CN103677757B - A kind of can the program parallelization degree control method of perception memory device bandwidth - Google Patents
A kind of can the program parallelization degree control method of perception memory device bandwidth Download PDFInfo
- Publication number
- CN103677757B CN103677757B CN201310477537.2A CN201310477537A CN103677757B CN 103677757 B CN103677757 B CN 103677757B CN 201310477537 A CN201310477537 A CN 201310477537A CN 103677757 B CN103677757 B CN 103677757B
- Authority
- CN
- China
- Prior art keywords
- degree
- parallelism
- efficiency monitoring
- function
- efficiency
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Landscapes
- Debugging And Monitoring (AREA)
Abstract
The invention discloses a kind of can the program parallelization degree control method of perception memory device bandwidth, the method is a kind of adaptive memory device combination property according to application program operation platform and real time load situation, the degree of parallelism of dynamic conditioning application program, i.e. process or number of threads, keeps the degree of parallelism that can obtain best I/O efficiency to make application program.The method, by the also records application program I/O efficiency information of monitoring in real time, controls as feedback information the degree of parallelism adjusting application program using it.The method, by progressively detecting the degree of parallelism of progressive increase program until reach the flex point of I/O actual efficiency, then carries out finely tuning to obtain the best degree of parallelism for different application platform; And then carrying out periodically according to the real time load situation of different application platform, the dynamic self-adapting that active accommodation combines with passive adjustment operates the best degree of parallelism having carried out application program and obtains.
Description
Technical field
The present invention relates to a kind of control being applicable to the concurrent program degree of parallelism of chip multi-core multiprocessor, particularly a kind of can the program parallelization degree control method of perception memory device I/O bandwidth.
Background technology
Chip multi-core multiprocessor (ChipMulit-Processor, CMP) is exactly be integrated in processor chips by multiple calculating kernel, thus improves computing power.Need the sharing and synchronization of carrying out data between the program of each core cpu execution of CMP processor, therefore its hardware configuration must support intercore communication.
After the appearance of CMP treatment technology, the parallelization of conventional serial program becomes the emphasis of research, most concurrent program is using the check figure of CMP as the important references index selecting degree of parallelism (process or number of threads), this processing policy achieves good effect in the application program of computation-intensive, if but for I/O intensity application only consider that the calculated performance of CMP may not reach maximum acceleration effect or throughput (I/O throughput is also referred to as I/O bandwidth), because application program may be limited to the I/O(I/O running this application platform) performance.
The I/O ability of different application platform differs greatly, and this depends on memory device (such as, the server of ordinary PC, band scsi interface, band disk array and the server of RAID card, magnetic disc i/o bandwidth has very big-difference).Based on the scale of memory device, create the hierarchical structure based on friction speed and capacity memory.As for a multi-level store hierarchical structure, please refer to " Computer Systems Organization--quantitative method " book of the translations such as in July, 2004 Zheng Weimin, Fig. 5 .1 in the design of this book the 5th chapter storage hierarchy is explained in detail.When different parallel tables runs for being applied in of frequent access I/O equipment, be difficult to directly determine the degree of parallelism that can obtain maximum I/O efficiency, the too low meeting of degree of parallelism causes I/O bandwidth to be not fully utilized wastes processor core resource simultaneously; Degree of parallelism is too high equally may not necessarily improving performance, because I/O bandwidth becomes bottleneck, excessively high degree of parallelism causes mutually disturbing aggravation between parallel execution unit, such as can make file prefetch mechanisms and the disk action confusion reigned of kernel, thus cause overall performance not rise counter falling.
For the Parallel application of I/O intensity, come according to the handling property of different application platform, determine that degree of parallelism cannot the advantage of maximum performance parallel processing.Therefore, how adaptively according to comprehensive hardware condition and the real time load situation of different application platform, can determine that corresponding best degree of parallelism becomes the key of dealing with problems dynamically.
Summary of the invention
The object of the invention is to propose a kind of adaptive memory device combination property according to application program operation platform and real time load situation, carry out the degree of parallelism of dynamic conditioning application program, i.e. process or number of threads, makes application program keep obtaining the degree of parallelism of best I/O efficiency.The method, by the also records application program I/O efficiency information of monitoring in real time, controls as feedback information the degree of parallelism adjusting application program using it.The method, by progressively detecting the degree of parallelism of progressive increase program until reach the flex point of I/O actual efficiency, then carries out finely tuning to obtain the best degree of parallelism for different application platform; And then carrying out periodically according to the real time load situation of different application platform, the dynamic self-adapting that active accommodation combines with passive adjustment operates the best degree of parallelism having carried out application program and obtains.
Of the present invention a kind of can the program parallelization degree control method of perception memory device bandwidth, the program that application platform runs can be divided into three states, i.e. initial state, stable state and final state according to program I/O efficiency;
Initial state refers to that application program opens a kind of state when running with lower degree of parallelism;
A kind of state when stable state refers to application program through entering stable operation after initial state;
Final state refers to that application program maintains a kind of state when higher I/O efficiency is run after stable state;
It is characterized in that: described program parallelization degree control method includes following treatment step:
Step one: the real-time monitoring of I/O efficiency
Step 101: the byte number of the read/write function of recording address intercepting-efficiency monitoring
At i-th I/O efficiency monitoring time T
iin, adopt the mode that address intercepts, the entrance function read of read operation operating system provided is labeled as and reads function c_read with I/O efficiency monitoring, and records the byte number that described c_read reads function and be designated as
the entrance function write of write operation operating system provided is labeled as and writes function c_write with I/O efficiency monitoring, and records the byte number that described c_write writes function and be designated as
then have, at T
ithe byte number set of the read/write function of the address intercepting-efficiency monitoring under condition is expressed as
In application program running time T
alwaysunder condition, the byte number of the read/write function of all addresses intercepting-efficiency monitoring is designated as
referred to as address intercepting-efficiency monitoring-byte number
Step 102: the byte number of the read/write function of record function overloading-efficiency monitoring
At the i-th+1 I/O efficiency monitoring time T
i+1in, adopt function overloading mode to provide reading function u_read and writing function u_write with monitoring function to user; The byte number that described u_read reads function is designated as
the byte number that described u_write writes function is designated as
then have, at T
i+1the byte number set of the read/write function of the function overloading-efficiency monitoring under condition is expressed as
At T
alwaysunder condition, the byte number of the read/write function of all function overloading-efficiency monitorings is designated as
referred to as function overloading-efficiency monitoring-byte number
Step 103: the byte number of the read/write function of record interface information-efficiency monitoring
At i-th I/O efficiency monitoring time T
iin, the I/O statistical information interface adopting operating system to provide, carries out the raw data of pre-service kernel generation, and completes the monitoring of I/O efficiency, and the system monitoring obtained is read function s_read and write function s_write; The byte number that described s_read reads function is designated as
the byte number that described s_write writes function is designated as
then have, at T
ithe byte number set of the read/write function of the interface message-efficiency monitoring under condition is expressed as
At T
alwaysunder condition, the byte number of the read/write function of total interface information-efficiency monitoring is designated as
referred to as interface message-efficiency monitoring-byte number
Step 2: the denoising of the byte number of read/write function
Application program includes different I/O efficiency monitoring type MODE when running, namely MODE includes address intercepting-efficiency monitoring type SYS, function overloading-efficiency monitoring type USR and interface message-efficiency monitoring type i NT, i.e. MODE={SYS, USR, INT};
According to denoising condition
filtering interfering process is carried out to the byte number of read/write function, obtains filtering postbyte number; If value is " 1 ", choose
and carry out degree of parallelism extraction process, then carry out quick position, if value is " 0 ", abandon
degree of parallelism extract, directly carry out quick position;
Q represents the number of I/O efficiency monitoring time;
I represents any one I/O efficiency monitoring time;
represent at I/O efficiency monitoring time T
iunder condition, the byte number of the read/write function of any I/O efficiency monitoring type MODE;
represent at I/O efficiency monitoring time T
i+1under condition, the byte number of the read/write function of any I/O efficiency monitoring type MODE;
C
threshold valuerepresent the critical value that the judgement wave phenomenon that denoising is arranged occurs;
According to denoising condition E to address intercepting-efficiency monitoring-byte number
denoising after, the function overloading-efficiency monitoring obtained-filtration postbyte number scale is
According to denoising condition E to function overloading-efficiency monitoring-byte number
denoising after, the function overloading-efficiency monitoring obtained-filtration postbyte number scale is
According to denoising condition E docking port information-efficiency monitoring-byte number
denoising after, the function overloading-efficiency monitoring obtained-filtration postbyte number scale is
Step 3: the quick position of degree of parallelism
The value of described degree of parallelism P is the power of 2, then have P=1,2,4,8 ..., 2
p, p represents exponential; Application program brings into operation with degree of parallelism P=1 after opening, and the filtration postbyte number scale obtained under the condition of P=1 is
after value P=1, degree of parallelism is according to P=2
prun, and obtain filtering postbyte number scale and be
To large, histogram structure is carried out to the filtration postbyte number different degree of parallelism from little according to degree of parallelism, obtain degree of parallelism-I/O efficiency histogram; By the byte number in more described degree of parallelism-I/O efficiency histogram, select maximum number of byte
this maximum number of byte
it is then the flex point of I/O efficiency monitoring;
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=1-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=2-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=4-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=8-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=16-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=32-filtration postbyte number scale be
By degree of parallelism P=2
pcondition under address intercepting-efficiency monitoring-filtration postbyte number scale of obtaining be
Be loaded into the filtration postbyte number under different degree of parallelism with degree of parallelism P for horizontal ordinate, byte number is in the histogram of ordinate, obtains address intercepting-efficiency monitoring histogram; By the byte number in the intercepting-efficiency monitoring histogram of more described address, select maximum number of byte
this maximum number of byte
it is then the flex point of the I/O efficiency monitoring of address intercepting;
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=1-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=2-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=4-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=8-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=16-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=32-filtration postbyte number scale be
By degree of parallelism P=2
pcondition under function overloading-efficiency monitoring-filtration postbyte number scale of obtaining be
Be loaded into the filtration postbyte number under different degree of parallelism with degree of parallelism P for horizontal ordinate, byte number is in the histogram of ordinate, obtains function overloading-efficiency monitoring histogram; By the byte number in more described function overloading-efficiency monitoring histogram, select maximum number of byte
this maximum number of byte
it is then the flex point of the I/O efficiency monitoring of function overloading;
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=1-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=2-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=4-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=8-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=16-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=32-filtration postbyte number scale be
By degree of parallelism P=2
pcondition under interface message-efficiency monitoring-filtration postbyte number scale of obtaining be
Be loaded into the filtration postbyte number under different degree of parallelism with degree of parallelism P for horizontal ordinate, byte number is in the histogram of ordinate, obtains interface message-efficiency monitoring histogram; By the byte number in more described interface message-efficiency monitoring histogram, select maximum number of byte
this maximum number of byte
it is then the flex point of the I/O efficiency monitoring of interface message;
Step 4: obtain best degree of parallelism
The thick degree of parallelism P recorded in histogram according to the filtration postbyte number under different degree of parallelism
slightly, and at thick degree of parallelism P
slightlyboth sides with the record of the read-write efficiency byte number reducing by 2 degree of parallelism units and carry out at every turn, to obtain best degree of parallelism P
essence; As the best degree of parallelism P of acquisition
slightlyafter, the program that application platform runs will according to P
slightlyrestart the operation of step one to step 4, until application program end of run;
At the best degree of parallelism P of acquisition
slightlyafter, application program can be carried out initiatively and passive dynamic self-adapting, so just can improve I/O bandwidth, and application program also can be made to obtain the throughput of larger acceleration effect and Geng Gao simultaneously.
The present invention can the advantage of program parallelization degree control method of perception storage system I/O bandwidth be:
1. the present invention is directed to the application of I/O intensity, by the I/O efficiency of monitoring application program, application program can be made to obtain the throughput of larger acceleration effect and Geng Gao.
2. the method for the present invention's dynamic self-adapting of using active accommodation to combine with passive adjustment is to obtain the degree of parallelism of maximum I/O efficiency, can adjust the best degree of parallelism of application program according to the change of system load accordingly.
3. degree of parallelism control method of the present invention is relative to traditional feedback, does not rely on outside input reference quantity, and reference quantity is progressively determined by the relativeness between adjustment process and the change of I/O efficiency.
4. the active accommodation in the present invention is the strategy do not had in traditional feedback control technology, and active accommodation can effectively reduce because external environment condition change has to carry out the negative effect that passive adjustment brings.
Accompanying drawing explanation
Fig. 1 is the different conditions schematic diagram of the program that application platform of the present invention runs.
Fig. 2 be of the present invention can the program parallelization degree control flow schematic diagram of perception memory device bandwidth.
Fig. 3 A is that the present invention adopts degree of parallelism to carry out the histogram of quick position.
Fig. 3 B is the histogram that the present invention adopts the inverse degree of parallelism after flex point.
Fig. 4 is active of the present invention and the schematic diagram of passive adjustment.
Embodiment
Below in conjunction with accompanying drawing, the present invention is described in further detail.
Shown in Figure 1, the program that application platform runs can be divided into three states, i.e. initial state, stable state and final state according to program I/O efficiency.
Initial state refers to that application program opens a kind of state when running with lower degree of parallelism.
A kind of state when stable state refers to application program through entering stable operation after initial state.
Final state refers to that application program maintains a kind of state when higher I/O efficiency is run after stable state.
Conversion regime between initial state and stable state is designated as quick position, and described quick position refers to that the exponential depth rule by 2 increases the process of degree of parallelism.
Conversion regime between stable state and final state is designated as fine setting, and described fine setting refers to the process detecting best degree of parallelism when stable state near determined parallel angle value.
Conversion regime between final state and stable state is designated as active accommodation, and described active accommodation refers to operation application program unconditionally being transferred to stable state from final state of periodic triggers.
Conversion regime between final state and initial state is designated as passive adjustment, and described passive adjustment refers to operation application program being transferred to initial state from final state that application program triggers when I/O efficiency change rate exceedes threshold value when final state.
Shown in Figure 2, the present invention be a kind of can the program parallelization degree control method of perception memory device bandwidth, the method is the adaptive comprehensive hardware condition according to different application platform and real time load situation, dynamically determine the best I/O efficiency performance that application program can obtain, and can the program parallelization degree control method of perception memory device bandwidth.Described program parallelization degree control method includes following treatment step:
Step one: the real-time monitoring of I/O efficiency
In the present invention, if the number of I/O efficiency monitoring time is q, first I/O efficiency monitoring time is designated as T
1, second I/O efficiency monitoring time is designated as T
2, last I/O efficiency monitoring time is designated as T
q.For convenience of description, any one I/O efficiency monitoring time is designated as T
i, described T
ithe last I/O efficiency monitoring time be called T
i-1, described T
ithe rear I/O efficiency monitoring time be called T
i+1.The program that application platform runs is designated as T the working time from initial state to stable state to final state
always(referred to as application program working time).At a T
alwayswill there be multiple I/O efficiency monitoring time working time in section, i.e. T
always=T
1+ T
2+ ... + T
i-1+ T
i+ T
i+1+ ... + T
q.The time interval between adjacent two I/O efficiency monitoring times is designated as T
p, usually set
Step 101: the byte number of the read/write function of recording address intercepting-efficiency monitoring
At first I/O efficiency monitoring time T
1in, adopt the mode that address intercepts, the entrance function read of read operation operating system provided is labeled as and reads function c_read with I/O efficiency monitoring, and records the byte number that described c_read reads function and be designated as
the entrance function write of write operation operating system provided is labeled as and writes function c_write with I/O efficiency monitoring, and records the byte number that described c_write writes function and be designated as
in the present invention, the read/write function mark adopting address interception way to carry out is to complete system-level I/O efficiency monitoring, then having, at T
1the byte number set of the read/write function of the address intercepting-efficiency monitoring under condition is expressed as
In like manner can obtain, at second I/O efficiency monitoring time T
2in, adopt the mode that address intercepts, the entrance function read of read operation operating system provided is labeled as and reads function c_read with I/O efficiency monitoring, and records the byte number that described c_read reads function and be designated as
the entrance function write of write operation operating system provided is labeled as and writes function c_write with I/O efficiency monitoring, and records the byte number that described c_write writes function and be designated as
then have, at T
2the byte number set of the read/write function of the address intercepting-efficiency monitoring under condition is expressed as
In like manner can obtain, at i-th I/O efficiency monitoring time T
iin, adopt the mode that address intercepts, the entrance function read of read operation operating system provided is labeled as and reads function c_read with I/O efficiency monitoring, and records the byte number that described c_read reads function and be designated as
the entrance function write of write operation operating system provided is labeled as and writes function c_write with I/O efficiency monitoring, and records the byte number that described c_write writes function and be designated as
then have, at T
ithe byte number set of the read/write function of the address intercepting-efficiency monitoring under condition is expressed as
In like manner can obtain, at the i-th+1 I/O efficiency monitoring time T
i+1in, adopt the mode that address intercepts, the entrance function read of read operation operating system provided is labeled as and reads function c_read with I/O efficiency monitoring, and records the byte number that described c_read reads function and be designated as
the entrance function write of write operation operating system provided is labeled as and writes function c_write with I/O efficiency monitoring, and records the byte number that described c_write writes function and be designated as
then have, at T
i+1the byte number set of the read/write function of the address intercepting-efficiency monitoring under condition is expressed as
In like manner can obtain, in the end an I/O efficiency monitoring time T
qin, adopt the mode that address intercepts, the entrance function read of read operation operating system provided is labeled as and reads function c_read with I/O efficiency monitoring, and records the byte number that described c_read reads function and be designated as
the entrance function write of write operation operating system provided is labeled as and writes function c_write with I/O efficiency monitoring, and records the byte number that described c_write writes function and be designated as
then have, at T
qthe byte number set of the read/write function of the address intercepting-efficiency monitoring under condition is expressed as
At T
alwaysunder condition, the byte number of the read/write function of all addresses intercepting-efficiency monitoring is designated as
referred to as address intercepting-efficiency monitoring-byte number
In the present invention, the read/write function mark adopting address interception way to carry out is to complete system-level I/O efficiency monitoring.
Step 102: the byte number of the read/write function of record function overloading-efficiency monitoring
At first I/O efficiency monitoring time T
1in, adopt function overloading mode to provide reading function u_read and writing function u_write with monitoring function to user; The byte number that described u_read reads function is designated as
the byte number that described u_write writes function is designated as
then have, at T
1the byte number set of the read/write function of the function overloading-efficiency monitoring under condition is expressed as
In like manner can obtain, at second I/O efficiency monitoring time T
2in, adopt function overloading mode to provide reading function u_read and writing function u_write with monitoring function to user; The byte number that described u_read reads function is designated as
the byte number that described u_write writes function is designated as
then have, at T
2the byte number set of the read/write function of the function overloading-efficiency monitoring under condition is expressed as
In like manner can obtain, at i-th I/O efficiency monitoring time T
iin, adopt function overloading mode to provide reading function u_read and writing function u_write with monitoring function to user; The byte number that described u_read reads function is designated as
the byte number that described u_write writes function is designated as
then have, at T
ithe byte number set of the read/write function of the function overloading-efficiency monitoring under condition is expressed as
In like manner can obtain, at the i-th+1 I/O efficiency monitoring time T
i+1in, adopt function overloading mode to provide reading function u_read and writing function u_write with monitoring function to user; The byte number that described u_read reads function is designated as
the byte number that described u_write writes function is designated as
then have, at T
i+1the byte number set of the read/write function of the function overloading-efficiency monitoring under condition is expressed as
In like manner can obtain, in the end an I/O efficiency monitoring time T
qin, adopt function overloading mode to provide reading function u_read and writing function u_write with monitoring function to user; The byte number that described u_read reads function is designated as
the byte number that described u_write writes function is designated as
then have, at T
qthe byte number set of the read/write function of the function overloading-efficiency monitoring under condition is expressed as
At T
alwaysunder condition, the byte number of the read/write function of all function overloading-efficiency monitorings is designated as
referred to as function overloading-efficiency monitoring-byte number
Step 103: the byte number of the read/write function of record interface information-efficiency monitoring
At first I/O efficiency monitoring time T
1in, the I/O statistical information interface adopting operating system to provide, carries out the raw data of pre-service kernel generation, and completes the monitoring of I/O efficiency, and the system monitoring obtained is read function s_read and write function s_write; The byte number that described s_read reads function is designated as
the byte number that described s_write writes function is designated as
then have, at T
1the byte number set of the read/write function of the interface message-efficiency monitoring under condition is expressed as
In like manner can obtain, at second I/O efficiency monitoring time T
2in, the I/O statistical information interface adopting operating system to provide, carries out the raw data of pre-service kernel generation, and completes the monitoring of I/O efficiency, and the system monitoring obtained is read function s_read and write function s_write; The byte number that described s_read reads function is designated as
the byte number that described s_write writes function is designated as
then have, at T
2the byte number set of the read/write function of the interface message-efficiency monitoring under condition is expressed as
In like manner can obtain, at i-th I/O efficiency monitoring time T
iin, the I/O statistical information interface adopting operating system to provide, carries out the raw data of pre-service kernel generation, and completes the monitoring of I/O efficiency, and the system monitoring obtained is read function s_read and write function s_write; The byte number that described s_read reads function is designated as
the byte number that described s_write writes function is designated as
then have, at T
ithe byte number set of the read/write function of the interface message-efficiency monitoring under condition is expressed as
In like manner can obtain, at the i-th+1 I/O efficiency monitoring time T
i+1in, the I/O statistical information interface adopting operating system to provide, carries out the raw data of pre-service kernel generation, and completes the monitoring of I/O efficiency, and the system monitoring obtained is read function s_read and write function s_write; The byte number that described s_read reads function is designated as
the byte number that described s_write writes function is designated as
then have, at T
i+1the byte number set of the read/write function of the interface message-efficiency monitoring under condition is expressed as
In like manner can obtain, in the end two I/O efficiency monitoring time T
qin, the I/O statistical information interface adopting operating system to provide, carries out the raw data of pre-service kernel generation, and completes the monitoring of I/O efficiency, and the system monitoring obtained is read function s_read and write function s_write; The byte number that described s_read reads function is designated as
the byte number that described s_write writes function is designated as
then have, at T
qthe byte number set of the read/write function of the interface message-efficiency monitoring under condition is expressed as
At T
alwaysunder condition, the byte number of the read/write function of total interface information-efficiency monitoring is designated as
referred to as interface message-efficiency monitoring-byte number
In the present invention, adopt multiple means to the monitoring of I/O efficiency, this is conducive to the user's request under adaptation different application scene.
Step 2: the denoising of the byte number of read/write function
Application program includes different I/O efficiency monitoring type MODE when running, in the present invention, described MODE includes address intercepting-efficiency monitoring type SYS, function overloading-efficiency monitoring type USR and interface message-efficiency monitoring type i NT, i.e. MODE={SYS, USR, INT}.
According to denoising condition
filtering interfering process is carried out to the byte number of read/write function, obtains filtering postbyte number.If value is " 1 ", choose
and carry out degree of parallelism extraction process, then carry out quick position, if value is " 0 ", abandon
degree of parallelism extract, directly carry out quick position.
Q represents the number of I/O efficiency monitoring time;
I represents any one I/O efficiency monitoring time;
represent at I/O efficiency monitoring time T
iunder condition, the byte number of the read/write function of any I/O efficiency monitoring type MODE;
represent at I/O efficiency monitoring time T
i+1under condition, the byte number of the read/write function of any I/O efficiency monitoring type MODE;
C
threshold valuerepresent the critical value that the judgement wave phenomenon that denoising is arranged occurs.
In the present invention, foundation denoising condition E is to address intercepting-efficiency monitoring-byte number
denoising after, the function overloading-efficiency monitoring obtained-filtration postbyte number scale is
In the present invention, foundation denoising condition E is to function overloading-efficiency monitoring-byte number
denoising after, the function overloading-efficiency monitoring obtained-filtration postbyte number scale is
In the present invention, according to denoising condition E docking port information-efficiency monitoring-byte number
denoising after, the function overloading-efficiency monitoring obtained-filtration postbyte number scale is
In the present invention, adopt denoising condition to distinguish interference and normal I/O efficiency change, the characteristic that denoising condition characterizes is exactly the continuity of time, disturbs instantaneous often, when continuous print interference occurs, can be referred to as fluctuation.By unified for above-mentioned multiple read/write function process, the monitoring of I/O efficiency can be carried out with the read-write efficiency in multiple continuous print unit interval.
Step 3: the quick position of degree of parallelism
Shown in Fig. 3 A, Fig. 3 B, in the present invention, using degree of parallelism P as horizontal ordinate, i.e. the process of corresponding application program or number of threads, ordinate is the byte number of read/write function, sets up the histogram of I/O efficiency monitoring.
The value of described degree of parallelism P is the power of 2, then have P=1,2,4,8 ..., 2
p, p represents exponential.Application program brings into operation with degree of parallelism P=1 after opening, and the filtration postbyte number scale obtained under the condition of P=1 is
after value P=1, degree of parallelism is according to P=2
prun, and obtain filtering postbyte number scale and be
To large, histogram structure is carried out to the filtration postbyte number different degree of parallelism from little according to degree of parallelism, obtain degree of parallelism-I/O efficiency histogram; By the byte number in more described degree of parallelism-I/O efficiency histogram, select maximum number of byte
this maximum number of byte
be then the flex point of I/O efficiency monitoring, as shown in Figure 3A.
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=1-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=2-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=4-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=8-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=16-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=32-filtration postbyte number scale be
By degree of parallelism P=2
pcondition under address intercepting-efficiency monitoring-filtration postbyte number scale of obtaining be
Be loaded into the filtration postbyte number under different degree of parallelism with degree of parallelism P for horizontal ordinate, byte number is in the histogram of ordinate, obtains address intercepting-efficiency monitoring histogram; By the byte number in the intercepting-efficiency monitoring histogram of more described address, select maximum number of byte
this maximum number of byte
it is then the flex point of the I/O efficiency monitoring of address intercepting.
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=1-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=2-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=4-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=8-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=16-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=32-filtration postbyte number scale be
By degree of parallelism P=2
pcondition under function overloading-efficiency monitoring-filtration postbyte number scale of obtaining be
Be loaded into the filtration postbyte number under different degree of parallelism with degree of parallelism P for horizontal ordinate, byte number is in the histogram of ordinate, obtains function overloading-efficiency monitoring histogram; By the byte number in more described function overloading-efficiency monitoring histogram, select maximum number of byte
this maximum number of byte
it is then the flex point of the I/O efficiency monitoring of function overloading.
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=1-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=2-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=4-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=8-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=16-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=32-filtration postbyte number scale be
By degree of parallelism P=2
pcondition under interface message-efficiency monitoring-filtration postbyte number scale of obtaining be
Be loaded into the filtration postbyte number under different degree of parallelism with degree of parallelism P for horizontal ordinate, byte number is in the histogram of ordinate, obtains interface message-efficiency monitoring histogram; By the byte number in more described interface message-efficiency monitoring histogram, select maximum number of byte
this maximum number of byte
it is then the flex point of the I/O efficiency monitoring of interface message.The degree of parallelism P at described flex point place is designated as thick degree of parallelism P
slightly.
In the present invention, the state machine in composition graphs 1 represents and is described below, and is in initial state, enters stable state, final state afterwards after the process of quick position when application program is opened.Be described below in conjunction with Fig. 3 A, run with degree of parallelism 1,2,4,8,16,32 successively according to mentioned above principle application program, degree of parallelism increases with exponential law, can see that the I/O efficiency of application program increases successively, apply afterwards when running with degree of parallelism 16 and can see that I/O efficiency starts to reduce, in order to ensure certain reliability, continue to increase degree of parallelism to 32, the trend that I/O efficiency is not recovered, just no longer increasing degree of parallelism and can obtain I/O efficiency performance flex point is so degree of parallelism 8, i.e. P
slightly=8.
Step 4: obtain best degree of parallelism
Flex point P is there is at Fig. 3 A
slightlyafter=8, by carrying out the Linear fine tuning of degree of parallelism with less degree of parallelism adjustment stride, to obtain best read-write efficiency, i.e. smart degree of parallelism P
essence.State machine in composition graphs 1 represents and is described below, and after application program enters stable state, the process through fine setting enters final state.Be described below in conjunction with Fig. 3 B, according to mentioned above principle application program at flex point P
slightlythe both sides of=8 are finely tuned to reduce by 2 degree of parallelism units at every turn, at every turn the method used in figure from degree of parallelism 14, reduces by 2 degree of parallelism unit linear successively successively decrease, until degree of parallelism is down to 4, the smart degree of parallelism P obtaining best I/O performance can be clear that
essence10(P in fact
essence=10), application program will with best degree of parallelism P afterwards
essencethe operation of=10 read/write function again.In the present invention, described best degree of parallelism P
essencefine setting be at thick degree of parallelism P
slightlyboth sides with the record of the read-write efficiency byte number reducing by 2 degree of parallelism units and carry out at every turn.
In the present invention, best degree of parallelism P is designated as by finely tuning the degree of parallelism obtained
essence.As the best degree of parallelism P of acquisition
essenceafter, the program that application platform runs will according to P
essencerestart the operation of step one to step 4, until application program end of run.At the best degree of parallelism P of acquisition
essenceafter, application program can be carried out initiatively and passive dynamic self-adapting, so just can improve I/O bandwidth, and application program also can be made to obtain the throughput of larger acceleration effect and Geng Gao simultaneously.
What the present invention proposed can the program parallelization degree control method of perception memory device bandwidth be bring into operation until application program terminates to be a control procedure from application program.
The dynamic self-adapting of best degree of parallelism
(1) active self-adaptative adjustment: in the present invention, described active accommodation refers to operation application program unconditionally being transferred to stable state from final state of periodic triggers.The status mechanism conversion that application program is run has initial state, stable state and final state.As the best degree of parallelism P of acquisition
essenceafter, final state carries out the real-time monitoring of I/O efficiency to stable state with active accommodation, and then carry out the re-treatment of step 2 to step 4.State machine in composition graphs 1 represents and is described below, and application program, after entering final state, when active accommodation is triggered, enters stable state by unconditional.In this process, only with best degree of parallelism P
essencecarry out the process of step one to step 4, to obtain the real-time change of the I/O bandwidth of application program, application program also can be made to obtain the throughput of larger acceleration effect and Geng Gao simultaneously.
(2) passive self-adaptative adjustment: in the present invention, described passive adjustment refers to operation application program being transferred to initial state from final state that application program triggers when I/O efficiency change rate exceedes threshold value when final state.As the best degree of parallelism P of acquisition
essenceafter, final state carries out the real-time monitoring of I/O efficiency to initial state with passive adjustment, and then carry out the re-treatment of step 2 to step 4.
Application program on application platform is at best degree of parallelism P
essenceduring lower operation, when the change of the I/O read-write efficiency rate of change B of application program is greater than the threshold parameter B of the passive adjustment of given triggering
threshold valuetime, then open the quick position of a new round and the process of fine setting immediately, passive adjustment that Here it is.State machine in composition graphs 1 represents and is described below, and application program, after entering final state, when passive adjustment is triggered, enters initial state by unconditional, to trigger the detection process of a new round.
The condition triggering passive adjustment is B > B
threshold value, and
represent at I/O efficiency monitoring time T
iunder condition, the byte number of the read/write function of any I/O efficiency monitoring type MODE.
represent at I/O efficiency monitoring time T
i+1under condition, the byte number of the read/write function of any I/O efficiency monitoring type MODE.
The technology of active accommodation and passive adjustment is described below in conjunction with Fig. 4, can see that the I/O efficiency of application program at I/O efficiency monitoring time T2 place is lower than present threshold value, passive adjustment is triggered, the quick position process of a new round is carried out in application immediately, and detection from lower degree of parallelism completes detection process arrival final state until arrive I/O efficiency monitoring time T3.After have employed active accommodation technology, be applied in I/O efficiency monitoring time T1 place and just trigger active accommodation, active accommodation only can cause the action of finely tuning, after fine setting, reach final state in the I/O efficiency monitoring time T3 moment.As can be seen from above process, after adjustment technology is taken the initiative in application, the active accommodation that then application program triggers when T1 can avoid application program to trigger passive adjustment when T2, thus make the average I/O efficiency of application program between T1 to T3 achieve larger raising, reduce dynamic development adjustment to the normal negative effect run of application
For I/O intensive applications, improving throughput is its important performance indexes, from the lifting that Physical layer is exactly its I/O efficiency, in order to can better control I/O efficiency, the I/O efficiency of application can be supplied to user's application as feedback information, maximize I/O efficiency information by feed-back regulatory mechanism.In feed-back regulatory mechanism, treat and adjust input obtaining original output by the adjustment algorithm of controller, and using original output be supplied to controller with reference to the difference that exports as feedback quantity and use, to being finally bordering on most the output of reference quantity.
Claims (4)
1. can the program parallelization degree control method of perception memory device bandwidth, the program that application platform runs can be divided into three states, i.e. initial state, stable state and final state according to program I/O efficiency;
Initial state refers to that application program opens a kind of state when running with lower degree of parallelism;
A kind of state when stable state refers to application program through entering stable operation after initial state;
Final state refers to that application program maintains a kind of state when higher I/O efficiency is run after stable state;
It is characterized in that: described program parallelization degree control method includes following treatment step:
Step one: the real-time monitoring of I/O efficiency
Step 101: the byte number of the read/write function of recording address intercepting-efficiency monitoring
At i-th I/O efficiency monitoring time T
iin, adopt the mode that address intercepts, the entrance function read of read operation operating system provided is labeled as and reads function c_read with I/O efficiency monitoring, and records the byte number that described c_read reads function and be designated as
the entrance function write of write operation operating system provided is labeled as and writes function c_write with I/O efficiency monitoring, and records the byte number that described c_write writes function and be designated as
then have, at T
ithe byte number set of the read/write function of the address intercepting-efficiency monitoring under condition is expressed as
In application program running time T
alwaysunder condition, the byte number of the read/write function of all addresses intercepting-efficiency monitoring is designated as
referred to as address intercepting-efficiency monitoring-byte number
Step 102: the byte number of the read/write function of record function overloading-efficiency monitoring
At the i-th+1 I/O efficiency monitoring time T
i+1in, adopt function overloading mode to provide reading function u_read and writing function u_write with monitoring function to user; The byte number that described u_read reads function is designated as
the byte number that described u_write writes function is designated as
then have, at T
i+1the byte number set of the read/write function of the function overloading-efficiency monitoring under condition is expressed as
At T
alwaysunder condition, the byte number of the read/write function of all function overloading-efficiency monitorings is designated as
referred to as function overloading-efficiency monitoring-byte number
Step 103: the byte number of the read/write function of record interface information-efficiency monitoring
At i-th I/O efficiency monitoring time T
iin, the I/O statistical information interface adopting operating system to provide, carries out the raw data of pre-service kernel generation, and completes the monitoring of I/O efficiency, and the system monitoring obtained is read function s_read and write function s_write; The byte number that described s_read reads function is designated as
the byte number that described s_write writes function is designated as
then have, at T
ithe byte number set of the read/write function of the interface message-efficiency monitoring under condition is expressed as
At T
alwaysunder condition, the byte number of the read/write function of total interface information-efficiency monitoring is designated as
referred to as interface message-efficiency monitoring-byte number
Step 2: the denoising of the byte number of read/write function
Application program includes different I/O efficiency monitoring type MODE when running, namely MODE includes address intercepting-efficiency monitoring type SYS, function overloading-efficiency monitoring type USR and interface message-efficiency monitoring type i NT, i.e. MODE={SYS, USR, INT};
According to denoising condition
filtering interfering process is carried out to the byte number of read/write function, obtains filtering postbyte number; If value is " 1 ", choose
and carry out degree of parallelism extraction process, then carry out quick position, if value is " 0 ", abandon
degree of parallelism extract, directly carry out quick position;
Q represents the number of I/O efficiency monitoring time;
I represents any one I/O efficiency monitoring time;
represent at I/O efficiency monitoring time T
iunder condition, the byte number of the read/write function of any I/O efficiency monitoring type MODE;
represent at I/O efficiency monitoring time T
i+1under condition, the byte number of the read/write function of any I/O efficiency monitoring type MODE;
C
threshold valuerepresent the critical value that the judgement wave phenomenon that denoising is arranged occurs;
According to denoising condition E to address intercepting-efficiency monitoring-byte number
denoising after, the function overloading-efficiency monitoring obtained-filtration postbyte number scale is
According to denoising condition E to function overloading-efficiency monitoring-byte number
denoising after, the function overloading-efficiency monitoring obtained-filtration postbyte number scale is
According to denoising condition E docking port information-efficiency monitoring-byte number
denoising after, the function overloading-efficiency monitoring obtained-filtration postbyte number scale is
Step 3: the quick position of degree of parallelism
The value of described degree of parallelism P is the power of 2, then have P=1,2,4,8 ..., 2
p, p represents exponential; Application program brings into operation with degree of parallelism P=1 after opening, and the filtration postbyte number scale obtained under the condition of P=1 is
after value P=1, degree of parallelism is according to P=2
prun, and obtain filtering postbyte number scale and be
To large, histogram structure is carried out to the filtration postbyte number different degree of parallelism from little according to degree of parallelism, obtain degree of parallelism-I/O efficiency histogram; By the byte number in more described degree of parallelism-I/O efficiency histogram, select maximum number of byte
this maximum number of byte
be then the flex point of I/O efficiency monitoring, as shown in Figure 3A;
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=1-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=2-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=4-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=8-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=16-filtration postbyte number scale be
By the address intercepting-efficiency monitoring obtained under the condition of degree of parallelism P=32-filtration postbyte number scale be
By degree of parallelism P=2
pcondition under address intercepting-efficiency monitoring-filtration postbyte number scale of obtaining be
Be loaded into the filtration postbyte number under different degree of parallelism with degree of parallelism P for horizontal ordinate, byte number is in the histogram of ordinate, obtains address intercepting-efficiency monitoring histogram; By the byte number in the intercepting-efficiency monitoring histogram of more described address, select maximum number of byte
this maximum number of byte
it is then the flex point of the I/O efficiency monitoring of address intercepting;
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=1-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=2-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=4-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=8-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=16-filtration postbyte number scale be
By the function overloading-efficiency monitoring obtained under the condition of degree of parallelism P=32-filtration postbyte number scale be
By degree of parallelism P=2
pcondition under function overloading-efficiency monitoring-filtration postbyte number scale of obtaining be
Be loaded into the filtration postbyte number under different degree of parallelism with degree of parallelism P for horizontal ordinate, byte number is in the histogram of ordinate, obtains function overloading-efficiency monitoring histogram; By the byte number in more described function overloading-efficiency monitoring histogram, select maximum number of byte
this maximum number of byte
it is then the flex point of the I/O efficiency monitoring of function overloading;
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=1-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=2-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=4-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=8-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=16-filtration postbyte number scale be
By the interface message-efficiency monitoring obtained under the condition of degree of parallelism P=32-filtration postbyte number scale be
By degree of parallelism P=2
pcondition under interface message-efficiency monitoring-filtration postbyte number scale of obtaining be
Be loaded into the filtration postbyte number under different degree of parallelism with degree of parallelism P for horizontal ordinate, byte number is in the histogram of ordinate, obtains interface message-efficiency monitoring histogram; By the byte number in more described interface message-efficiency monitoring histogram, select maximum number of byte
this maximum number of byte
it is then the flex point of the I/O efficiency monitoring of interface message;
Step 4: obtain best degree of parallelism
The thick degree of parallelism P recorded in histogram according to the filtration postbyte number under different degree of parallelism
slightly, and at thick degree of parallelism P
slightlyboth sides with the record of the read-write efficiency byte number reducing by 2 degree of parallelism units and carry out at every turn, to obtain best degree of parallelism P
essence; As the best degree of parallelism P of acquisition
essenceafter, the program that application platform runs will according to P
essencerestart the operation of step one to step 4, until application program end of run;
At the best degree of parallelism P of acquisition
essenceafter, application program can be carried out initiatively and passive dynamic self-adapting, so just can improve I/O bandwidth, and application program also can be made to obtain the throughput of larger acceleration effect and Geng Gao simultaneously.
2. according to claim 1 can the program parallelization degree control method of perception memory device bandwidth, it is characterized in that: the active self-adaptative adjustment that application program is carried out is: as the best degree of parallelism P of acquisition
essenceafter, final state carries out the real-time monitoring of I/O efficiency to stable state with active accommodation, and then carry out the re-treatment of step 2 to step 4.
3. according to claim 1 can the program parallelization degree control method of perception memory device bandwidth, it is characterized in that: the passive self-adaptative adjustment that application program is carried out is: as the best degree of parallelism P of acquisition
essenceafter, final state carries out the real-time monitoring of I/O efficiency to initial state with passive adjustment, and then carry out the re-treatment of step 2 to step 4; Application program on application platform is at best degree of parallelism P
essenceduring lower operation, when the change of the I/O read-write efficiency rate of change B of application program is greater than the threshold parameter B of the passive adjustment of given triggering
threshold valuetime, then open the quick position of a new round and the process of fine setting immediately; The condition triggering passive adjustment is B > B
threshold value, and
represent at I/O efficiency monitoring time T
iunder condition, the byte number of the read/write function of any I/O efficiency monitoring type MODE,
represent at I/O efficiency monitoring time T
i+1under condition, the byte number of the read/write function of any I/O efficiency monitoring type MODE.
4. according to claim 1 can the program parallelization degree control method of perception memory device bandwidth, it is characterized in that: the method is the adaptive comprehensive hardware condition according to different application platform and real time load situation, dynamically determine the best I/O efficiency performance that application program can obtain, and can the program parallelization degree control method of perception memory device bandwidth.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310477537.2A CN103677757B (en) | 2013-10-14 | 2013-10-14 | A kind of can the program parallelization degree control method of perception memory device bandwidth |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310477537.2A CN103677757B (en) | 2013-10-14 | 2013-10-14 | A kind of can the program parallelization degree control method of perception memory device bandwidth |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103677757A CN103677757A (en) | 2014-03-26 |
| CN103677757B true CN103677757B (en) | 2016-01-06 |
Family
ID=50315436
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310477537.2A Expired - Fee Related CN103677757B (en) | 2013-10-14 | 2013-10-14 | A kind of can the program parallelization degree control method of perception memory device bandwidth |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103677757B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109324894A (en) * | 2018-08-13 | 2019-02-12 | 中兴飞流信息科技有限公司 | PC cluster method, apparatus and computer readable storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7437546B2 (en) * | 2005-08-03 | 2008-10-14 | Intel Corporation | Multiple, cooperating operating systems (OS) platform system and method |
| CN102521047A (en) * | 2011-11-15 | 2012-06-27 | 重庆邮电大学 | Method for realizing interrupted load balance among multi-core processors |
-
2013
- 2013-10-14 CN CN201310477537.2A patent/CN103677757B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7437546B2 (en) * | 2005-08-03 | 2008-10-14 | Intel Corporation | Multiple, cooperating operating systems (OS) platform system and method |
| CN102521047A (en) * | 2011-11-15 | 2012-06-27 | 重庆邮电大学 | Method for realizing interrupted load balance among multi-core processors |
Non-Patent Citations (1)
| Title |
|---|
| 多核处理器大规模并行系统中的任务分配问题及算法;刘轶等;《小型微型计算机系统》;20080515;第29卷(第5期);第972页至第975页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103677757A (en) | 2014-03-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8239613B2 (en) | Hybrid memory device | |
| US8019790B2 (en) | System and method of dynamically changing file representations | |
| US11392565B1 (en) | Optimizing data compression in a storage system | |
| KR101629879B1 (en) | Multicore computer system with cache use based adaptive scheduling | |
| US10089014B2 (en) | Memory-sampling based migrating page cache | |
| US20110191534A1 (en) | Dynamic management of destage tasks in a storage controller | |
| US20080229071A1 (en) | Prefetch control apparatus, storage device system and prefetch control method | |
| Niu et al. | Hybrid storage systems: A survey of architectures and algorithms | |
| WO2014209234A1 (en) | Method and apparatus for hot data region optimized dynamic management | |
| US20190065404A1 (en) | Adaptive caching in a storage device | |
| KR20170002866A (en) | Adaptive Cache Management Method according to the Access Chracteristics of the User Application in a Distributed Environment | |
| US11500577B2 (en) | Method, electronic device, and computer program product for data processing | |
| Wen et al. | Openmem: Hardware/software cooperative management for mobile memory system | |
| US11640243B2 (en) | Supervised learning with closed loop feedback to improve input output consistency of solid state drives | |
| CN103677757B (en) | A kind of can the program parallelization degree control method of perception memory device bandwidth | |
| CN115129228A (en) | Data relocation system | |
| US20240061782A1 (en) | Method and device for data caching | |
| US20230100110A1 (en) | Computing resource management method, electronic equipment and program product | |
| Chen et al. | DA-GC: A dynamic adjustment garbage collection method considering wear-leveling for SSD | |
| Micheloni et al. | Hybrid storage | |
| Micheloni et al. | Hybrid storage systems | |
| Garg et al. | CrossPrefetch: Accelerating I/O Prefetching for Modern Storage | |
| US12019532B2 (en) | Distributed file system performance optimization for path-level settings using machine learning | |
| US20230178136A1 (en) | Memory device detecting weakness of operation pattern and method of operating the same | |
| US20220334944A1 (en) | Distributed file system performance optimization for path-level settings using machine learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210420 Address after: 100160, No. 4, building 12, No. 128, South Fourth Ring Road, Fengtai District, Beijing, China (1515-1516) Patentee after: Kaixi (Beijing) Information Technology Co.,Ltd. Address before: 100191 Haidian District, Xueyuan Road, No. 37, Patentee before: BEIHANG University |
|
| TR01 | Transfer of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160106 Termination date: 20211014 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |