Controller, memory system and operating method thereof
In accordance with an embodiment of the present disclosure, an operating method of a controller for controlling a nonvolatile memory device may include: generating pre-read information based on a first read request, reading out 1 st sub-chunks respectively included in a plurality of data chunks from the nonvolatile memory device, and providing a host with the read 1 st sub-chunks, wherein the first read request includes respective addresses of the 1 st sub-chunks; starting, after the 1 st sub-chunks are provided to the host, a pre-read operation of reading out 2 nd sub-chunks respectively included in the plurality of data chunks from the nonvolatile memory device based on the pre-read information and storing the read 2 nd sub-chunks into a memory in the controller; and providing, after the pre-read operation is started, the host with the 2 nd sub-chunks stored in the memory in response to a second read request received from the host.
1. An operating method of a controller for controlling a nonvolatile memory device, the operating method comprising:
generating pre-read information based on a first read request, reading out 1 st sub-chunks respectively included in a plurality of data chunks from the nonvolatile memory device, and providing a host with the read 1 st sub-chunks, wherein the first read request includes respective addresses of the 1 st sub-chunks;
starting, after the 1 st sub-chunks are provided to the host, a pre-read operation of reading out 2 nd sub-chunks respectively included in the plurality of data chunks from the nonvolatile memory device based on the pre-read information and storing the read 2 nd sub-chunks into a memory in the controller; and
providing, after the pre-read operation is started, the host with the 2 nd sub-chunks stored in the memory in response to a second read request received from the host,
wherein the nonvolatile memory device stores machine learning data that includes the plurality of data chunks, and each of the plurality of data chunks includes a plurality of sub-chunks including the 1 st and 2 nd sub-chunks.
2. The operating method of claim 1 , wherein the machine learning data is to be used for a machine learning operation based on Recurrent Neural Networks (RNN) or Long Short-Term Memory models (LSTM).
3. The operating method of claim 2 ,
wherein the first read request includes information indicating that the first read request is for data to be used in the machine learning operation, and
wherein the pre-read information is generated when the information indicating that the first read request is for data to be used in the machine learning operation is detected.
4. The operating method of claim 1 , wherein the pre-read information includes respective addresses of the second sub-chunk to the last sub-chunk among the plurality of sub-chunks within each of the plurality of data chunks.
5. The operating method of claim 1 , wherein the machine learning data is stored in the nonvolatile memory device through a sequential write operation.
6. The operating method of claim 1 , wherein a number of the plurality of data chunks included in the machine learning data is a batch size.
7. The operating method of claim 1 , wherein a size of each of the plurality of data chunks is a batch length.
8. The operating method of claim 1 , wherein each of the plurality of sub-chunks is a unit of learning processed in a machine learning operation based on Recurrent Neural Networks (RNN) or Long Short-Term Memory models (LSTM).
9. A memory system including:
a nonvolatile memory device configured to store machine learning data; and
a controller configured to control the nonvolatile memory device,
wherein the machine learning data includes a plurality of data chunks, and each of the plurality of data chunks includes a plurality of sub-chunks, and
wherein the controller is further configured to:
generate pre-read information based on a first read request, read out 1 st sub-chunks respectively included in the plurality of data chunks from the nonvolatile memory device, and provide a host with the read 1 st sub-chunks, wherein the first read request includes respective addresses of the 1 st sub-chunks;
start, after the 1 st sub-chunks are provided to the host, a pre-read operation of reading out 2 nd sub-chunks respectively included in the plurality of data chunks from the nonvolatile memory device based on the pre-read information and store the read 2 nd sub-chunks into a memory in the controller; and
provide, after the pre-read operation is started, the host with the 2 nd sub-chunks stored in the memory in response to a second read request received from the host.
10. The memory system of claim 9 , wherein the machine learning data is to be used for a machine learning operation based on Recurrent Neural Networks (RNN) or Long Short-Term Memory models (LSTM).
11. The memory system of claim 10 ,
wherein the first read request includes information indicating that the first read request is for data to be used in the machine learning operation, and
wherein the controller generates the pre-read information when the information indicating that the first read request is for data to be used in the machine learning operation is detected.
12. The memory system of claim 9 , wherein the pre-read information includes respective addresses of the second sub-chunk to the last sub-chunk among the plurality of sub-chunks within each of the plurality of data chunks.
13. The memory system of claim 9 , wherein the machine learning data is stored in the nonvolatile memory device through a sequential write operation.
14. The memory system of claim 9 , wherein a number of the plurality of data chunks included in the machine learning data is a batch size.
15. The memory system of claim 9 , wherein a size of each of the plurality of data chunks is a batch length.
16. The memory system of claim 9 , wherein each of the plurality of sub-chunks is a unit of learning processed in a machine learning operation based on Recurrent Neural Networks (RNN) or Long Short-Term Memory models (LSTM).