Method and device for supporting FPGA (Field Programmable Gate Array) training in TensorFlow

An operator and equipment technology, applied in the field of supporting FPGA training, can solve problems such as limiting FPGA usage scenarios, and achieve the effect of expanding usage scenarios
CN110929883AInactive Publication Date: 2020-03-27SUZHOU LANGCHAO INTELLIGENT TECH CO LTD
0 Cites 7 Cited by

Patent Information

Authority / Receiving Office
CN · China
Current Assignee / Owner
SUZHOU LANGCHAO INTELLIGENT TECH CO LTD
Publication Date
2020-03-27
Estimated Expiration
Not applicable · inactive patent

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention provides a method for supporting FPGA (Field Programmable Gate Array) training in TensorFlow, which comprises the following steps of: adding registration and discovery of FPGA equipmentin the TensorFlow so as to eanble the name of the FPGA equipment to be in an equipment list of the TensorFlow; calling an operator registration interface of TensorFlow to register an operator supporting the FPGA equipment according to the name of the FPGA equipment, and enabling the name of the operator to be the same as the names of the operators of all the equipment supported by the TensorFlow;and compiling execution functions of the operator by utilizing opencl, including a host end execution function and an FPGA equipment end execution function, so as to execute data interaction between aCPU on the host and the FPGA. According to the invention, the usage scenario of the FPGA is expanded, and the method is suitable for scenarios requiring online training and model updating.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The present invention relates to the computer field, and more specifically, relates to a method and device for supporting FPGA training in TensorFlow. Background technique

[0002] TensorFlow is currently the most widely used deep learning framework in the field of deep learning. Many deep learning models are implemented based on TensorFlow. Many hardware manufacturers, including ASIC and FPGA manufacturers, regard TensorFlow as the primary support framework for deep learning.

[0003] However, most of the current manufacturers only support the reasoning of TensorFlow models (convert the model into an intermediate layer supported by FPGA to run on FPGA), and only CPU, GPU, and TPU support TensorFlow training. Some manufacturers have implemented FPGAs to support TensorFlow reasoning, but do not support FPGA training. Since FPGAs do not support TensorFlow training and cannot be accelerated by FPGAs for training, models that are deployed online and requ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More