NVIDIA Tesla M40 Professional Graphics Card

Product status: Official | Last Update: 2015-11-10 | Report Error
Overview
Manufacturer
NVIDIA
Original Series
Tesla Maxwell
Release Date
November 10th, 2015
Model
NVIDIA PG600 SKU 202
Graphics Processing Unit
GPU Model
GM200-895 (GM200)
Architecture
Maxwell
Fabrication Process
28 nm
Die Size
601 mm2
Transistors Count
8B
Transistors Density
13.3M TRAN/mm2
CUDA Cores
3072
SMMs
24
TMUs
192
ROPs
96
Clocks
Base Clock
948 MHz
Boost Clock
1114 MHz
Memory Clock
1500 MHz
Effective Memory Clock
6000 Mbps
Memory Configuration
Memory Size
12288 MB
Memory Type
GDDR5
Memory Bus Width
384-bit
Memory Bandwidth
288.0 GB/s

Physical
Interface
PCI-Express 3.0 x16
Height
2-slot
Power Connectors
1× 8-pin
TDP/TBP
250 W
Recommended PSU
600 W
API Support
DirectX
12.0
Vulkan
1.0
OpenGL
4.5
OpenCL
3.0

Performance
Pixel Fillrate
106.9 GPixels/s
Texture Fillrate
213.9 GTexel/s
Peak FP32
6.8 TFLOPS
FP32 Perf. per Watt
27.4 GFLOPS/W
FP32 Perf. per mm2
11.4 GFLOPS/mm2




 ModelCoresBoost ClockMemory ClockMemory Config.
Thumbnail
NVIDIA Tesla M10
 
2560
-
 
6 Gbps
 
128 GB GD5 128b
Thumbnail
NVIDIA Tesla M60
 
4096
 
1184 MHz
 
6 Gbps
 
32 GB GD5 256b
Thumbnail
NVIDIA Tesla M40
 
3072
 
1114 MHz
 
6 Gbps
 
12 GB GD5 384b
Thumbnail
NVIDIA Tesla M6
 
1536
 
1051 MHz
 
4.6 Gbps
 
8 GB GD5 256b
Thumbnail
NVIDIA Tesla M4
 
1024
-
 
5.5 Gbps
 
4 GB GD5 128b
 ModelCoresBoost ClockMemory ClockMemory Config.
Thumbnail
NVIDIA Quadro VCA
 
24576
 
1140 MHz
 
6.6 GB/s
 
768 GB GD5 384b
Thumbnail
NVIDIA GeForce GTX TITAN X
 
3072
 
1075 MHz
 
7 GB/s
 
12 GB GD5 384b
Thumbnail
NVIDIA Quadro M6000
 
3072
 
1140 MHz
 
6.6 GB/s
 
12 GB GD5 384b
Thumbnail
NVIDIA Tesla M40
 
3072
 
1114 MHz
 
6 GB/s
 
12 GB GD5 384b
Thumbnail
NVIDIA GeForce GTX 980 Ti
 
2816
 
1076 MHz
 
7 GB/s
 
6 GB GD5 384b

NVIDIA Tesla M40 GPU Accelerator
The NVIDIA Tesla M40 GPU accelerator allows data scientists to save days, even weeks, of time while training their deep neural networks against massive amounts of data for higher overall accuracy. Key features include:

  • Optimized for Machine Learning – Reduces training time by 8X compared with CPUs (1.2 days vs. 10 days for a typical AlexNet training).
  • Built for 24/7 reliability – Designed and tested for high reliability in data center environments.
  • Scale-out performance – Support for NVIDIA GPUDirect allowing fast multi-node neural network training.