Memory activity driven adaptive performance measurement

• US 10,067,672 B2
• Filed: 08/31/2015
• Issued: 09/04/2018
• Est. Priority Date: 08/31/2015
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Chinnakkonda Vidyapoornachary, Diyanesh B., Purushotham, Srinivas B.
• Primary Examiner(s): Bertram, Ryan
• Assistant Examiner(s): Ta, Trang K
• Application Number:
  US14/840,503
• Publication Number:
  US 20170060476A1
• Time in Patent Office: 1,100 Days
• CPC Class Codes: G06F 1/20 Cooling means
• CPC Class Codes: G06F 1/206 comprising thermal management
• CPC Class Codes: G06F 3/0604 Improving or facilitating administration, e.g. storage management
• CPC Class Codes: G06F 3/0634 by changing the state or mode of one or more devices
• CPC Class Codes: G06F 3/0673 Single storage device
• CPC Class Codes: G11C 16/10 Programming or data input circuits
• CPC Class Codes: G11C 2029/0409 Online test
• CPC Class Codes: G11C 29/027 in fuses
• CPC Class Codes: G11C 29/028 with adaption or trimming of parameters
• CPC Class Codes: G11C 5/143 Detection of memory cassette insertion or removal; Continuity checks of supply or ground lines; Detection of supply variations, interruptions or levels G11C5/148 takes precedence; Switching between alternative supplies G11C5/141 takes precedence
• CPC Class Codes: G11C 7/04 with means for avoiding disturbances due to temperature effects
• CPC Class Codes: G11C 7/24 Memory cell safety or protection circuits, e.g. arrangements for preventing inadvertent reading or writing; Status cells; Test cells
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A method of monitoring memory performance comprises selecting a first portion of memory from two or more portions of memory in accordance with an adaptive mode indicated by configuration bits in a control register; monitoring memory accesses to the selected portion of memory during a first sampling period; selecting a different portion of memory from the two or more portions of memory in accordance with the adaptive mode for monitoring the different portion of memory in a subsequent sampling period; monitoring memory accesses to the different portion of memory during the subsequent sampling period; recording a respective number of memory accesses for each portion of memory over a plurality of sampling periods; and generating one or more interrupts to output data regarding the monitored memory accesses for data analysis.

Claims

1. A memory controller comprising:

a plurality of memory interfaces, each of the plurality of memory interfaces communicatively coupled to a respective memory module comprising at least one rank and at least one bank;

a profiler configured to maintain a record of respective memory accesses to each rank and to each bank of each memory module;

a control unit configured to adaptively select which portion of the respective memory modules coupled to the plurality of memory interfaces to monitor traffic access patterns for each of a plurality of sampling periods based on one of a track hotness mode or a track coldness mode as indicated by bits in a control register; and

a counter configured to count the respective number of memory accesses to the selected portion of the memory modules during the respective sampling period;

wherein the control unit is further configured to generate interrupts periodically to provide data from the counter to a processor for workload and usage data analysis;

wherein the control unit is further configured to analyze, using the workload and usage data analysis, how the memory modules behave when peak bandwidth workload is run and how the power management is handled when low workload bandwidth is run;

wherein when configured to adaptively select which portion of the respective memory modules to monitor based on the track hotness mode, the control unit is configured to retrieve data from the profiler regarding the number of memory accesses for each rank and to select, for monitoring in a next sampling period of the plurality of sampling periods, only the portion of the respective memory modules having the highest number of memory accesses during a given time window; and

wherein when configured to adaptively select which portion of the respective memory modules to monitor based on the track coldness mode, the control unit is configured to retrieve data from the profiler regarding the number of memory accesses for each rank and to select, for monitoring in a next sampling period of the plurality of sampling periods, only the portion of the respective memory modules having the lowest number of memory accesses during a given time window.

2. The memory controller of claim 1, wherein the control unit is configured to adaptively select which portion of the respective memory modules to monitor based on the round robin mode;

wherein the control unit is configured to rotate selection of each rank of the respective memory modules coupled to the plurality of memory interfaces such that each rank is monitored in a respective sampling period.

3. The memory controller of claim 2, wherein the memory controller further comprises a memory buffer;

wherein the control unit is configured to rotate selection of each rank of the respective memory modules coupled to the plurality of memory interfaces and the memory buffer such that each rank and memory buffer events are monitored in a respective sampling period.

4. The memory controller of claim 1, wherein the control unit is further configured to generate interrupts at predetermined time intervals.

5. The memory controller of claim 1, wherein the control unit is further configured to generate an interrupt when the counter has reached a predetermined limit.

6. The memory controller of claim 1, wherein the given time period is a moving time window of previous sampling periods;

wherein, in the track hotness mode, the control unit is configured to select, for monitoring in the next sampling period, only the portion of the respective memory modules having the highest cumulative count of memory accesses for the moving time window; and

wherein, in the track coldness mode, the control unit is configured to select, for monitoring in the next sampling period, only the portion of the respective memory modules having the lowest cumulative count of memory accesses for the moving time window.

7. The memory controller of claim 6, wherein, as the moving time window changes, the control unit is configured to determine if a different portion other than the selected portion has the highest cumulative count of memory accesses for the moving time window;

and, in response to determining that the different portion has the highest cumulative count of memory accesses for the moving time window, select the different portion for monitoring in a subsequent sampling period.

8. The memory controller of claim 1, wherein the plurality of sampling periods are defined by one or more user inputs.

9. The memory controller of claim 1, wherein the plurality of sampling periods includes a first sampling period and a second sampling period, and wherein the second sampling period is defined dynamically based on the record of respective memory accesses to the selected portion of the respective memory modules during the first sampling period.

10. The memory controller of claim 1, wherein the control register contains configuration bits in an adaptive mode field, a rank select field, a bank select field, and a device level select field.

11. The memory controller of claim 1, wherein the usage data analysis enables modification of an amount of energy supplied to memory modules.

12. The memory controller of claim 1, wherein the developing the detailed memory characterization for both the peak bandwidth driving workload and the low bandwidth workload further comprises developing a rank level stress distribution across the two or more portions of memory.

13. The memory controller of claim 1, wherein the developing the detailed memory characterization for both the peak bandwidth driving workload and the low bandwidth workload further comprises collecting data regarding how power management is handled when a low workload bandwidth is run.

14. A memory sub-system comprising:

a processor;

a plurality of memory modules each comprising at least one rank and at least one bank; and

a memory controller having a plurality of memory interfaces, each memory interface communicatively coupled to a respective one of the plurality of memory modules;

wherein the memory controller further comprises a performance measurement component configured to maintain a record of respective memory accesses to each rank and to each bank of each memory module and to adaptively select which portion of the respective memory modules to monitor traffic access patterns for each of a plurality of sampling periods based on one of a track hotness mode or a track coldness mode as indicated by bits in a control register;

wherein the performance measurement component is further configured to count the respective number of memory accesses to the selected portion of the memory modules during the respective sampling period and to generate interrupts periodically to provide the counted number of memory accesses to the processor for workload and usage data analysis;

wherein the performance measurement component is further configured to analyze, using the workload and usage data analysis, how the memory modules behave when peak bandwidth workload is run and how the power management is handled when low workload bandwidth is run;

wherein when configured to adaptively select which portion of the respective memory modules to monitor based on the track hotness mode, the performance measurement component is configured to select, for monitoring in a next sampling period of the plurality of sampling periods, only the portion of the respective memory modules having the highest number of memory accesses during a given time window; and

wherein when configured to adaptively select which portion of the respective memory modules to monitor based on the track coldness mode, the performance measurement component is configured to select, for monitoring in a next sampling period of the plurality of sampling periods, only the portion of the respective memory modules having the lowest number of memory accesses during a given time window.

15. The memory sub-system of claim 14, wherein the performance measurement component is configured to adaptively select which portion of the respective memory modules to monitor based on the round robin mode;

wherein the performance measurement component is configured to rotate selection of each rank of the respective memory modules coupled to the plurality of memory interfaces such that each rank is monitored in a respective sampling period.

16. The memory sub-system of claim 15, wherein the memory controller further comprises a memory buffer;

wherein the performance measurement component is configured to rotate selection of each rank of the respective memory modules coupled to the plurality of memory interfaces and the memory buffer such that each rank and memory buffer events are monitored in a respective sampling period.

17. The memory sub-system of claim 14, wherein the performance measurement unit is further configured to generate interrupts at predetermined time intervals.

18. The memory sub-system of claim 14, wherein the given time period is a moving time window of previous sampling periods;

wherein, in the track hotness mode, the performance measurement component is configured to select, for monitoring in the next sampling period, only the portion of the respective memory modules having the highest cumulative count of memory accesses for the moving time window; and

wherein, in the track coldness mode, the performance measurement component is configured to select, for monitoring in the next sampling period, only the portion of the respective memory modules having the lowest cumulative count of memory accesses for the moving time window.

19. The memory sub-system of claim 18, wherein, as the moving time window changes, the performance measurement component is configured to determine if a different portion other than the selected portion has the highest cumulative count of memory accesses for the moving time window;

and, in response to determining that the different portion has the highest cumulative count of memory accesses for the moving time window, select the different portion for monitoring in a subsequent sampling period.