System and method to improve data compression ratios for fixed block sizes in a smart data accelerator interface device
An information handling system for compressing data includes multiple compression engines, a source data buffer to provide compression data to the compression engines, at least one destination data buffer to receive compressed data from the compression engines, and a compression engine driver. Each compression engine is configured to provide a different compression function. The compression engine driver directs each compression engine to compress data from the source data buffer, and retrieves select compressed data from a first one of the compression engines from the at least one destination data buffer. The selection is based upon a selection criterion.
1. An information handling system for compressing data, the information handling system comprising:
a processor;
a memory device including a source data buffer and at least one destination data buffer; and
a smart data accelerator interface (SDXI) hardware accelerator including a plurality of compression engines, each compression engine configured to provide a different compression function;
wherein:
the source data buffer provides compression data to the compression engines;
the at least one destination data buffer receives compressed data from the compression engines; and
the processor instantiates a compression engine driver configured to instruct the SDXI hardware accelerator 1) to process a first command to determine that first data from the source data buffer is a first type of data, to direct each compression engine to compress the first data, and to retrieve select first compressed data from a first one of the compression engines from the at least one destination data buffer, the selection being based upon a first selection criterion, and 2) to process a second command to determine that second data from the source data buffer is the first type of data, to direct the first compression engine to compress the second data, and to retrieve second compressed data from the first compression engine;
wherein, the compression engine driver is further configured to process a third command to determine that third data from the source data buffer is a second type of data, to direct each compression engine to compress the third data, and to retrieve select third compressed data from a second one of the compression engines from the at least one destination data buffer, the selection being based upon a second selection criterion.
2. The information handling system of claim 1 , wherein the different compression functions include different compression algorithms.
3. The information handling system of claim 1 , wherein the different compression functions include a common compression algorithm, and further include different algorithmic settings.
4. The information handling system of claim 3 , wherein the different algorithmic settings include at least one of a different compression level, a different minimum match length, a different maximum match length, a different hash depth, a different match level, and a different static library.
5. The information handling system of claim 1 , wherein the first selection criterion includes selecting the selected compression engine because the compressed data from the selected compression engine has a highest compression ratio.
6. The information handling system of claim 1 , wherein the selection first criterion includes selecting the selected compression engine because the compression of the data from the selected compression engine has a compression ratio that is greater than a compression ratio threshold.
7. The information handling system of claim 6 , wherein the selection first criterion further includes selecting the selected compression engine because the data from the selected compression engine was a first compressed data with the compression ratio that is greater than the compression ratio threshold.
8. The information handling system of claim 1 wherein the selection first criterion includes selecting the selected compression engine because the compression of the data from the selected compression engine has a fastest compression speed.
9. The information handling system of claim 1 , wherein the at least one destination buffer includes a plurality of destination buffers, each destination buffer to receive compressed data from an associated one of the compression engines, and wherein the compression engine driver is further configured to select the first compression engine.
10. A method for compressing data, the method comprising:
providing, on an information handling system, a memory including a source data buffer and at least one destination data buffer;
providing, on the information handling system, a smart data accelerator interface (SDXI) hardware accelerator including a plurality of compression engines, each compression engine configured to provide a different compression function;
instantiating, on a processor of the information handling system, a compression engine driver;
instructing, by the compression engine driver, the SDXI hardware accelerator to process a first command;
processing, by the SDXI hardware accelerator, the first command, wherein in processing the first command, the method further comprises:
determining that first data from a source data buffer of the information handling system is a first type of data;
directing each compression engine to compress the first data; and
retrieving, by the compression engine driver, select compressed data from a first one of the compression engines stored in a destination data buffer of the information handling system, the selection being based upon a first selection criterion; and
processing, by the SDXI hardware accelerator, the second command, wherein in processing the second command, the method further comprises:
determining that second data from the source data buffer is the first type of data;
directing the first compression engine to compress the first data; and
retrieving, by the compression engine driver, select compressed data from the first compression engines stored in the destination data buffer; and
processing, by the SDXI hardware accelerator, a third command, wherein in processing the third command, the method further comprises:
determining that third data from the source data buffer is a second type of data;
directing each compression engine to compress the third data; and
retrieving, by the compression engine driver, select third compressed data from a second one of the compression engines from the at least one destination data buffer, the selection being based upon a second selection criterion.
11. The method of claim 10 , wherein the different compression functions include different compression algorithms.
12. The method of claim 10 , wherein the different compression functions include a common compression algorithm, and further include different algorithmic settings.
13. The method of claim 12 , wherein the different algorithmic settings include at least one of a different compression level, a different minimum match length, a different maximum match length, a different hash depth, a different match level, and a different static library.
14. The method of claim 10 , wherein the different algorithmic settings include at least one of a different compression level, a different minimum match length, a different maximum match length, a different hash depth, a different match level, and a different static library.
15. The method of claim 10 , wherein the first selection criterion includes selecting the selected compression engine because the compression of the data from the selected compression engine has a compression ratio that is greater than a compression ratio threshold.
16. The method of claim 15 , wherein the first selection criterion further includes selecting the selected compression engine because the data from the selected compression engine was a first compressed data with the compression ratio that is greater than the compression ratio threshold.
17. The method of claim 10 wherein the first selection criterion includes selecting the selected compression engine because the compression of the data from the selected compression engine has a fastest compression speed.
18. The method of claim 10 , wherein the at least one destination buffer includes a plurality of destination buffers, each destination buffer to receive compressed data from an associated one of the compression engines, and wherein the compression engine driver is further configured to select the first compression engine.
19. A smart data accelerator interface hardware device for compressing data, the hardware device comprising:
a plurality of compression engines each configured to provide a different compression function;
a source data buffer to provide compression data to the compression engines; and
at least one destination data buffer to receive compressed data from the compression engines;
wherein the hardware device executes code to direct each compression engine 1) to process a first command to determine that first compression data is a first type of data, to compress the first compression data, and to store select first compressed data from a first one of the compression engines to the one destination data buffer, the selection being based upon a first selection criterion, 2) To process a second command to determine that second compression data is the first type of data, to compress the second compression data, and to store compressed data from the first compression engine to the destination data buffer, and 3) to process a third command to determine that third compression data is a second type of data, to compress the second compression data, and to store select second compressed data from a second one of the compression engines to destination data buffer, the selection being based upon a second selection criterion.