At GTC 2016 in China NVIDIA released two new GPUs aimed at deep learning. The NVIDIA Tesla P4 and NVIDIA Tesla P40 cards are the newest generation of Pascal GPUs designed for the machine learning market that NVIDIA is aggressively targeting.
Like many GPUs that NVIDIA has been launching recently, the Tesla P4 and Tesla P40 are a pair of GPUs released much like the Tesla M4 and M40 released at GTC 2016 in San Jose earlier this year.
The NVIDIA Tesla P4 is a 50W TDP card that is meant for dense installations and low power inference workloads. Model training is more appropriate for the higher power part, the Tesla P40 which is a 250W TDP GPU. The NVIDIA Tesla P4 is an 8GB card while the P40 is a 24GB card.
NVIDIA Tesla P4 and Tesla P40 Specifications
Here are the specifications for the two new cards from NVIDIA:
| Specification | Tesla P4 | Tesla P40 |
| Single Precision FLOPS* | 5.5 | 12 |
| INT8 TOPS* (Tera-Operations Per Second) | 22 | 47 |
| CUDA Cores | 2,560 | 3,840 |
| GPU GDDR5 Memory | 8GB | 24GB |
| Memory Bandwidth | 192GB/s | 346GB/s |
| Power | 50 Watt (or higher) | 250 Watt |
* With boost clock on
Availability
The NVIDIA Tesla P4 and P40 are planned to be available in November and October, respectively, in qualified servers offered by ODM, OEM and channel partners. We have been expecting PCIe Pascal cards for some time now and it appears as though we will have them within a month or so.
More Information
You can see more about the two cards at NVIDIA’s official press release which you can find here.
