server Toronto Kyoto
Desktop Llano Trinity Richland Kaveri Carrizo Bristol Ridge Raven Ridge Picasso Kabini
Mobile Llano Trinity Richland Kaveri Carrizo Bristol Ridge Dalí Raven Ridge Picasso Renoir Desna, Ontario, Zacate Kabini, Temash Beema, Mullins Carrizo-L Stoney Ridge
Embedded Trinity Bald Eagle Merlin Falcon,
Brown Falcon
Banded Kestrel Great Horned Owl Ontario, Zacate Kabini Steppe Eagle, Crowned Eagle,
Prairie Falcon
Platform High, standard and low power Low and ultra-low power
Released Aug 2011 Oct 2012 Jun 2013 Jan 2014 Jun 2015 Jun 2016 Apr 2019 Oct 2017 Jan 2019 2020 Jan 2011 May 2013 Apr 2014 May 2015 Feb 2016
CPU microarchitecture K10 Piledriver Steamroller Excavator "Excavator+"[16] Zen Zen+ Zen 2 Bobcat Jaguar Puma Puma+[17] "Excavator+"
Socket Desktop FM1 FM2 FM2+ FM2+[a], AM4 AM4 AM4 N/A AM1 N/A
Other FS1 FS1+, FP2 FP3 FP4 FP5 FP6 FT1 FT3 FT3b FP4
PCI Express version 2.0 3.0 2.0 3.0
Fab. (nm) GlobalFoundries 32SHP
GlobalFoundries 28SHP
(HKMG bulk)
GlobalFoundries 14LPP
(FinFET bulk)
GlobalFoundries 12LP
(FinFET bulk)
(FinFET bulk)
(HKMG bulk)
GlobalFoundries 28SHP
(HKMG bulk)
die area (mm2) 228 246 245 245 250 210[18] 210 149 75 (+ 28 FCH) 107 ? 125
Min TDP 35W 17W 12W 4.5W 4W 3.95W 10W 6W
Max APU TDP 100W 95W 65W 25W 65W 45W 18W 25W
Max stock APU base clock (GHz) 3 3.8 4.1 3.7 3.8 3.3 3.6 3.7 3 1.75 2.2 2 2.2 3.2
Max APU cores 4 2 4 8 2 4 2
Max threads per core 1 2 1
Integer structure 3+3 2+2 4+2 4+2+1 1+1+1+1 2+2
FPUs per core 1 0.5 1 1 0.5
FPU pipe width 128-bit 256-bit 64-bit 128-bit
CPU instruction set SIMD level SSE3 AVX AVX2 SSSE3 AVX AVX2
L1 data cache per core (KiB) 64 16 32 32
L1 data cache associativity (ways) 2 4 8 8
L1 instruction caches per core 1 0.5 1 1 0.5
Max APU total L1 instruction cache (KiB) 256 128 192 128 256 64 128 96
L1 instruction cache associativity (ways) 2 3 4 8 2 3
L2 caches per core 1 0.5 1 1 0.5
Max APU total L2 cache (MiB) 4 2 1 2 4 1 2 1
L2 cache associativity (ways) 16 8 16
APU total L3 cache (MiB) N/A 4 8 N/A
APU L3 cache associativity (ways) 16
L3 cache scheme victim N/A victim
Max stock DRAM support DDR3-1866 DDR3-2133 DDR3-2133, DDR4-2400 DDR4-2400 DDR4-2933 DDR4-3200, LPDDR4-4266 DDR3L-1333 DDR3L-1600 DDR3L-1866 DDR3-1866, DDR4-2400
DRAM channels 2 1
Max stock DRAM bandwidth (GB/s) 29.866 34.132 38.400 46.932 68.256 10.666 12.800 14.933 19.200
GPU microarchitecture TeraScale 2 (VLIW5) TeraScale 3 (VLIW4) GCN 2nd gen GCN 3rd gen GCN 5th gen[19] TeraScale 2 (VLIW5) GCN 2nd gen GCN 3rd gen[19]
GPU instruction set TeraScale instruction set GCN instruction set TeraScale instruction set GCN instruction set
Max stock GPU base clock (MHz) 600 800 844 866 1108 1100 1250 1400 1750 538 600 ? 847 900
Max stock GPU base GFLOPS[b] 480 614.4 648.1 886.7 1134.5 460.8 1760 1971.2 1792 86 ? ? ? 345.6
3D engine[c] Up to 400:20:8 Up to 384:24:6 Up to 512:32:8 Up to 192:?:? Up to 704:44:16[20] Up to 512:?:? 80:8:4 128:8:4 Up to 192:?:?
Video decoder UVD 3.0 UVD 4.2 UVD 6.0 VCN 1.0[21] UVD 3.0 UVD 4.0 UVD 4.2 UVD 6.0 UVD 6.3
Video encoder N/A VCE 1.0 VCE 2.0 VCE 3.1 N/A VCE 2.0 VCE 3.1
GPU power saving PowerPlay PowerTune PowerPlay PowerTune[22]
TrueAudio N/A Yes[23] N/A Yes
FreeSync 1
HDCP[d] ? 1.4 1.4
? 1.4
PlayReady[d] N/A 3.0 not yet N/A
Supported displays[e] 2–3 2–4 3 4 3 (desktop)
4 (mobile)
2 3
/drm/radeon[f][25][26] Yes N/A Yes N/A
/drm/amdgpu[f][27] N/A Yes[28] Yes N/A Yes[28] Yes
  1. ^ APU models: A8-7680, A6-7480. CPU only: Athlon X4 845.
  2. ^ Single-precision performance is calculated from the base (or boost) core clock speed based on a FMA operation.
  3. ^ Unified shaders : texture mapping units : render output units
  4. ^ a b To play protected video content, it also requires card, operating system, driver, and application support. A compatible HDCP display is also needed for this. HDCP is mandatory for the output of certain audio formats, placing additional constraints on the multimedia setup.
  5. ^ To feed more than two displays, the additional panels must have native DisplayPort support.[24] Alternatively active DisplayPort-to-DVI/HDMI/VGA adapters can be employed.
  6. ^ a b DRM (Direct Rendering Manager) is a component of the Linux kernel. Support in this table refers to the most current version.


ARM's Bifrost microarchitecture, as implemented in the Mali-G71,[29] is fully compliant with the HSA 1.1 hardware specifications. As of June 2016, ARM has not announced software support that would use this hardware feature.

See also[edit]


  1. ^ Tarun Iyer (30 April 2013). "AMD Unveils its Heterogeneous Uniform Memory Access (hUMA) Technology". Tom's Hardware.
  2. ^ a b George Kyriazis (30 August 2012). Heterogeneous System Architecture: A Technical Review (PDF) (Report). AMD. Archived from the original (PDF) on 28 March 2014. Retrieved 26 May 2014.
  3. ^ "What is Heterogeneous System Architecture (HSA)?". AMD. Archived from the original on 21 June 2014. Retrieved 23 May 2014.
  4. ^ Joel Hruska (26 August 2013). "Setting HSAIL: AMD explains the future of CPU/GPU cooperation". ExtremeTech. Ziff Davis.
  5. ^ Linaro. "LCE13: Heterogeneous System Architecture (HSA) on ARM". slideshare.net.
  6. ^ a b "Heterogeneous System Architecture: Purpose and Outlook". gpuscience.com. 9 November 2012. Archived from the original on 1 February 2014. Retrieved 24 May 2014.
  7. ^ "Heterogeneous system architecture: Multicore image processing using a mix of CPU and GPU elements". Embedded Computing Design. Retrieved 23 May 2014.
  8. ^ "Kaveri microarchitecture". SemiAccurate. 15 January 2014.
  9. ^ Michael Larabel (21 July 2014). "AMDKFD Driver Still Evolving For Open-Source HSA On Linux". Phoronix. Retrieved 21 January 2015.
  10. ^ a b "Linux kernel 3.19, Section 1.3. HSA driver for AMD GPU devices". kernelnewbies.org. 8 February 2015. Retrieved 12 February 2015.
  11. ^ "HSA-Runtime-Reference-Source/README.md at master". github.com. 14 November 2014. Retrieved 12 February 2015.
  12. ^ "Linux Kernel 4.14 Announced with Secure Memory Encryption and More". 13 November 2017.
  13. ^ Alex Woodie (26 August 2013). "HSA Foundation Aims to Boost Java's GPU Prowess". HPCwire.
  14. ^ "Bolt on github".
  15. ^ AMD GPUOpen (19 April 2016). "CodeXL 2.0 includes HSA profiler".
  16. ^ "AMD Announces the 7th Generation APU: Excavator mk2 in Bristol Ridge and Stoney Ridge for Notebooks". 31 May 2016. Retrieved 3 January 2020.
  17. ^ "AMD Mobile "Carrizo" Family of APUs Designed to Deliver Significant Leap in Performance, Energy Efficiency in 2015" (Press release). 20 November 2014. Retrieved 16 February 2015.
  18. ^ "The Mobile CPU Comparison Guide Rev. 13.0 Page 5 : AMD Mobile CPU Full List". TechARP.com. Retrieved 13 December 2017.
  19. ^ a b "AMD VEGA10 and VEGA11 GPUs spotted in OpenCL driver". VideoCardz.com. Retrieved 6 June 2017.
  20. ^ Cutress, Ian (1 February 2018). "Zen Cores and Vega: Ryzen APUs for AM4 – AMD Tech Day at CES: 2018 Roadmap Revealed, with Ryzen APUs, Zen+ on 12nm, Vega on 7nm". Anandtech. Retrieved 7 February 2018.
  21. ^ Larabel, Michael (17 November 2017). "Radeon VCN Encode Support Lands in Mesa 17.4 Git". Phoronix. Retrieved 20 November 2017.
  22. ^ Tony Chen; Jason Greaves, "AMD's Graphics Core Next (GCN) Architecture" (PDF), AMD, retrieved 13 August 2016
  23. ^ "A technical look at AMD's Kaveri architecture". Semi Accurate. Retrieved 6 July 2014.
  24. ^ "How do I connect three or More Monitors to an AMD Radeon™ HD 5000, HD 6000, and HD 7000 Series Graphics Card?". AMD. Retrieved 8 December 2014.
  25. ^ Airlie, David (26 November 2009). "DisplayPort supported by KMS driver mainlined into Linux kernel 2.6.33". Retrieved 16 January 2016.
  26. ^ "Radeon feature matrix". freedesktop.org. Retrieved 10 January 2016.
  27. ^ Deucher, Alexander (16 September 2015). "XDC2015: AMDGPU" (PDF). Retrieved 16 January 2016.
  28. ^ a b Michel Dänzer (17 November 2016). "[ANNOUNCE] xf86-video-amdgpu 1.2.0". lists.x.org.
  29. ^ "ARM Bifrost GPU Architecture". 30 May 2016.

External links[edit]