Previous reports stated that Nvidia would most likely choose to release the GeForce RTX 5050 in July. In addition to the independent graphics card used in desktop computers, there is a high probability that a laptop version will also be released together. Unlike the GeForce RTX 50 series graphics cards used in other desktops and laptops, which use GDDR7, Nvidia will switch to GDDR6 this time and is rumored to have chosen Samsung and SK Hynix as video memory suppliers.
Recently, some netizens revealed that Nvidia plans to choose to carry 20Gbps GDDR6 on the desktop GeForce RTX 5050, which is the same as AMD’s graphics card based on RDNA 4 architecture GPU. In addition, the GeForce RTX 5050 will use a 3+2 phase power supply design, corresponding to the GPU and video memory, while the previous generation GeForce RTX 4060 has 1 phase in the video memory part.
The RTX 5050 will be equipped with the Blackwell architecture GB207. The PCB design is PG152-F01 SKU 50 and has 2560 CUDA cores, which is one-third less than the RTX 5060. The bit width of both is 128-bit, and the capacity is obviously 8GB. The video memory is downgraded from GDDR7 to GDDR6. As the rate will also decrease, the video memory bandwidth corresponding to the RTX 5050 will decrease, from 448GB/s of the RTX 5060 to 320GB/s, but it is higher than the 272GB/s of the RTX 4060 (the rate is 17Gbps). In addition, the power consumption of the entire RTX 5050 card is 130W, which is 15W lower than the RTX 5060.
It is rumored that this generation of Blackwell architecture has five chips: GB202, GB203, GB205, GB206 and GB207. With the arrival of RTX 5050, the last GB207 will also appear. Nvidia has not yet determined the specific release time of the GeForce RTX 5050, but it is believed that it will soon complete the update of this generation of gaming graphics card product lines.