Intel Graphics Technology[4] (GT)[a] is the collective name for a series of integrated graphics processors (IGPs) produced by Intel that are manufactured on the same package or die as the central processing unit (CPU). It was first introduced in 2010 as Intel HD Graphics and renamed in 2017 as Intel UHD Graphics.
Intel Iris Graphics and Intel Iris Pro Graphics are the IGP series introduced in 2013 with some models of Haswell processors as the high-performance versions of HD Graphics. Iris Pro Graphics was the first in the series to incorporate embedded DRAM.[5] Since 2016 Intel refers to the technology as Intel Iris Plus Graphics with the release of Kaby Lake.
In the fourth quarter of 2013, Intel integrated graphics represented, in units, 65% of all PC graphics processor shipments.[6] However, this percentage does not represent actual adoption as a number of these shipped units end up in systems with discrete graphics cards.
The previous Intel integrated graphics solution, Intel GMA, had a reputation of lacking performance and features, and therefore was not considered to be a good choice for more demanding graphics applications, such as 3D gaming. The performance increases brought by Intel's HD Graphics made the products competitive with integrated graphics adapters made by its rivals, Nvidia and ATI/AMD.[7] Intel HD Graphics, featuring minimal power consumption that is important in laptops, was capable enough that PC manufacturers often stopped offering discrete graphics options in both low-end and high-end laptop lines, where reduced dimensions and low power consumption are important.
Intel HD and Iris Graphics are divided into generations, and within each generation are divided into 'tiers' of increasing performance, denominated by the 'GTx' label. Each generation corresponds to the implementation of a Gen[8] graphics microarchitecture with a corresponding GEN instruction set architecture[9][10][11] since Gen4.[12]
Gen5 architecture
Westmere
In January 2010, Clarkdale and Arrandale processors with Ironlake graphics were released, and branded as Celeron, Pentium, or Core with HD Graphics. There was only one specification:[13] 12 execution units, up to 43.2 GFLOPS at 900 MHz. It can decode a H.264 1080p video at up to 40 fps.
Its direct predecessor, the GMA X4500, featured 10 EUs at 800 MHz, but it lacked some capabilities.[14]
Sandy Bridge Celeron and Pentium have Intel HD, while Core i3 and above have either HD 2000 or HD 3000. HD Graphics 2000 and 3000 include hardware video encoding and HD postprocessing effects.
Gen7 architecture
Ivy Bridge
On 24 April 2012, Ivy Bridge was released, introducing the "third generation" of Intel's HD graphics:[15]
For some low-power mobile CPUs there is limited video decoding support, while none of the desktop CPUs have this limitation. HD P4000 is featured on the Ivy Bridge E3 Xeon processors with the 12X5 v2 descriptor, and supports unbuffered ECC RAM.
Gen7.5 architecture
Haswell
In June 2013, Haswell CPUs were announced, with four tiers of integrated GPUs:
The 128 MB of eDRAM in the Iris Pro GT3e is in the same package as the CPU, but on a separate die manufactured in a different process. Intel refers to this as a Level 4 cache, available to both CPU and GPU, naming it Crystalwell. The Linux drm/i915 driver is aware and capable of using this eDRAM since kernel version 3.12.[16][17][18]
Gen8 architecture
Broadwell
In November 2013, it was announced that Broadwell-K desktop processors (aimed at enthusiasts) would also carry Iris Pro Graphics.[19]
The following models of integrated GPU are announced for Broadwell processors:[20][better source needed]
The Skylake line of processors, launched in August 2015, retires VGA support, while supporting multi-monitor setups of up to three monitors connected via HDMI 1.4, DisplayPort 1.2 or Embedded DisplayPort (eDP) 1.3 interfaces.[21][22]
The following models of integrated GPU are available or announced for the Skylake processors:[23][24][better source needed]
The Apollo Lake line of processors was launched in August 2016.
Model number
CPU model
Tier
Execution units
Shading units
Clock speed (MHz)
HD Graphics 500
E3930
GT1
12
96
400 – 550
E3940
400–600
N3350
200–650
N3450
200–700
J3355
250–700
J3455
250–750
HD Graphics 505
E3950
18
144
500–650
N4200
200–750
J4205
250–800
Gen9.5 architecture
Kaby Lake
The Kaby Lake line of processors was introduced in August 2016. New features: speed increases, support for 4K UHD "premium" (DRM encoded) streaming services, media engine with full hardware acceleration of 8- and 10-bit HEVC and VP9 decode.[26][27]
New features: 10 nm Gen 11 GPU microarchitecture, two HEVC 10-bit encode pipelines, three 4K display pipelines (or 2× 5K60, 1× 4K120), variable rate shading (VRS),[31][32][33] and integer scaling.[34]
While the microarchitecture continues to support double-precision floating-point as previous versions did, the mobile configurations of it do not include the feature and therefore on these it is supported only through emulation.[35]
These are based on the Intel Xe-LP microarchitecture, the low power variant of the Intel Xe GPUarchitecture[40] also known as Gen 12.[41][42] New features include Sampler Feedback,[43] Dual Queue Support,[43] DirectX12 View Instancing Tier2,[43] and AV1 8-bit and 10-bit fixed-function hardware decoding.[44] Support for FP64 was removed.[45]
Intel Quick Sync Video is Intel's hardware video encoding and decoding technology, which is integrated into some of the Intel CPUs. The name "Quick Sync" refers to the use case of quickly transcoding ("syncing") a video from, for example, a DVD or Blu-ray Disc to a format appropriate to, for example, a smartphone. Quick Sync was introduced with the Gen 6 in Sandy Bridge microprocessors on 9 January 2011.
Graphics Virtualization Technology
Graphics Virtualization Technology (GVT) was announced 1 January 2014 and introduced at the same time as Intel Iris Pro. Intel integrated GPUs support the following sharing methods:[53][54]
Direct passthrough (GVT-d): the GPU is available for a single virtual machine without sharing with other machines
Paravirtualized API forwarding (GVT-s): the GPU is shared by multiple virtual machines using a virtual graphics driver; few supported graphics APIs (OpenGL, DirectX), no support for GPGPU
Full GPU virtualization (GVT-g): the GPU is shared by multiple virtual machines (and by the host machine) on a time-sharing basis using a native graphics driver; similar to AMD's MxGPU and Nvidia's vGPU, which are available only on professional line cards (Radeon Pro and Nvidia Quadro)
Full GPU virtualization in hardware (SR-IOV): The gpu can be partitioned and used/shared by multiple virtual machines and the host with support built-in hardware, unlike GVT-g that does this in software(driver).[55]
Gen9 (i.e. Graphics powering 6th through 9th generation Intel processors) is the last generation of the software-based vGPU solution GVT-G (Intel® Graphics Virtualization Technology –g).
SR-IOV (Single Root IO Virtualization) is supported only on platforms with 11th Generation Intel® Core™ "G" Processors (products formerly known as Tiger Lake) or newer. This leaves Rocket Lake (11th Gen Intel Processors) without support for GVT-g and/or SR-IOV. This means Rocket Lake has no full virtualization support.[56] Started from 12th Generation Intel® Core™ Processors, both desktop and laptop Intel CPUs have GVT-g and SR-IOV support.
HD 2500 and HD 4000 GPUs in Ivy Bridge CPUs are advertised as supporting three active monitors, but this only works if two of the monitors are configured identically, which covers many[57] but not all three-monitor configurations. The reason for this is that the chipsets only include two phase-locked loops (PLLs) for generating the pixel clocks timing the data being transferred to the displays.[58]
Therefore, three simultaneously active monitors can only be achieved when at least two of them share the same pixel clock, such as:
Using two or three DisplayPort connections, as they require only a single pixel clock for all connections.[59] Passive adapters from DisplayPort to some other connector do not count as a DisplayPort connection, as they rely on the chipset being able to emit a non-DisplayPort signal through the DisplayPort connector. Active adapters that contain additional logic to convert the DisplayPort signal to some other format count as a DisplayPort connection.
Using two non-DisplayPort connections of the same connection type (for example, two HDMI connections) and the same clock frequency (like when connected to two identical monitors at the same resolution), so that a single unique pixel clock can be shared between both connections.[57]
Another possible three-monitor solution uses the Embedded DisplayPort on a mobile CPU (which does not use a chipset PLL at all) along with any two chipset outputs.[59]
Haswell
ASRock Z87- and H87-based motherboards support three displays simultaneously.[60]Asus H87-based motherboards are also advertised to support three independent monitors at once.[61]
OpenCL 2.1 and 2.2 possible with software update on OpenCL 2.0 hardware (Broadwell+) with future software updates.[93]
Support in Mesa is provided by two Gallium3D-style drivers, with the Iris driver supporting Broadwell hardware and later,[94] while the Crocus driver supports Haswell and earlier.[95] The classic Mesa i965 driver was removed in Mesa 22.0, although it would continue to see further maintenance as part of the Amber branch.[96]
New OpenCL driver is Mesa RustiCL and this driver written in new language Rust is OpenCL 3.0 conformant for Intel XE Graphics with Mesa 22.3. Intel Broadwell and higher will be also conformant to 3.0 with many 2.x features. For Intel Ivy Bridge and Haswell target is OpenCL 1.2. Actual development state is available in mesamatrix.
NEO compute runtime driver supports openCL 3.0 with 1.2, 2.0 and 2.1 included for Broadwell and higher and Level Zero API 1.3 for Skylake and higher.[97]
All GVT virtualization methods are supported since the Broadwell processor family with KVM[98] and Xen.[99]
Intel developed a dedicated SIP core which implements multiple video decompression and compression algorithms branded Intel Quick Sync Video. Some are implemented completely, some only partially.
Hardware-accelerated algorithms
Hardware-accelerated video compression and decompression algorithms present in Intel Quick Sync Video
Intel releases programming manuals for most of Intel HD Graphics devices via its Open Source Technology Center.[118] This allows various open source enthusiasts and hackers to contribute to driver development, and port drivers to various operating systems, without the need for reverse engineering.
^The abbreviation "GT" appears in certain monitoring tools, such as Intel Power Gadget in reference to the graphics core on Intel processors.
^VP9 media codec GPU accelerator to be supported post TTM, for non-Windows operating systems only.
^Resolution details for media codec on open source Linux OS depends on platform features and drivers used. Decode/Encode features may not align to Table 8-4 that is specific to Win8.1 and Win7 operating systems.
^All capabilities dependent on OS. Here HW support is mentioned. For more info, see Table 8-4 on page 80 of PDF.
^Junkins, Stephen (14 August 2015). The Compute Architecture of Intel Processor Graphics Gen9(PDF) (White Paper). Intel. p. 2. Retrieved 9 September 2020. At Intel, architects colloquially refer to Intel processor graphics architecture as simply 'Gen', shorthand for Generation.
^ abMichael Larabel (2011-10-06). "Details On Intel Ivy Bridge Triple Monitor Support". Phoronix. A limitation of this triple monitor support for Ivy Bridge is that two of the pipes need to share a PLL. Ivy Bridge has three planes, three pipes, three transcoders, and three FDI (Flexible Display Interface) interfaces for this triple monitor support, but there's only two pipe PLLs. This means that two of the three outputs need to have the same connection type and same timings. However, most people in a triple monitor environment will have at least two — if not all three — of the monitors be identical and configured the same, so this shouldn't be a terribly huge issue.
^LG Nilsson (2012-03-12). "Most desktop Ivy Bridge systems won't support three displays". VRZone. Archived from the original on 2012-04-01. Despite the fact that Intel has been banging its drums about support for up to three displays on the upcoming 7-series motherboards in combination with a shiny new Ivy Bridge based CPU, this isn't likely to be the case. The simple reason behind this is that very few, if any motherboards will sport a pair of DisplayPort connectors.
^ abDavid Galus (February 2013). "Migration to New Display Technologies on Intel Embedded Platforms"(PDF). Intel. Archived from the original(PDF) on 2013-02-01. The Intel 7 Series Chipset based platform allows for the support of up to three concurrent displays with independent or replicated content. However, this comes with the requirement that either one of the displays is eDP running off the CPU or two DP interfaces are being used off the PCH. When configuring the 2 DP interfaces from the PCH, one may be an eDP if using Port D. This limitation exists because the 7 Series Intel PCH contains only two display PLLs (the CPU has one display PLL also) which will control the clocking for the respective displays. All display types other than DP have an external variable clock frequency associated with the display resolution that is being used. The DP interface has an embedded clocking scheme that is semi- variable, either at 162 or 270 MHz depending on the bandwidth required. Therefore, Intel only allows sharing of a display PLL with DP related interfaces.Alt URL
^"Z87E-ITX". ASRock. This motherboard supports Triple Monitor. You may choose up to three display interfaces to connect monitors and use them simultaneously.
^"H87I-PLUS". Asus. Connect up to three independent monitors at once using video outputs such as DisplayPort, Mini DisplayPort, HDMI, DVI, or VGA. Choose your outputs and set displays to either mirror mode or collage mode.
^Michael Larabel (15 June 2021). "Mesa's New "Crocus" OpenGL Driver Is Performing Well For Old Intel Hardware". Phoronix. Retrieved 2023-07-03. Crocus does allow for OpenGL 4.6 on Haswell compared to OpenGL 4.5 being exposed on the i965 driver. Additionally, Crocus allows for OpenGL ES 3.2 rather than OpenGL ES 3.1 on Haswell. Aside from that the drivers are in similar shape for the most part.