The report found that Intel’s official Arc manages presentation video, specs of a separate desktop GPU that can be displayed for a split second. In addition, the Arc organizes Intel’s new GPU supervising software for Arc Alchemist and Xe Iris distinct GPUs. The presentation illustrates various features of the new software; its new Performance Tab for GPU is supervising, and some basic overclocking.
Data such as GPU power or core and memory regularity will be publicized through the Arc control software or a built-in screen placed on top that can be facilitated in games. Intel is showing this tab continuously in the video, but only for a second. We can see the specs that do not match any mobile Arc GPU declared two days ago.
According to those two screenshots, the GPU clock would reach 2200-2250 MHz with over 99% GPU implementation. It can be assumed that this is an enhanced clock for Intel’s new Xe-HPG GPU architecture. For desktop GPUs, a 2.2 GHz clock should not be particular.
The Arc A350M Geek Bench statistics depict that the GPU can get parallel regularity through the Open CL test. On the record, the core clock for this GPU is 1150 MHz; as a result, the base clock boosts up to 2.2 GHz.
A question arises then how we will know the Arc Control video shows desktop GPU?
First, provide the GPU Power of 175W. Not any mobile part offered by Intel has such a high power draw. But Arc A770M has the highest GPU power of 150W. But due to Intel Deep Link Technology, it can reach optimum power.
Since this is GPU power only, other desktop card components will still integrate more potency, so the whole card might facilely consume more than 200W, 3DCenter soothsays. We do not ken if Intel’s GPU Power accounts for the recollection subsystem yet. Nevertheless, GPU power leaked a while ago for the SOC1 GPU (ACM-G10) for desktop components. Do note that there is a distinction between TGP, TDP, or GPU Power metrics, but the details on Intel’s implementation are yet to be corroborated once desktop Arc GPUs are relinquished.
Furthermore, the VRAM clock of 1093 MHz suggests we are optically canvassing an efficacious peed haste of 17.5 Gbps, while the most expeditious mobile GPU has 16 Gbps. Recollection overclocking for mobile GPUs is a prodigiously unlikely feature; thus, this might attest that desktop Arc will not be more gradual than, for instance, GeForce RTX 3070 (14 Gbps).
| AMD Radeon RX 6700XT | NVIDIA GeForce RTX 3070 | Intel ARC A780 (?) | |
|---|---|---|---|
| Picture |  |  |  |
| GPU | AMD Navi 22 | NVIDIA GA104-300 | ACM-G10 |
| Fabrication Node | TSMC N7 | SAMSUNG 8N | TSMC N6 |
| Die Size | 336 mm² | 392 mm² | 406 mm² |
| Transistor Count | 17.2 B | 17.4 B | 21.7 B |
| FP32 Units | 2560 | 5888 | 4096 |
| Memory | 12GB GDDR6 129-bit | 8GB GDDR6 256-bit | 16GB GDDR6 256-bit |
| Memory Speed | 16 Gbps | 14 Gbps | 17.5 Gbps |
| Card Power | 230W (TGP) | 220 (TDP) | 175W (GPU Power) |
New Source: videocardz.com
