Apple 60-Core M3 Ultra vs NVIDIA GeForce RTX 5090 Full Specs

Summary

7,680 Shaders 1.38GHz	21,760 Shaders 2.41GHz
Shared Memory819.2GB/sLPDDR5	32GB GDDR71.79TB/s placeholder
Peak·FP32·FP16 21.2 TFLOPS	Peak·FP32·FP16 104.8 TFLOPS
Form Factor iGPU	Form Factor PCIe Card
TDP Shared	TDP 575W
Power Connectors -	Power Connectors 1x 16-Pin 12VHPWR

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 TDP

60-Core M3 Ultra21.2 TFLOPSFP16

x1

GeForce RTX 50903.35 PFLOPSFP4 Tensor Sparse

x158.14

Show Top Score

Theoretical Performance

Clock Speed1380MHz UC·Boost·OC	Clock Speed2407MHz UC·Base·Boost·OC
Peak OPS 21.2 TFLOPSFP16	Peak OPS 3.35 PFLOPSFP4 Tensor Sparse
-	Tensor FP4 1.68 PFLOPS FP4 Tensor Sparse 3.35 PFLOPS
-	Tensor FP8-16 838 TFLOPS FP8-16 Tensor Sparse 1.68 PFLOPS Tensor FP8-32 419 TFLOPS FP8-32 Tensor Sparse 838 TFLOPS
-	Tensor FP16-16 419 TFLOPS FP16-16 Tensor Sparse 838 TFLOPS Tensor FP16-32 209.5 TFLOPS FP16-32 Tensor Sparse 419 TFLOPS
-	BF16 104.8 TFLOPS Tensor BF16 209.5 TFLOPS BF16 Tensor Sparse 419 TFLOPS
Tensor TF32 -	Tensor TF32 104.8 TFLOPS
FP32 21.2 TFLOPS	FP32 104.8 TFLOPS
-	FP64 1.64 TFLOPS
-	Tensor INT8 838 TOPS
Ray -	Ray 317.7 TOPS
Pixel Rate 331.2 GPixel/s	Pixel Rate 423.6 GPixel/s
Texture Rate 662.4 GTexel/s	Texture Rate 1.64 TTexel/s

Core Config

Shaders 7,680 Shaders	Shaders 21,760 Shaders
TMUs 480 TMUs	TMUs 680 TMUs
ROPs 240 ROPs	ROPs 176 ROPs
Tensor Cores -	Tensor Cores 680 T-Cores
RT Cores 60 RT-Cores	RT Cores 170 RT-Cores
EUs 960 EUs	SMs 170 SMs

Clocks

Base Clock -	Base Clock 2.01GHz
Boost Clock 1.38GHz	Boost Clock 2.41GHz

Cache

L2 Cache 16.4MB shared

L2 Cache 98.3MB shared

Memory

Shared MemoryLPDDR5	32GB GDDR7 placeholder
Memory Bus 1024-bit	Memory Bus 512-bit
Memory Speed 6.4GT/s	Memory Speed 28GT/s
Memory Bandwidth 819.2GB/s	Memory Bandwidth 1.79TB/s

Power & Thermals

TDP Shared	TDP 575W
Max Temp -	Max Temp 90°C Max

Video Output

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

Ports

-	3x DisplayPort 2.1
-	1x HDMI 2.1

Encoder

Encoder Model 4x Apple Media Engine 3

Encoder Model 3x NVENC 9

Decoder

Decoder Model 2x Apple Media Engine 3

Decoder Model 2x NVDEC 6

Card

Form Factor iGPU	Form Factor PCIe Card
PCIe -	PCIe 2.1-Slots
-	Height 137 mm (5.39") Width 304 mm (11.97") Depth 42 mm (1.65")
Cooling Passive -	Cooling Open-Air 2x Fans
Power Connectors -	Power Connectors 1x 16-Pin 12VHPWR

Chip

Manufacturer Apple	Manufacturer NVIDIA
Chip Designer Apple	Chip Designer NVIDIA
Architecture Apple Gen 6	Architecture Blackwell
Family M Series	Family GeForce 50
Branding	Branding
Codename G15D	Codename NV192
Chip Variant -	Chip Variant GB202-250-A1
Market Segment Workstation	Market Segment Desktop
Release Date Mar 5, 2025	Release Date Jan 6, 2025

Fabrication

Foundry TSMC	Foundry TSMC
Fabrication Node N3B	Fabrication Node 4NP
Die Size 954mm²	Die Size 762mm²
Transistor Count 184 Billion	Transistor Count 92.2 Billion
Transistor Density 193 MTr/mm²	Transistor Density 121.1 MTr/mm²

Images

No images available

No images available

Used In

CPUs M3 Ultra (28C60G)	CPUs -
Desktops Mac Studio(M3 Ultra, Early 2025)	Desktops Tinybox Green v2(Mid 2025)

Competitors

Apple60-Core M3 Ultra vs GPU Apple80-Core M3 Ultra Apple60-Core M3 Ultra vs GPU NVIDIARTX PRO 4000 Blackwell SFF Apple60-Core M3 Ultra vs GPU Apple40-Core M5 Max Show 12 more

NVIDIAGeForce RTX 5090 vs GPU NVIDIARTX PRO 6000 Blackwell Server NVIDIAGeForce RTX 5090 vs GPU NVIDIARTX PRO 6000 Blackwell NVIDIAGeForce RTX 5090 vs GPU NVIDIARTX PRO 6000 Blackwell Max-Q Show 12 more