The GeForce RTX 2050, which was recently announced by NVIDIA, is an odd GPU that no one asked for, but the green team desperately needs it because of the upcoming Intel ARC Alchemist DG2-128 graphics chip for mobile devices.
In the discrete GPU segment, it appears that NVIDIA is already feeling the heat from Intel’s upcoming competition. The GeForce RTX 2050 and MX500 series GPUs, which were just announced, appear to be Nvidia’s answer to AMD’s Rembrandt APUs and Intel’s ARC Alchemist DG2-128 GPUs. NVIDIA hasn’t released any performance figures for either chip, but recent leaks have revealed disappointing results when compared to existing APUs and entry-level products.
Intel’s ARC Alchemist DG2-128 GPU appears to be getting ready to take on the GeForce RTX 2050 4 GB graphics card. Intel’s entry-level mobility GPU based on the Xe-HPG architecture will have a base TDP of 35W and a maximum PL2 rating of 50W, according to the latest leak from Moore’s Law is Dead.
Intel’s decision to move its power limit numbering from CPUs to GPUs is intriguing. This puts the entry-level GPU in direct competition with the RTX 2050 mobile, which has a comparable TDP design rated at 30-45W.
Also observe how NVIDIA’s GeForce RTX 2050 has a 64-bit bus interface, which is quite similar to AMD’s Navi 24 and Intel’s ARC 128 entry-level GPUs’ 64-bit bus interfaces. Furthermore, all of these GPUs are expected to be released in the first quarter of 2022. (Spring). It appears like NVIDIA is keen to gain a decent market share in the entry-level class, but with Intel as the segment’s second and most powerful competitor, things may become difficult for both NVIDIA and AMD.
AMD will continue to keep some market share thanks to its APUs, but only until Intel releases faster CPUs with faster Xe GPUs onboard. NVIDIA’s only hope of avoiding ARC is to become more aggressive in their product positioning or lower the prices of their mobile chips.
Intel Xe-HPG 128 EU GPU Powered ARC Alchemist Gaming Graphics Lineup
Last but not least, the Intel Xe-HPG Alchemist 128 EU (8 Xe Cores) components. The full-fat SKU with 1024 cores is available in two configurations: a 96-bit and a 64-bit variant with 6 GB and 4 GB memory capacity, respectively.
The stripped-down version will include 96 EUs (768 cores) and 4 GB GDDR6 memory on a 64-bit bus interface. The processor will also have a clock speed of roughly 2.2-2.5 GHz and a power consumption of less than 75W, implying that entry-level graphics cards will be connector-free.
Following are all the ARC 128 EU variants that we can expect based on the DG2-SOC2:
- 128 EU (1024 ALUs) / 6 GB @ Up To 16 Gbps / 96-bit / ~75W (Desktops)
- 128 EU (1024 ALUs) / 4 GB @ Up To 16 Gbps / 64-bit / 35-30W (Laptops)
- 96 EU (768 ALUs) / 4 GB @ Up To 16 Gbps / 64-bit / ~35W (Laptops)
Performance should be comparable to the GeForce GTX 1650 and GTX 1650 SUPER, but with raytracing. One significant advantage Intel may have over AMD and NVIDIA with these cards is that they may be able to penetrate the sub-$250 US market, which has been largely ignored in the current generation of cards. So far, the GeForce RTX 3050 series has only been available on laptops, with the RTX 3060 servicing the mainstream Ampere sector for $329 US and the RX 6600 serving as its major GPU competition for roughly $329 US.
This GPU will be very similar to the DG1 GPU-based discrete SDV board, but Alchemist will have a better architecture design and will outperform the first-gen Xe GPU architecture in terms of performance. Based on the specifications, this lineup is aimed at the entry-level desktop discrete market.
Intel ARC Alchemist vs NVIDIA GA106 & AMD Navi 24 GPUs
|Graphics Card Name||Intel ARC A350?||NVIDIA GeForce RTX 2050||AMD Radeon RX 6500 XT||Intel ARC A380?||NVIDIA GeForce RTX 3050|
|GPU Name||ARC Alchemist DG2-128||NVIDIA GA107||AMD Navi 24||ARC Alchemist DG2-128||NVIDIA GA106|
|Process Node||TSMC 6nm||Samsung 8nm||TSMC 7nm||TSMC 6nm||Samsung 8nm|
|FP32 Cores||8 Xe Cores||16 SM Units||16 Compute Units||8 Xe Cores||24 SM Units|
|Memory Capacity||4 GB GDDR6||4 GB GDDR6||4 GB GDDR6||6 GB GDDR6||8 GB GDDR6|
|Launch||Q1 2022||Q1 2022||Q2 2022||Q1 2022||Q1 2022|