NVIDIA L40 vs NVIDIA Tesla S870
We compared two Professional market GPUs: 48GB VRAM L40 and 1536MB VRAM Tesla S870 to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
NVIDIA L40 's Advantages
Released 15 years and 5 months late
Boost Clock2490MHz
More VRAM (48GB vs 1536GB)
Larger VRAM bandwidth (864.0GB/s vs 76.80GB/s)
18048 additional rendering cores
Lower TDP (300W vs 800W)
Score
Benchmark
FP32 (float)
L40
+26061%
90.52 TFLOPS
Tesla S870
0.346 TFLOPS
Graphics Card
Oct 2022
Release Date
May 2007
Tesla Ada
Generation
Tesla
Professional
Type
Professional
PCIe 4.0 x16
Bus Interface
PCIe 1.0 x16
Clock Speeds
735 MHz
Base Clock
-
2490 MHz
Boost Clock
-
2250 MHz
Memory Clock
800 MHz
Memory
48GB
Memory Size
1536MB
GDDR6
Memory Type
GDDR3
384bit
Memory Bus
384bit
864.0GB/s
Bandwidth
76.80GB/s
Render Config
-
-
-
142
SM Count
16
18176
Shading Units
128
568
TMUs
32
192
ROPs
24
568
Tensor Cores
-
142
RT Cores
-
128 KB (per SM)
L1 Cache
-
96 MB
L2 Cache
96 KB
-
-
-
Theoretical Performance
478.1 GPixel/s
Pixel Rate
14.40 GPixel/s
1414 GTexel/s
Texture Rate
38.40 GTexel/s
90.52 TFLOPS
FP16 (half)
-
90.52 TFLOPS
FP32 (float)
345.6 GFLOPS
1414 GFLOPS
FP64 (double)
-
Board Design
300W
TDP
800W
700 W
Suggested PSU
1200 W
4x DisplayPort 1.4a
Outputs
No outputs
1x 16-pin
Power Connectors
-
Graphics Processor
AD102
GPU Name
G80
AD102-895-A1
GPU Variant
-
Ada Lovelace
Architecture
Tesla
TSMC
Foundry
TSMC
4 nm
Process Size
90 nm
76.3 billion
Transistors
0.681 billion
608 mm²
Die Size
484 mm²
Graphics Features
12 Ultimate (12_2)
DirectX
11.1 (10_0)
4.6
OpenGL
3.3
3.0
OpenCL
1.1 (1.0)
1.3
Vulkan
N/A
8.9
CUDA
1.0
6.6
Shader Model
4.0
