AMD Radeon RX 9070 vs NVIDIA GeForce RTX 5070

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

Better figure, where one is clearly better

Radeon RX 9070 compared with GeForce RTX 5070
AMDRadeon RX 9070NVIDIAGeForce RTX 5070
Overview
ReleasedMarch 2025March 2025
Performance
Chip makerAMDNVIDIA
ArchitectureRDNA 4Blackwell
Shader cores3,5846,144
Boost clock2,520 MHz (better)2,512 MHz
Architecture56 compute units, 128 ROPs; game clock 2070 MHz; 53.9 billion transistorsGB205 GPU; 48 SMs
Manufacturing process–TSMC 4nm (4N NVIDIA custom process); 31.1 billion transistors; 263 mm² die
Ray tracing56 third-generation Ray Accelerators48 fourth-generation RT Cores; 94 RT TFLOPS
AI cores112 AI Accelerators; up to 578 TOPS INT8 or 1156 TOPS INT4 with structured sparsity192 fifth-generation Tensor Cores; 988 AI TOPS
Upscaling and frame generationAMD FSR "Redstone" (including FSR 4 upscaling), AMD Fluid Motion Frames, Radeon Super ResolutionDLSS 4.5 with Super Resolution, DLAA, Ray Reconstruction, Frame Generation, Multi Frame Generation and Dynamic Multi Frame Generation
Base clock–2.33 GHz
Cache64 MB AMD Infinity Cache48 MB L2
Memory
Video memory16 GB GDDR6 (better)12 GB GDDR7
Memory bus256-bit (better)192-bit
Memory bandwidth640 GB/s672 GB/s (better)
Memory detailGDDR6 at up to 20 GbpsGDDR7 at 28 Gbps
Connections and power
Board power220 W (better)250 W
InterfacePCIe 5.0 x16PCIe 5.0
Display outputsDisplayPort 2.1a, HDMI 2.1b (number of ports depends on the card)3 × DisplayPort 2.1b (UHBR20), 1 × HDMI 2.1b; up to 4 displays; HDCP 2.3
Maximum resolution–4K at 480 Hz or 8K at 165 Hz with DSC
Power connector2 × 8-pin1 × 300 W or greater PCIe Gen 5 cable, or 2 × PCIe 8-pin cables with the adapter in the box
Recommended power supply650 W650 W
Slot width–2-slot (Founders Edition)
Size (reference card)–242 × 112 mm, 2-slot (Founders Edition)
Video encode and decodeH.264, HEVC and AV1 encode and decode1 × ninth-generation NVENC, 1 × sixth-generation NVDEC; AV1 encode and decode

More comparisons