NVIDIA GeForce RTX 3090 Ti vs NVIDIA RTX PRO 4000 Blackwell Full Specs

Summary

10,752 Shaders 1.86GHz	8,960 Shaders 2.45GHz
24GB GDDR6X1.01TB/s	24GB GDDR7672GB/s
Peak·FP32·FP16 40 TFLOPS	Peak·FP32·FP16 43.94 TFLOPS
Form Factor PCIe Card	Form Factor PCIe Card
TDP 450W	TDP 140W
Power Connectors - 1x 12-Pin	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 Tensor INT4 Memory Bandwidth TDP

GeForce RTX 3090 Ti1.28 POPSINT4 Tensor Sparse

x1

RTX PRO 4000 Blackwell1.41 PFLOPSFP4 Tensor Sparse

x1.10

Show Top Score

Theoretical Performance

Clock Speed1860MHz UC·Base·Boost·OC	Clock Speed2452MHz UC·Boost·OC
Peak OPS 1.28 POPSINT4 Tensor Sparse	Peak OPS 1.41 PFLOPSFP4 Tensor Sparse
-	Tensor FP4 703 TFLOPS FP4 Tensor Sparse 1.41 PFLOPS
-	Tensor FP8-16 351.5 TFLOPS FP8-16 Tensor Sparse 703 TFLOPS Tensor FP8-32 175.8 TFLOPS FP8-32 Tensor Sparse 351.5 TFLOPS
Tensor FP16-16 160 TFLOPS FP16-16 Tensor Sparse 320 TFLOPS Tensor FP16-32 79.99 TFLOPS FP16-32 Tensor Sparse 160 TFLOPS	Tensor FP16-16 175.8 TFLOPS FP16-16 Tensor Sparse 351.5 TFLOPS Tensor FP16-32 87.88 TFLOPS FP16-32 Tensor Sparse 175.8 TFLOPS
BF16 40 TFLOPS Tensor BF16 79.99 TFLOPS BF16 Tensor Sparse 160 TFLOPS	BF16 43.94 TFLOPS Tensor BF16 87.88 TFLOPS BF16 Tensor Sparse 175.8 TFLOPS
Tensor TF32 40 TFLOPS	Tensor TF32 43.94 TFLOPS
FP32 40 TFLOPS	FP32 43.94 TFLOPS
FP64 625 GFLOPS	FP64 686.6 GFLOPS
Tensor INT4 640 TOPS	Tensor INT4 -
Tensor INT8 320 TOPS	Tensor INT8 351.5 TOPS
Ray 60.32 TOPS	Ray 133.3 TOPS
Pixel Rate 208.3 GPixel/s	Pixel Rate 235.4 GPixel/s
Texture Rate 625 GTexel/s	Texture Rate 686.6 GTexel/s

Core Config

Shaders 10,752 Shaders	Shaders 8,960 Shaders
TMUs 336 TMUs	TMUs 280 TMUs
ROPs 112 ROPs	ROPs 96 ROPs
Tensor Cores 336 T-Cores	Tensor Cores 280 T-Cores
RT Cores 84 RT-Cores	RT Cores 70 RT-Cores
SMs 84 SMs	SMs 70 SMs

Clocks

Base Clock 1.56GHz	Base Clock -
Boost Clock 1.86GHz	Boost Clock 2.45GHz

Cache

L2 Cache 6.1MB shared

L2 Cache 49.1MB shared

Memory

24GB GDDR6X	24GB GDDR7
Memory Bus 384-bit	Memory Bus 192-bit
Memory Speed 21GT/s	Memory Speed 28GT/s
Memory Bandwidth 1.01TB/s	Memory Bandwidth 672GB/s
ECC -	ECC Yes

Power & Thermals

TDP 450W	TDP 140W
Max Temp 92°C Max	Max Temp -

Video Output

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

Ports

-	4x DisplayPort 2.1
3x DisplayPort 1.4	-
1x HDMI 2.1	-

Encoder

Encoder Model NVENC 7

Encoder Model 2x NVENC 9

Decoder

Decoder Model NVDEC 5

Decoder Model 2x NVDEC 6

Card

Form Factor PCIe Card	Form Factor PCIe Card
PCIe 3-Slots	PCIe 1-Slots
Height 140 mm (5.51") Width 336 mm (13.23") Depth 61 mm (2.4")	Height 111.8 mm (4.4") Width 241.3 mm (9.5") Depth 20 mm (0.79")
Cooling Open-Air 2x Fans	Cooling Blower 1x Fan
Power Connectors 1x 12-Pin	Power Connectors 1x 16-Pin 12VHPWR

Chip

Manufacturer NVIDIA	Manufacturer NVIDIA
Chip Designer NVIDIA	Chip Designer NVIDIA
Architecture Ampere	Architecture Blackwell
Family GeForce 30	Family RTX PRO Blackwell
Branding	Branding
Codename NV172	Codename NV193
Chip Variant GA102-200-KD-A1	Chip Variant GB203-200-A1
Market Segment Desktop	Market Segment Workstation
Release Date Mar 29, 2022	Release Date Mar 18, 2025

Fabrication

Foundry Samsung	Foundry TSMC
Fabrication Node 8N	Fabrication Node 4NP
Die Size 628mm²	Die Size 378mm²
Transistor Count 28.3 Billion	Transistor Count 45.6 Billion
Transistor Density 45.04 MTr/mm²	Transistor Density 120.6 MTr/mm²

Images

No images available

No images available

Used In

Desktops XPS(XD8950, Late 2021)

Desktops -

Competitors

NVIDIAGeForce RTX 3090 Ti vs GPU NVIDIAGeForce RTX 4090 D NVIDIAGeForce RTX 3090 Ti vs GPU NVIDIAGeForce RTX 5070 Ti NVIDIAGeForce RTX 3090 Ti vs GPU NVIDIAGeForce RTX 5080 Show 12 more

NVIDIARTX PRO 4000 Blackwell vs GPU NVIDIARTX PRO 4500 Blackwell NVIDIARTX PRO 4000 Blackwell vs GPU NVIDIARTX PRO 4000 Blackwell SFF NVIDIARTX PRO 4000 Blackwell vs GPU NVIDIARTX 4500 Ada Show 12 more