Intel's ray tracing performance is giving them a strong starting point
Published: 2nd September 2022 | Source: Intel | Author: Mark Campbell
Intel details ARC Alchemist's ray tracing performance, a strong point of Intel's discrete graphics technology
Intel's ARC Alchemist series graphics architecture is launching later this year, bringing a third player into the gaming GPU market. While Intel is not targeting the high-end linkups of AMD or Nvidia with their first generation ARC series products, they are aiming to deliver great value for money to their customers, often comparing their graphics cards to Nvidia's RTX 3060.
With their latest GPU comparison, Intel has compared their ARC A770 with Nvidia's RTX 3060, showcasing higher framerates in games that have ray tracing enabled. On average, Intel has showcased a 14% performance advantage over Nvidia's RTX 3060, placing Intel in a great position moving forward.
Ray tracing performance is becoming an increasingly important aspect of GPU performance, and Intel's strong ray tracing performance with their ARC Alchemist graphics cards will place Intel in a strong position moving forward. With their current levels of RT performance, Intel has already surpassed the ray tracing capabilities of AMD's RDNA 2 architecture, placing them ahead of one of their rivals. Assuming that Intel can scale up their graphics architectures and further grow their capabilities with next-generation products, Intel has the potential to grow to become a strong player within the GPU market.
Intel's Ray Tracing Hardware
Intel's ARC Alchemist series graphics cards support Microsoft's DirectX 12 Ultimate feature set and supports ray tracing using the DXR 1.0, DXR 1.1, and Vulkan APIs. Intel's Xe cores contain a Ray Tracing Unit (RTU) and a Thread Sorting Unit (TSU), both of which are important aspects of Intel's ray tracing "secret sauce".
Intel's Thread Sorting Units schedules work around Intel's Xe core and Ray Tracing Units, allowing Intel to tackle ray tracing workloads efficiently. Each of Intel's RTUs contain two ray traversal pipelines, which are used to calculate ray intersections and BVH structures. Intel's A-series COPUs can calculate 12 box intersections per cycle, 1 triangle intersection per cycle and features a dedicated BVH cache.
Using a technique called "Asynchronous Ray Tracing", Intel has utilised their TSU to organise ray tracing instructions with similar hit shaders and optimally allocates them unified shader resources. This allows Intel to best utilise their hardware resources and accelerates ray tracing workloads.
Intel's Ray Tracing Performance VS Nvidia's RTX 3060
At 1080p, Intel has claimed that their A770 graphics card can offer gamers a 14% performance advantage on average over Intel's RTX 3060. In Ghostwire Tokyo, new driver optimisations are expected to give Intel's A770 a 25% performance advantage.
Since AMD's ray tracing performance is often far behind Nvidia's, the results below are a strong showing for Intel. These results show that Intel has the potential to challenge Nvidia in a workload where they are currently a dominant player. That said, it remains to be seen how ray tracing performance will evolve in the future.
Ray Tracing + XeSS
By adding XeSS into the mix, Intel's ARC A770 graphics card is capable of delivering strong performance levels in games with ray tracing enabled at 1440p. That said, we have not seen XeSS in action yet, so it remains to be seen whether or not XeSS can deliver upcaling results as strong as Nvidia's DLSS technology.
You can join the discussion on Intel's ARC A770's ray tracing performance on the OC3D Forums .