Intel Launches Arc B580 12 GB & B570 10 GB “Battlemage” GPUs For $249 & $219, Faster RT Than RTX 4060 & 10% Faster On Average
Intel has finally unveiled its next-generation Arc B580 and B570 "Battlemage" GPUs, which are targeted at mainstream gamers under $250 US.
Intel Battlemage Debuts With Sub-$250 US Pricing, B580 12 GB For $249 & B570 10 GB For $219 US
It has been two years since Intel released its first discrete graphics card lineup, the Alchemist A-Series. Today, the company is unveiling its second-generation discrete graphics line based on the Xe2 architecture, the Battlemage B-Series. The Battlemage lineup debuts with the BMG-G21 SoC, which is targeted at mainstream gamers and eyes the 1440p/1080p segment. We have already laid out the technical aspects of the Xe2 architecture which was first utilized by the Lunar Lake SoC & now, the architecture gets its first performance outing.
Related Story Intel’s Arc Battlemage B580 “Limited Edition” GPU Listed Online For $259
Starting with the details of the Xe2-based BMG-G21 SoC, the chip features a maximum of five render slices which include four Xe2 cores each for a total of 20 Xe2 cores. Each Xe2 core has 8 512-bit vector engines, 8 2048-bit XMX engines, 64b atomic ops support, and an upgraded 256KB L1$/SLM cache. The Xe2 cores also include a dedicated RT (Ray Tracing) unit, and each render slice carries four Sampler, Geometry, Rasterizer, HiZ, and two Pixel Backend blocks.
The BMG-G21 SoC measures 272mm2 and features a total of 19.6 million transistors, making it 33% smaller than the Alchemist ACM-G10 die which featured 21.7 million transistors. The chip itself has been fabricated on the TSMC 5nm (N5) process node.
According to Intel, the Battlemage BMG-G21 discrete GPU offers an incredible 70% performance improvement per Xe core and a 50% performance per watt uplift versus the prior generation. The updated Xe2 IP leads to lower execution time versus the Alchemist architecture, allowing for better utilization of the silicon and faster performance than its predecessors. So, with the architecture bits out of the way, let's start with the
Read more on wccftech.com
