NVIDIA RTX PRO 3000 vs NVIDIA RTX PRO 4000 graphics comparison: performance score, cores and clock speeds, VRAM, TGP, cache, and RT and AI units. The better value in each row is highlighted in green.
Summary
NVIDIA RTX PRO 4000 offers more compute units (CUDA cores: 7680 vs 5888), more VRAM (16 GB vs 12 GB), a wider memory bus (256 bit vs 192 bit), higher memory bandwidth (896 GB/s vs 672 GB/s), and more L2 cache (64 MB vs 48 MB).
Performance score
0
Coming soon
0
Coming soon
Key specs
CUDA cores5888
CUDA cores7680
BoostN/A
BoostN/A
VRAM12 GB
VRAM16 GB
Memory bus192 bit
Memory bus256 bit
TGP60–140 W
TGP80–175 W
MemoryGDDR7
MemoryGDDR7
General
SeriesRTX PRO Blackwell Mobile
SeriesRTX PRO Blackwell Mobile
GPU architectureBlackwell
GPU architectureBlackwell
GPU chipGB205
GPU chipGB203
FoundryTSMC
FoundryTSMC
Process node5 nm
Process node5 nm
Die size263 mm²
Die size378 mm²
Transistors31.1 billion
Transistors45.6 billion
InterfacePCIe 5.0 x16
InterfacePCIe 5.0 x16
Release dateMarch 18, 2025
Release dateMarch 18, 2025
Cores and clocks
CUDA cores5888
CUDA cores7680
46× SM
60× SM
Boost clock (max)N/A
Boost clock (max)N/A
Memory
VRAM12 GB
VRAM16 GB
MemoryGDDR7
MemoryGDDR7
Memory bus192 bit
Memory bus256 bit
Memory speed28 Gb/s
Memory speed28 Gb/s
Bandwidth672 GB/s
Bandwidth896 GB/s
TGP
TGP range60 - 140 W
TGP range80 - 175 W
Cache
L2 cache48 MB
L2 cache64 MB
L1 cache128 KB
L1 cache128 KB
Units
TMUs184
TMUs240
ROPs80
ROPs96
RT cores46
RT cores60
Tensor cores184
Tensor cores240
SM blocks46
SM blocks60
AI
AI performance992 TOPS
AI performance1334 TOPS