Granted Patent B2

US 12,321,638 · App. 18/154,434 · Granted Jun 3, 2025

Inventors: Xiaoli Rui (Shanghai, CN); Yuan-mao Chang (Shanghai, CN); Gang Hu (Shanghai, CN); Zining Wu (Shanghai, CN)

Assignee: INNOGRIT TECHNOLOGIES CO., LTD.

G06F3/0659G06F3/064G06F3/0679G06F3/0604

Abstract

Claims (22)

1. A method of simulating hard disk mode to access a compression accelerator card based on data intelligent processing technology, comprising:

inserting the compression accelerator card into a host;

loading a hard disk driver and an application, not a PCIe adapter driver, and successfully detecting the compression accelerator card as a virtual compression hard disk by the host, the host accessing the compression accelerator card as if accessing a normal hard disk;

using the application to perform functions of setting a compression identifier and sending a write command to the virtual compression hard disk, and the write command is sent to a compression accelerator card via the virtual compression hard disk, the compression accelerator card is driven by the hard disk driver and detected as the virtual compression hard disk to the host;

compressing data by the compression accelerator card according to the compression identifier and caching the compressed data within the compression accelerator card, and returning a message to the application;

using the application to perform functions of setting a decompression identifier and sending a read command to the virtual compression hard disk, the read command is sent to the compression accelerator card via the virtual compression hard disk, wherein a logical address to be read by the read command is a same logical address as was used for the write command; and

not decompressing the compressed data cached in the compression accelerator card according to the decompression identifier set by the application, and returning the compressed data cached in the compression accelerator card to the application by the virtual compression hard disk without decompressing the compressed data cached in the compression accelerator card,

wherein the compression identifier is added to the write command and sent to the virtual compression hard disk, and the decompression identifier is added to the read command and sent to the virtual compression hard disk, and

wherein the compression identifier and the decompression identifier are respectively arranged in a register of the compression accelerator card.

2. The method of claim 1 , wherein the hard disk driver is an NVMe driver.

3. The method of claim 1 , wherein logical addresses of all input/output interfaces of the compression accelerator card are based on 32K alignment.

4. A method of simulating hard disk mode to access a compression accelerator card based on data intelligent processing technology, comprising:

inserting the compression accelerator card into a host;

loading a hard disk driver and an application, not a PCIe adapter driver, and successfully detecting the compression accelerator card as a virtual compression hard disk by the host, the host accessing the compression accelerator card as if accessing a normal hard disk;

using the application to perform functions of setting a compression identifier and sending a write command to the virtual compression hard disk, and the write command is sent to a compression accelerator card via the virtual compression hard disk, the compression accelerator card is driven by the hard disk driver and detected as the virtual compression hard disk to the host;

not compressing data by the compression accelerator card according to the compression identifier set by the application and caching previously compressed data within the compression accelerator card, and returning a message to the application;

using the application to perform functions of setting a decompression identifier and sending a read command to the virtual compression hard disk, the read command is sent to the compression accelerator card via the virtual compression hard disk, wherein a logical address to be read by the read command is a same logical address as was used for the write command; and

decompressing the previously compressed data according to the decompression identifier by the compression accelerator card, and returning the decompressed data to the application,

wherein the compression identifier is added to the write command and sent to the virtual compression hard disk, and the decompression identifier is added to the read command and sent to the virtual compression hard disk, and

wherein the compression identifier and the decompression identifier are respectively arranged in a register of the compression accelerator card.

5. The method of claim 4 , wherein the hard disk driver is an NVMe driver.

6. The method of claim 4 , wherein logical addresses of all input/output interfaces of the compression accelerator card are based on 32K alignment.

Assignments (1)

Continuity (1)

References Cited (29)