NVIDIA GB10 vs Apple 80-Core M3 Ultra Full Specs

Summary

6,144 Shaders 2.52GHz	10,240 Shaders 1.38GHz
Shared Memory273.1GB/sLPDDR5X	Shared Memory819.2GB/sLPDDR5
Peak·FP32·FP16 31.03 TFLOPS	Peak·FP32·FP16 28.26 TFLOPS
Form Factor iGPU	Form Factor iGPU
TDP Shared	TDP Shared

Highlights

Peak OPS

Peak OPS FP32 FP16 Tensor FP4 FP4 Tensor Sparse Tensor FP8-16 FP8-16 Tensor Sparse Tensor FP8-32 FP8-32 Tensor Sparse Tensor FP16-16 FP16-16 Tensor Sparse Tensor FP16-32 FP16-32 Tensor Sparse FP64 BF16 Tensor BF16 BF16 Tensor Sparse Tensor TF32 Tensor INT8 Memory Bandwidth

GB10992.9 TFLOPSFP4 Tensor Sparse

x35.13

80-Core M3 Ultra28.26 TFLOPSFP16

x1

Show Top Score

Theoretical Performance

Clock Speed2525MHz UC·Base·Boost·OC	Clock Speed1380MHz UC·Boost·OC
Peak OPS 992.9 TFLOPSFP4 Tensor Sparse	Peak OPS 28.26 TFLOPSFP16
Tensor FP4 496.4 TFLOPS FP4 Tensor Sparse 992.9 TFLOPS	-
Tensor FP8-16 248.2 TFLOPS FP8-16 Tensor Sparse 496.4 TFLOPS Tensor FP8-32 124.1 TFLOPS FP8-32 Tensor Sparse 248.2 TFLOPS	-
Tensor FP16-16 124.1 TFLOPS FP16-16 Tensor Sparse 248.2 TFLOPS Tensor FP16-32 62.05 TFLOPS FP16-32 Tensor Sparse 124.1 TFLOPS	-
BF16 62.05 TFLOPS Tensor BF16 62.05 TFLOPS BF16 Tensor Sparse 124.1 TFLOPS	-
Tensor TF32 31.03 TFLOPS	Tensor TF32 -
FP32 31.03 TFLOPS	FP32 28.26 TFLOPS
FP64 15.51 TFLOPS	-
Tensor INT8 248.2 TOPS	-
Ray 94.1 TOPS	Ray -
Pixel Rate 80.8 GPixel/s	Pixel Rate 441.6 GPixel/s
Texture Rate 484.8 GTexel/s	Texture Rate 883.2 GTexel/s

Core Config

Shaders 6,144 Shaders	Shaders 10,240 Shaders
TMUs 192 TMUs	TMUs 640 TMUs
ROPs 32 ROPs	ROPs 320 ROPs
Tensor Cores 192 T-Cores	Tensor Cores -
RT Cores 48 RT-Cores	RT Cores 80 RT-Cores
SMs 48 SMs	EUs 1,280 EUs

Clocks

Base Clock 2.42GHz	Base Clock -
Boost Clock 2.52GHz	Boost Clock 1.38GHz

Cache

L2 Cache 24.6MB shared

L2 Cache 16.4MB shared

Memory

Shared MemoryLPDDR5X	Shared MemoryLPDDR5
Memory Bus 256-bit	Memory Bus 1024-bit
Memory Speed 8.5GT/s	Memory Speed 6.4GT/s
Memory Bandwidth 273.1GB/s	Memory Bandwidth 819.2GB/s

Video Output

Multi-Monitor 4	Multi-Monitor 8
HDCP HDCP 2.3	HDCP -

Encoder

Encoder Model NVENC 9

Encoder Model 4x Apple Media Engine 3

Decoder

Decoder Model NVDEC 6

Decoder Model 2x Apple Media Engine 3

Card

Form Factor iGPU	Form Factor iGPU
Cooling Open-Air	Cooling Passive

Chip

Manufacturer NVIDIA	Manufacturer Apple
Chip Designer NVIDIA	Chip Designer Apple
Architecture Blackwell	Architecture Apple Gen 6
Family -	Family M Series
Branding	Branding
Codename Thor	Codename G15D
Market Segment Embedded	Market Segment Workstation
Release Date Jan 6, 2025	Release Date Mar 5, 2025

Fabrication

Foundry TSMC	Foundry TSMC
Fabrication Node N3E	Fabrication Node N3B
Die Size -	Die Size 954mm²
Transistor Count -	Transistor Count 184 Billion
Transistor Density -	Transistor Density 193 MTr/mm²

Images

No images available

No images available

Used In

CPUs Grace Blackwell GB10	CPUs M3 Ultra (32C80G)
Desktops DGX Spark(Late 2025)	Desktops Mac Studio(M3 Ultra, Early 2025)

Competitors

NVIDIAGB10 vs GPU Apple60-Core M3 Ultra NVIDIAGB10 vs GPU Apple40-Core M4 Max NVIDIAGB10 vs GPU Apple40-Core M5 Max Show 2 more

Apple80-Core M3 Ultra vs GPU Apple60-Core M3 Ultra Apple80-Core M3 Ultra vs GPU NVIDIARTX PRO 4000 Blackwell SFF Apple80-Core M3 Ultra vs GPU NVIDIARTX PRO 4000 Blackwell Show 12 more