GPU Comparison
GEFORCE
NVIDIA B200
CORE STATE
GB100
VRAM
90 GB
CLOCK SPEED
1965 MHz
TDP
1000 W
BUS WIDTH
4096 bit
ARCHITECTURE
Blackwell
PROCESS
5 nm
LAUNCH DATE
—
VS
GEFORCE
TITAN RTX
CORE STATE
TU102
VRAM
24 GB
CLOCK SPEED
1770 MHz
TDP
280 W
BUS WIDTH
384 bit
ARCHITECTURE
Turing
PROCESS
12 nm
LAUNCH DATE
2018
PERFORMANCE BENCHMARKS
geekbench_opencl
3dmark_3dmark_steel_nomad_dx12
geekbench_vulkan
passmark_directx_10
passmark_directx_11
passmark_directx_12
passmark_directx_9
passmark_g2d
passmark_g3d
passmark_gpu_compute
DETAILED SPECIFICATIONS
SPECIFICATION
B200
TITAN RTX
Core Specs
Shading Units
18,944
4,608
-75.7%
Shaders
18,944
4,608
-75.7%
TMUs
592
288
-51.4%
ROPs
24
96
+300.0%
SM Count
148
72
-51.4%
Clocks
Base Clock
700 MHz
1350 MHz
Boost Clock
1965 MHz
1770 MHz
Memory Clock
2000 MHz
8 Gbps effective
1750 MHz
14 Gbps effective
Memory
Memory Size
90 GB
24 GB
VRAM (MB)
92,160
24,576
-73.3%
Memory Type
HBM3e
GDDR6
Memory Bus
4096 bit
384 bit
Bandwidth
4.10 TB/s
672.0 GB/s
Cache
L1 Cache
256 KB (per SM)
64 KB (per SM)
L2 Cache
50 MB
6 MB
Performance
Pixel Rate
47.16 GPixel/s
169.9 GPixel/s
Texture Rate
1,163.3 GTexel/s
509.8 GTexel/s
FP32 (TFLOPS)
74.45 TFLOPS
16.31 TFLOPS
FP64 (TFLOPS)
37.22 TFLOPS (1:2)
509.8 GFLOPS (1:32)
FP16 (TFLOPS)
1,191.2 TFLOPS (16:1)
32.62 TFLOPS (2:1)
AI/RT
RT Cores
—
72
Tensor Cores
592
576
-2.7%
Power
TDP
1000 W
280 W
TDP (W)
1,000
280
-72.0%
Suggested PSU
1400 W
600 W
Power Connectors
—
2x 8-pin
Architecture
Architecture
Blackwell
Turing
GPU Name
GB100
TU102
Generation
Server Blackwell
(Bxx)
GeForce 20
Process Size
5 nm
12 nm
Transistors
104,000 million
18,600 million
Die Size
—
754 mm²
Foundry
TSMC
TSMC
Density
—
24.7M / mm²
API Support
DirectX
—
12 Ultimate (12_2)
OpenGL
—
4.6
Vulkan
—
1.4
OpenCL
3.0
3.0
CUDA
10.0
7.5
Shader Model
—
6.8
Physical
Slot Width
SXM Module
Dual-slot
Length
—
267 mm
10.5 inches
Height
—
116 mm
4.6 inches
Outputs
No outputs
1x HDMI 2.03x DisplayPort 1.4a1x USB Type-C
Bus Interface
PCIe 5.0 x16
PCIe 3.0 x16
Other
Launch Price
—
2,499 USD
Production
Active
End-of-life
Predecessor
Server Hopper
GeForce 10
Successor
Server Rubin
GeForce 30