Intel Arc A580 aims at Radeon RX 6600 and GeForce RTX 3050, while NVIDIA gears up to launch an even slower RTX 3050
Today, a new slide has surfaced unveiling Intel’s long-forgotten Arc A580 GPU, accompanied by two custom models derived from this particular SKU. As previously mentioned in our earlier report, the A580 was officially announced last year but has not been made available until now. The recently leaked slide, credited to @momomo_us, suggests that Intel did indeed make some slight adjustments to the specifications, although these changes are relatively minor.
In Intel’s discrete GPU lineup, the A580 with its 24 Xe-Cores is set to become the first and only A5 model. This addition will firmly establish the Alchemist lineup with a fifth SKU, featuring a board power rating of 185W. Notably, the TBP (Total Board Power) is the aspect that has been altered from the originally disclosed 2022 specifications, which initially indicated a 175W rating.
These additional 10 watts of power are what Intel deemed necessary to compete with the NVIDIA GeForce RTX 3050 and AMD Radeon RX 6600 models, as per the information provided in the official slide, making them direct comparable in terms of product offerings:
Interestingly, another slide which shows a roadmap for all Intel Arc products appears to confirm that only two brands are planing A580 GPU, Sparkle and ASRock, both models were leaked earlier today.
NVIDIA with RTX 3050 6GB
NVIDIA doesn’t appear to be fazed by this development at the Intel camp, as there are reports suggesting that they are actively working on an even less powerful RTX 3050 model. This card, which some argue shouldn’t bear the RTX 3050 name, is rumored to be equipped with the GA107-325 GPU variant, a departure from the GA106-150 used in the 2022 model. Most notably, it is expected to have a lower boost clock speed of 1470 MHz, representing roughly a 20% decrease in performance.
Additionally, it is speculated that this card will feature a narrower 96-bit memory bus, as opposed to the previous 128-bit configuration, resulting in a reduction in memory bandwidth. Furthermore, the memory capacity is expected to be reduced to 6GB from the previous 8GB. The one aspect that still remains unconfirmed is the specific configuration of the GPU core.