NVIDIA A40 PCIe vs NVIDIA GeForce GTX 1650 GDDR6
We compared two Desktop platform GPUs: 48GB VRAM A40 PCIe and 4GB VRAM GeForce GTX 1650 GDDR6 to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
NVIDIA A40 PCIe 's Advantages
Released 6 months late
Boost Clock has increased by 9% (1740MHz vs 1590MHz)
More VRAM (48GB vs 4GB)
Larger VRAM bandwidth (695.8GB/s vs 192.0GB/s)
9856 additional rendering cores
NVIDIA GeForce GTX 1650 GDDR6 's Advantages
Lower TDP (75W vs 300W)
Score
Benchmark
FP32 (float)
A40 PCIe
+1213%
37.42 TFLOPS
GeForce GTX 1650 GDDR6
2.849 TFLOPS
Blender
A40 PCIe
+657%
3990
GeForce GTX 1650 GDDR6
527
Graphics Card
Oct 2020
Release Date
Apr 2020
Tesla
Generation
GeForce 16
Desktop
Type
Desktop
PCIe 4.0 x16
Bus Interface
PCIe 3.0 x16
Clock Speeds
1305 MHz
Base Clock
1410 MHz
1740 MHz
Boost Clock
1590 MHz
1812 MHz
Memory Clock
1500 MHz
Memory
48GB
Memory Size
4GB
GDDR6
Memory Type
GDDR6
384bit
Memory Bus
128bit
695.8GB/s
Bandwidth
192.0GB/s
Render Config
-
-
-
84
SM Count
14
10752
Shading Units
896
336
TMUs
56
112
ROPs
32
336
Tensor Cores
-
84
RT Cores
-
128 KB (per SM)
L1 Cache
64 KB (per SM)
6 MB
L2 Cache
1024 KB
-
-
-
Theoretical Performance
194.9 GPixel/s
Pixel Rate
50.88 GPixel/s
584.6 GTexel/s
Texture Rate
89.04 GTexel/s
37.42 TFLOPS
FP16 (half)
5.699 TFLOPS
37.42 TFLOPS
FP32 (float)
2.849 TFLOPS
584.6 GFLOPS
FP64 (double)
89.04 GFLOPS
Board Design
300W
TDP
75W
700 W
Suggested PSU
250 W
3x DisplayPort 1.4a
Outputs
1x DVI
1x HDMI 2.0
1x DisplayPort 1.4a
8-pin EPS
Power Connectors
None
Graphics Processor
GA102
GPU Name
TU117
-
GPU Variant
TU117-300-A1
Ampere
Architecture
Turing
Samsung
Foundry
TSMC
8 nm
Process Size
12 nm
28.3 billion
Transistors
4.7 billion
628 mm²
Die Size
200 mm²
Graphics Features
12 Ultimate (12_2)
DirectX
12 (12_1)
4.6
OpenGL
4.6
3.0
OpenCL
3.0
1.3
Vulkan
1.3
8.6
CUDA
7.5
6.6
Shader Model
6.6
