NVIDIA GeForce RTX 3050 vs NVIDIA GeForce RTX 5090

Every spec side by side, from each maker's published figures.

Only show differences

Better figure, where one is clearly better

Share comparison

GeForce RTX 3050 compared with GeForce RTX 5090

	NVIDIAGeForce RTX 3050	NVIDIAGeForce RTX 5090
Overview
Released
Released	January 2022	January 2025
Performance
Chip maker
Chip maker	NVIDIA	NVIDIA
Architecture
Architecture	Ampere	Blackwell
Shader cores
Shader cores	2,560 or 2,304	21,760
Boost clock
Boost clock	1,780 MHz or 1,470 MHz	2,407 MHz
Architecture
Architecture	NVIDIA reference GPU figures; photo shows PNY's stock-clock 8 GB VERTO Dual Fan partner card. The 6 GB revision went on sale on 2 February 2024. NVIDIA also lists different OEM clocks and core counts; a complete independent OEM specification is not published here.	GB202 GPU; 170 SMs, 680 texture units, 176 ROPs
Manufacturing process
Manufacturing process	–	TSMC 4nm (4N NVIDIA custom process); 92.2 billion transistors; 750 mm² die
Ray tracing
Ray tracing	Second-generation RT Cores	170 fourth-generation RT Cores; 318 RT TFLOPS
AI cores
AI cores	Third-generation Tensor Cores	680 fifth-generation Tensor Cores; 3352 AI TOPS
Upscaling and frame generation
Upscaling and frame generation	NVIDIA DLSS upscaling	DLSS 4.5 with Super Resolution, DLAA, Ray Reconstruction, Frame Generation, Multi Frame Generation and Dynamic Multi Frame Generation
Base clock
Base clock	Original 8 GB is 1.55 GHz and 6 GB is 1.04 GHz on NVIDIA's rounded table; PNY specifies 1,552 MHz and 1,042 MHz respectively. NVIDIA rounds the original boost to 1.78 GHz; PNY gives 1,777 MHz.	2.01 GHz
Cache
Cache	–	96 MB L2
Memory
Video memory
Video memory	8 GB GDDR6 or 6 GB GDDR6	32 GB GDDR7
Memory bus
Memory bus	128-bit or 96-bit	512-bit
Memory bandwidth
Memory bandwidth	224 GB/s or –	1,792 GB/s
Memory detail
Memory detail	PNY specifies 14 Gbps GDDR6 for both capacities. Its original 8 GB page publishes 224 GB/sec bandwidth; the 6 GB PDF labels its bandwidth figure as Gbps, so an unambiguous numeric bandwidth is omitted for that revision.	GDDR7 at 28 Gbps
Connections and power
Board power
Board power	130 W or 70 W	575 W
Interface
Interface	PCIe 4.0	PCIe 5.0
Display outputs
Display outputs	PNY 8 GB VERTO has 3 × DisplayPort 1.4a and 1 × HDMI 2.1; PNY 6 GB VERTO lists DisplayPort 1.4a, HDMI 2.1 and DVI-DL, for 3 displays. Check the chosen partner card.	3 × DisplayPort 2.1b (UHBR20), 1 × HDMI 2.1b; up to 4 displays; HDCP 2.3
Maximum resolution
Maximum resolution	7680 × 4320 on PNY's specifications; HDMI supports 4K 120 Hz HDR, 8K 60 Hz HDR and variable refresh rate	4K at 480 Hz or 8K at 165 Hz with DSC
Power connector
Power connector	Original reference specification and PNY 8 GB VERTO require 1 × 8-pin; NVIDIA and PNY specify no auxiliary connector for the 6 GB version	1 × 600 W PCIe Gen 5 cable, or 4 × PCIe 8-pin cables with the adapter in the box
Recommended power supply
Recommended power supply	NVIDIA specifies 550 W for the original 8 GB card and 300 W for 6 GB; requirements depend on the rest of the system	1000 W
Slot width
Slot width	Partner designs vary; PNY's pictured 8 GB VERTO and its 6 GB VERTO are dual-slot cards	2-slot (Founders Edition)
Size (reference card)
Size (reference card)	No Founders Edition card; pictured PNY 8 GB VERTO is 8.99 × 4.49 × 1.71 inches. PNY 6 GB VERTO is 186.2 × 113 × 38.50 mm.	304 × 137 mm, 2-slot (Founders Edition)
Video encode and decode
Video encode and decode	PNY specifies seventh-generation NVIDIA Encoder and fifth-generation Decoder; its 8 GB board is PCIe 4.0 x8 and 6 GB is PCIe 4.0 x16 with x8 active	3 × ninth-generation NVENC, 2 × sixth-generation NVDEC; AV1 encode and decode