Intel Arc is a brand of graphics processing units designed by Intel. These are discrete GPUs mostly marketed for the high-margin gaming PC market. The brand also covers Intel's consumer graphics software and services.
Arc competes with Nvidia's GeForce and AMD's Radeon lines.[2] The Arc-A series for laptops was launched on March 30, 2022, with the A750 and A770 both released in Q3 2022.[3][4][5] Intel missed their initial Q2 2022 release target, with most discrete Arc GPUs not launching until October 2022.[6]
Intel officially launched the Arc Pro workstation GPUs on August 8, 2022.[7][8]
Intel officially announced the 2nd-generation Battlemage GPUs on December 3rd, 2024.[9][10] The first model, the B580, was released on December 13th, 2024.
Etymology
According to Intel, the brand is named after the concept of story arcs found in video games.[11] Each generation of Arc is named after character classes sorted by each letter of the Latin alphabet in ascending order. They begin with A, then B, then C, and so on. The first generation is named Alchemist, while Battlemage, Celestial and Druid are the respective names for the second, third and fourth Arc generations.[12]
Graphics processor generations
Alchemist
Developed under the previous codename "DG2", the first generation of Intel Arc GPUs (codenamed "Alchemist") released on March 30, 2022.[1][13] It comes in both add-on desktop card and laptop form factors. TSMC manufactures the die, using their N6 process.[14]
Intel confirmed ASTC support has been removed from hardware starting with Alchemist and future Arc GPU microarchitectures will also not support it.[17]
Arc Alchemist does not support SR-IOV[18] or Direct3D 9 natively, instead falling back on the D3D9On12 wrapper which translates Direct3D 9 calls to their Direct3D 12 equivalents.[19][20]
^ Shading cores (ALU): texture mapping units (TMU): render output units (ROP): ray tracing units (tensor cores (XMX): execution units: render slices)
^Texture fillrate is calculated as the number of texture mapping units (TMUs) multiplied by the base (or boost) core clock speed.
^ Boost values (if available) are stated below the base value in italic.
^Pixel fillrate is calculated as the lowest of three numbers: number of ROPs multiplied by the base core clock speed, number of rasterizers multiplied by the number of fragments they can generate per rasterizer multiplied by the base core clock speed, and the number of streaming multiprocessors multiplied by the number of fragments per clock that they can output multiplied by the base clock rate.
^ Shading cores (ALU): texture mapping units (TMU): render output units (ROP): ray tracing units (tensor cores (XMX): execution Units: render slices)
^ Boost values (if available) are stated below the base value in italic.
^ Pixel fillrate is calculated as the lowest of three numbers: number of ROPs multiplied by the base core clock speed, number of rasterizers multiplied by the number of fragments they can generate per rasterizer multiplied by the base core clock speed, and the number of streaming multiprocessors multiplied by the number of fragments per clock that they can output multiplied by the base clock rate.
^ Texture fillrate is calculated as the number of texture mapping units (TMUs) multiplied by the base (or boost) core clock speed.
Battlemage
Battlemage (Xe2) is the second-generation Xe architecture that debuted with its low power variant in Lunar Lake mobile processors that released in September 2024.[25] On December 3, 2024, Intel announced two Arc B-Series desktop graphics cards based on the Xe2-HPG graphics architecture.[26]
^Core boost values (if available) are stated below the base value in italics.
Future generations
Intel also revealed future generations of Intel Arc GPUs under development: Celestial (Xe3), and Druid (Xe4).[27][5]
Intel revealed that Meteor Lake and later generations of CPU SoCs uses an Intel Arc Tile GPU.[28][29]
Intel XeSS
Intel XeSS is a real-timedeep learning image upsampling technology developed primarily for use in video games as a competitor to Nvidia's DLSS and AMD's FSR technologies. Additionally, XeSS is not restricted to Arc graphics cards. It utilizes XMX instructions exclusive to Arc graphics cards, but will fall back to utilizing DP4a instructions on competing GPUs that have support for DP4a instructions. XeSS is trained with 64 samples per pixel as opposed to Nvidia DLSS's 16 samples per pixel (16K reference images).[30][31]
^ The algorithm does not necessarily need to be implemented using these presets; it is possible for the implementer to define custom input and output resolutions.
^ The linear scale factor used for upsampling the input resolution to the output resolution. For example, a scene rendered at 540p with a 2.00× scale factor would have an output resolution of 1080p.
^ The linear render scale, compared to the output resolution, that the technology uses to render scenes internally before upsampling. For example, a 1080p scene with a 50% render scale would have an internal resolution of 540p.
Issues
Drivers
Performance on Intel Arc GPUs has suffered from poor driver support, particularly at launch. An investigation by Gamers Nexus discovered 43 known driver issues with Arc GPUs, prompting a response and acknowledgement of the issues from Intel.[33][34][35] Intel CEO Pat Gelsinger also blamed driver problems as a reason for Arc's delayed launch.[36] A beta driver from October 2022 accidentally reduced the memory clock by 9% on the Arc A770 from 2187MHz to 2000MHz, resulting in a 17% reduction in memory bandwidth.[37] This particular issue was later fixed.[38] Intel provides an open source driver for Linux.[39]
A December 2022 driver update improved Arc compatibility and performance with DirectX 9-based games.[42] According to Intel, the driver update made Arc GPUs up to 1.8x faster in DirectX 9 games.[43] A February 2023 driver update further improved Arc's performance on DirectX 9-based games.[44]
Legacy BIOS compatibility
Intel Arc requires a UEFI BIOS with resizable BAR support for optimal performance.[45]
Footnotes
^In OpenCL 3.0, OpenCL 1.2 functionality has become a mandatory baseline, while all OpenCL 2.x and OpenCL 3.0 features were made optional.