Method and apparatus for performing simulated write operations - Patents.com
Simulated writes address power fluctuations in memory systems by incrementally adjusting power levels, stabilizing the power supply and preventing errors during read/write transitions.
Patent Information
- Application Number
- JP2024535341
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2021-12-20
- Filing Date
- 2022-10-26
- Publication Date
- 2024-12-05
AI Technical Summary
The sudden and significant fluctuations in power requirements during simultaneous memory operations cause potential errors and strain on the power supply due to large power increases and decreases when transitioning from an idle state to read or write states.
Implementing simulated write operations, also known as phantom writes, which incrementally increase and decrease power levels to smooth the transition from idle to read/write states without fully activating the physical layer circuitry, thereby managing power fluctuations.
This approach reduces power strain on the supply by smoothing power transitions, preventing errors and ensuring stable operation during memory transactions.
Smart Images

Figure 2024545212000001_ABST
Abstract
Description
[Technical field]
[0001] (CROSS REFERENCE TO RELATED APPLICATIONS) This application claims the benefit of U.S. Patent Application No. 17 / 556,958, filed December 20, 2021, the contents of which are incorporated herein by reference. [Background technology]
[0002] During periods when data is being written to or read from memory from an idle period, there is a power increase that must occur to perform the read or write. Depending on how many memory channels require some degree of simultaneous reading or writing, the amount of power ramp up to go from an idle state to a read or write state can be quite large.
[0003] A more detailed understanding may be had from the following description, given by way of example in conjunction with the accompanying drawings, in which: [Brief description of the drawings]
[0004] [Figure 1] 1 is a block diagram of an example device in which one or more features of the present disclosure may be implemented. [Diagram 2] FIG. 2 is a block diagram of a portion of the example system of FIG. [Diagram 3] 2 is an exemplary timing diagram illustrating read and write operations to memory regions of the exemplary system of FIG. 1. [Figure 4] FIG. 2 is an exemplary timing diagram illustrating read and write operations to a memory region including a series of simulated writes. [Diagram 5] FIG. 1 illustrates an example block diagram of the physical (PHY) layer during a simulated write operation. [Figure 6] 4 is a flow diagram of an exemplary method for performing a simulated write operation, according to one embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0005] The method and apparatus are expanded upon in more detail below, but a method for performing a simulated write (also referred to herein as a phantom write) is briefly described herein. During a read or write operation from an idle operation, a power increase occurs to perform the operation.
[0006] Depending on the number of reads / writes that may occur substantially simultaneously across multiple memory channels, the amount of power required to bring the circuitry required to perform a read or write up to a level to perform the operation may have significant variations over a very short period of time. That is, if many read or write operations are performed at once, the circuitry in the physical layer (PHY) that accesses memory banks in the memory may need to draw more power from the power system than is immediately available, potentially resulting in errors.
[0007] To mitigate this occurrence, techniques are described herein for introducing a phantom write, a simulated write operation that does not contain all of the attributes of a traditional write operation. A phantom write operation incrementally increases power from an idle state before or after a read / write event in the PHY to avoid large individual increases or decreases in power.
[0008] Additionally, as will be described in more detail below, an SOC is a device in which many components of an entire system reside on a chip. For example, an SOC may include a processor, memory, storage, input and output drivers, and other components on a single chip.
[0009] A method for performing a simulated write in a computer system includes, in response to a scheduled memory operation determined by a memory controller, sending a simulated write operation to a physical layer circuit (PHY) to increase circuit power without enabling an output of the PHY until the memory operation begins, and in response to the memory operation being completed, sending the simulated write operation to the PHY to decrease circuit power.
[0010] A computer system for performing simulated writes includes a physical layer circuit (PHY) and a memory controller operatively coupled to and in communication with the PHY. The memory controller transmits a simulated write operation to the PHY in response to a scheduled memory operation to increase circuit power without enabling an output of the PHY until the memory operation begins. The memory controller transmits a simulated write operation to the PHY in response to the memory operation being completed to reduce circuit power.
[0011] A memory controller operatively coupled to and in communication with a physical layer circuit (PHY) includes a processor. The processor, in response to a scheduled memory operation, sends a simulated write operation to the PHY to increase circuit power without enabling an output of the PHY until the memory operation begins. The processor, in response to the memory operation being completed, sends a simulated write operation to the PHY to reduce circuit power.
[0012] 1 is a block diagram of an example device 100 capable of implementing one or more features of the present disclosure. Device 100 may include, for example, a computer, a server, a gaming device, a handheld device, a set-top box, a television, a mobile phone, or a tablet computer. Device 100 includes a processor 102, memory 104, storage 106, one or more input devices 108, and one or more output devices 110. For purposes of illustration, output device 110 is shown as a display 110, but it should be understood that other output devices may be included.
[0013] Device 100 may also optionally include input drivers 112 and output drivers 114. Additionally, device 100 includes a memory controller 115 in communication with processor 102 and memory 104, and which may also communicate with external memory 116. In some embodiments, memory controller 115 is included within processor 102. It should be understood that device 100 may include additional components not shown in FIG.
[0014] As described above, the processor 102, memory 104, storage 106, input drivers 112, output drivers 114, and memory controller 115 may be included on the SOC 101. Additionally, the memory controller 115 may include a processor or processing circuitry for performing operations.
[0015] In various alternatives, the processor 102 may include a central processing unit (CPU), a graphics processing unit (GPU), a CPU and a GPU located on the same die, or one or more dies, where each processor die may be a CPU or a GPU. In various alternatives, the memory 104 may be located on the same die as the processor 102 or may be located separately from the processor 102. The memory 104 may include volatile or non-volatile memory (e.g., random access memory (RAM), dynamic RAM, cache).
[0016] Storage 106 includes fixed or removable storage (e.g., hard disk drive, solid state drive, optical disk, flash drive). Input devices 108 include, but are not limited to, a keyboard, a keypad, a touch screen, a touch pad, a detector, a microphone, an accelerometer, a gyroscope, a biometric scanner, or a network connection (e.g., a wireless local area network card for transmitting and / or receiving wireless IEEE 802 signals). Output devices 110 include, but are not limited to, a display, a speaker, a printer, a haptic feedback device, one or more optical, antennae, or network connections (e.g., a wireless local area network card for transmitting and / or receiving wireless IEEE 802 signals).
[0017] The input driver 112 communicates with the processor 102 and the input device 108, allowing the processor 102 to receive input from the input device 108. The output driver 114 communicates with the processor 102 and the output device 110, allowing the processor 102 to send output to the output device 110. It should be noted that the input driver 112 and the output driver 114 are optional components and that the device 100 operates in the same manner when the input driver 112 and the output driver 114 are not present.
[0018] The external memory 116 may be similar to the memory 104 and may exist in the form of off-chip memory. Additionally, the external memory may be memory resident in a server with which the memory controller 115 communicates via a network interface to access the external memory 116. For example, the external memory may be a dynamic random access memory (DRAM) that is external to the SOC 101. Additionally, the external memory 116 may be external to the system 100.
[0019] Figure 2 is a block diagram of a portion of the example system 100 of Figure 1. In particular, Figure 2 shows a memory controller 115 (e.g., shown as MC) and a physical layer circuit (PHY) 119 as part of SOC 101. PHY 119 communicates with memory banks (shown as B0, B1, ..., Bn) of external memory 116. MC 115 controls the flow of data to PHY 119, which then effects reads and writes to the memory banks of memory 104 via bus wires 210.
[0020] Each of the bus wires 210 communicates from the PHY 119 to memory bank B of the external memory 116, allowing data to be written to memory bank B when a command to write the data is received by the PHY 119. Additionally, each of the bus wires 210 allows data to be transmitted from memory bank B back to the PHY when a command to read the data is received by the PHY 119.
[0021] Although memory controller 115 is shown as communicating with external memory 116 (DRAM), it should be noted that in the context of the following description, the memory controller may also be in communication with memory 104 or any other memory utilized by system 100.
[0022] 2, when an operation is requested from memory controller 115 to PHY 119 to perform an operation on data in memory 116, for example, a signal is sent by memory controller 115 to PHY 119 to perform the operation. For example, when a read or write operation is passed from MC 115 to PHY 119, PHY 119 transmits and receives from memory 104 over bus wires 210. These operations utilize power when transitioning from an idle state to a read or write state, as described below in FIG.
[0023] 1 and 2 together, when a command is issued by memory controller 115 to perform a read or write operation, memory controller 115 passes the command to PHY 119 for the actual transmission or reception of data. Each of bus wires 210 is connected to a memory bank B, and PHY 119 enables data to be transmitted to one or more of memory banks B, depending on the need of the read or write operation.
[0024] As mentioned above, each bus wire 210 transmits / carries information independently of the others, and therefore each bus wire 210 incurs increased power when that bus wire 210 is utilized to transmit data.
[0025] For example, when a first one of bus wires 210 is energized to transmit data from PHY 119 to memory bank B in external memory 116, a power increase occurs to power the transmission. When additional bus wires 210 are energized, an additional power increase occurs to power the additional bus wires.
[0026] Depending on the number of writes or reads occurring, PHY 119 incurs different power requirements: when an additional simultaneous read or write occurs, power is increased to power PHY 119 to transmit over bus wires 210, or to power bus wires 210 to receive data from the DRAM.
[0027] 3 is an example timing diagram illustrating read and write operations to a memory region of the example system 100 of FIG. 1, for example, at PHY 119. As shown in FIG. 3, the x-axis represents time passage and the y-axis represents power (current).
[0028] 3, during idle periods, power utilization in PHY 119 is at a low level. In this state, the power requirements in PHY 119 are low because the circuits within PHY 119 do not need to be powered up to write or read data.
[0029] When a write or read operation is performed, the power level increases to a higher level. As shown in Figure 3, a write operation utilizes less power than a read operation, but this is shown as an example only. For example, when memory controller 115 sends a command to PHY 119 to perform a write, the power requirements at PHY 119 increase.
[0030] This increase, as described above, allows the PHY 119 circuitry to be powered up to transmit data to DRAM bank B. Additionally, when memory controller 115 sends a command to PHY 119 to perform a read operation, the power in the PHY is increased to allow the PHY 119 circuitry directed to receiving data from DRAM bank B to be energized.
[0031] Additionally, the bus wires 210 are energized during either a read or write operation to transmit data along the wires, and as additional wires of the bus wires 210 are utilized to transmit or receive data, more power is generated.
[0032] During the idle periods shown in Figure 3, the current requirements (and therefore power requirements) are low because no circuitry is energized to allow PHY 119 to transmit or receive data. Again, during a write operation, the current is increased to a higher level to accommodate the circuitry requirements within PHY 119 to perform the write. As also shown in Figure 3, during a read operation, the current is increased to an even higher level than a read in PHY 119.
[0033] For example, if a write is to occur, the circuitry within PHY 119 for transmitting data over wire 210 is energized by increasing the current. If a read operation is to occur, the current is increased in the PHY 119 circuitry, energizing the circuitry to receive data from DRAM bank B over wire 210.
[0034] As can be seen in Figure 3, the increase in current going from the idle state to either the read or write operation state causes a significant increase in current draw from the power supply, and once the power requirement is removed after each write and read operation is completed, the current power drop is significantly reduced back to the idle level.
[0035] These large increases and decreases may prevent the power supply from adequately supplying power for all of the increases, especially if many simultaneous reads or writes are required on the bus wires 210. That is, each read and write on each of the bus wires 210 requires its own increase in power from an idle state to a read or write state.
[0036] Although each of the DRAM banks B may only see some of the transactions (e.g., reads or writes), PHY 119 is utilized for all of the transactions and is therefore required to increase or decrease the circuit power requirements for all transactions. Thus, by smoothing the current requirements for each scheduled read and write on all DRAM banks, the current increase or jump from an idle state to the level of power required to perform a read or write operation can help avoid a large drain on the power supply.
[0037] That is, by incrementally increasing the current level in PHY 119 from the idle power level to the read or write level power requirement prior to a read or write operation, distortion on the power supply is reduced. Additionally, by decreasing the current level in PHY 119 from either the read or write power level to the idle state power level, power distortion on the power supply is further managed.
[0038] To manage the power ramping before and after read and write operations, a simulated write (phantom write) may be performed. A simulated write simulates a write operation to increase power from an idle state to a level above the idle state. However, a simulated write may not increase the current level in PHY 119 to the full level utilized for either a read or write operation.
[0039] 2, memory controller 115 sends a simulated write (phantom write) to PHY 119, which is described in more detail below. However, again simply, a simulated write is an operation that partially powers up PHY 119 without allowing the output of PHY 119 to transmit to a memory bank (e.g., in memory 104).
[0040] Additionally, as described in more detail below, the simulated writes may increase the power level current to a level above the idle state power level but not the full current level utilized to perform the actual read or write operations. This can be accomplished by a technique of sub-channel throttling, where the actual increase is incrementally throttled during the increase in current level to bring the current level up to the read or write power level, or during the decrease in current level from the read or write operation to bring the current level down to the idle current level.
[0041] Figure 4 is an example timing diagram showing read and write operations to a memory region including a series of simulated writes. Figure 4 shows a similar diagram to the timing diagram of Figure 3. However, the timing diagram of Figure 4 shows the effect of sending simulated write operations to PHY 119.
[0042] For example, as shown in FIG. 4, a smoothed series of dashed lines 410 indicate intended triangles for incrementally increasing power levels by transmitting simulated writes (PW) 420.
[0043] A more detailed description of the method for performing simulated writes is briefly described below, but this is done by detecting where a write or read is about to occur and scheduling a series of simulated writes (420) to incrementally raise the power level prior to the read or write event.
[0044] These simulated writes are scheduled internally by memory controller 115 so as not to interfere with and not to introduce latency into the actual read and write transactions. As seen in Figure 4, each simulated write 420 causes a power increase by drawing current from an idle state that increases the power level from the supply power level to a level higher than the idle state power level.
[0045] 4, however, each simulated write 420 increases the power level by an additional step from the idle power state level to the read or write state power level. This incremental step up in power level with each simulated write 420 smoothes out the current draw from the power supply, effectively replicating triangle 410 and causing a smoother increase in power from the idle state power level to the read or write power level.
[0046] This incremental increase in power level may be accomplished by transmitting a predetermined value of data from the memory controller without enabling the transmitter output of PHY 119 (described below). That is, a first number of bytes may be used / transferred from memory controller 115 to PHY 119 to accomplish a first simulated write 420. A second, larger amount of bytes may be utilized to provide a greater increase in power in the second simulated write 420, etc., to bring the current level up to that of an actual read or write operation.
[0047] Once the event is complete, a series of simulated writes are scheduled to transition the power level to idle. That is, a simulated write 420 is introduced at a lower byte level than the actual read or write operation once the read or write event is completed. Each subsequent simulated write 420 is then introduced at a lower byte level than the previous simulated write 410 until the current level transitions back to the idle power level.
[0048] Thus, the power level increases in a fairly smooth increment along slope 410 as the byte size of each simulated write 420 is incrementally increased to draw more power, and then decreases along a descending slope 410 after a read or write event occurs with each simulated write 420 whose byte size is incrementally decreased. The smoothing of the power curve, which ramps up to each read or write operation and transitions back to an idle power state at the end of each read or write event, reduces strain on the power supply system.
[0049] 5 is an example block diagram of PHY layer 119 during a simulated write operation. PHY 119 includes a transmitter 121 that includes an output enable input that turns transmitter 121 on or off. PHY 119 also includes a termination circuit 122 and a receiver 123.
[0050] The transmitter 121, termination circuitry 122 and receiver 123 communicate with the DRAM memory 116 (e.g., a DDR bank) to send information during write operations and receive information during read operations. The communication can be via bus wires 210 shown in Figure 2. The PHY 119 receives data flow from the memory controller 115, including read operations, write operations and a simulated write operation that does not toggle the output enable.
[0051] As described above, during an actual read or write operation, the output enable of PHY 119 is enabled (e.g., set to 1) to allow PHY 119 to transmit and receive data from the DRAM bank. During a simulated write, data is transferred between memory controller 115 and PHY 119, causing some power up in PHY 119, but the output enable is not enabled (e.g., set to or remains at 0). Thus, the transmit and receive circuits are not powered and are not active for the actual read and write operations.
[0052] FIG. 6 is a flow diagram of an example method 600 for performing a simulated write operation, according to one embodiment.
[0053] An idle state is detected or determined (e.g., by memory controller 115) in step 610. To properly schedule a simulated write from the idle state, memory controller 115 then detects whether a read or write is to occur (step 620).
[0054] If a read or write occurs (step 620), a simulated write is performed to increase power before the read or write event (step 630). The simulated write instructs PHY 119 to increase power from an idle power state to a higher power level state. However, the simulated write does not enable the output enable bit of transmitter 121 of PHY 119. For example, the simulated write in step 630 does not toggle the output enable bit from 0 to 1 and does not change the state of the internal interface between PHY 119 and the DRAM.
[0055] If the power level of PHY 119 is not close to the level required for the read or write operation (step 640), memory controller 115 performs another simulated write operation (step 630). In the subsequent simulated write operation, the power level is increased above the power level of the first simulated write operation.
[0056] In this manner, an incremental increase in power along a smoothly increasing ramp is effectively generated, such as ramp 410 shown in Figure 4. That is, each subsequent simulated write in step 630 increases its power before a read or write operation, similar to the exemplary simulated write 420 shown in Figure 4.
[0057] That is, each subsequent simulated write 420 causes a higher increase in power level than the previous simulated write 420. This allows the power to be ramped up in stages towards the power level required to perform the actual read or write operation.
[0058] In step 640, if the power level is close to the power level required for a read or write operation and the read or write event is completed (step 650), the memory controller 115 performs a simulated write operation 660 to reduce the power level back to the idle power level until the power level is close to the idle power level (step 670).
[0059] That is, each subsequent simulated write 660 uses less power than the previous simulated write operation, similar to the example simulated write 420 shown in FIG. 4 after a read or write operation.
[0060] As mentioned above, the simulated write operation may be power modified via the memory controller instructing the PHY to modulate the number of data pins to perform the simulated write. For example, individual byte enables may be utilized to provide a granularity of eight data pins to perform the simulated write. That is, each step up or down may be an increase or decrease of eight data pins in the subsequent simulated write.
[0061] It should also be noted that criteria may be utilized to schedule the simulated write. For example, the simulated write may be scheduled between the load and load release (idle read) steps. The memory controller may know when the current goes from active to idle so that it can schedule the simulated write according to the active state (read vs. write). The memory controller may also schedule a predetermined period between the read and write operations for the simulated write.
[0062] Due to the DDR protocol, a gap between read and write operations is required. Thus, the memory controller may schedule a gap between read and write operations during which a simulated write may be performed. Because the simulated write is internal to the memory controller and PHY, communication of the simulated write is not communicated to the DRAM. Thus, the simulated write may be scheduled even when it would not otherwise be permitted by the DDR protocol.
[0063] Additionally, the simulated write operation may be power modified via the memory controller instructing the PHY to modulate the number of data pins to perform the simulated write. This may be implemented using individual byte enables, for example, allowing a granularity of 8 data pins.
[0064] The provided methods can be implemented in a general purpose computer, processor, or processor core. Suitable processors include, by way of example, a general purpose processor, a purpose processor, a conventional processor, a digital signal processor (DSP), multiple microprocessors, one or more microprocessors associated with a DSP core, a controller, a microcontroller, an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) circuit, any other type of integrated circuit (IC), and / or a state machine. Such a processor can be manufactured by configuring a manufacturing process using the results of processed hardware description language (HDL) instructions and other intermediate data (instructions that can be stored in a computer readable medium) including a netlist. The results of such processing can be a mask work that is used in a subsequent semiconductor manufacturing process to manufacture a processor that implements the features of the present disclosure. Furthermore, although the above-described methods and apparatus are described in the context of controlling and configuring memory physical links, the methods and apparatus may be utilized in any interconnect protocol in which link widths are negotiated.
[0065] The methods or flow charts provided herein may be implemented in a computer program, software, or firmware embodied in a non-transitory computer-readable storage medium for execution by a general-purpose computer or processor. Examples of non-transitory computer-readable storage media include read-only memory (ROM), random access memory (RAM), registers, cache memory, semiconductor memory devices, magnetic media such as internal hard disks and removable disks, magneto-optical media, and optical media such as CD-ROM disks and digital versatile disks (DVDs). For example, the methods described above may be implemented in processor 102 or on any other processor in computer system 100. In addition, the methods described above may be executed in a controller, processor, or circuitry in any component (e.g., a component of SOC 101).
Claims
1. 1. A method for performing simulated writes in a computer system, comprising: sending a simulated write operation to a physical layer circuit (PHY) to increase circuit power without enabling an output of the PHY until a memory operation is initiated; and in response to the memory operation being completed, sending a simulated write operation to the PHY to reduce circuit power. method.
2. the simulated write instructs the PHY to increase circuit power to a level above an idle state that is lower than a power level state of the memory operation; 2. The method of claim 1.
3. a first simulated write increases circuit power of the PHY to a first power level above the idle state; The method of claim 2.
4. a second simulated write increases a circuit power of the PHY to a second power level above the first power level; The method of claim 3.
5. performing the first simulated write prior to the second simulated write. The method of claim 4.
6. the first simulated write and the second simulated write are performed prior to a memory operation. The method of claim 5.
7. performing the first simulated write after the second simulated write. The method of claim 4.
8. the first simulated write and the second simulated write are performed upon completion of a memory operation. The method of claim 7.
9. the first simulated write includes a first number of data bytes and the second simulated write includes a second number of data bytes that is greater than the first number of data bytes. The method of claim 4.
10. scheduling the simulated writes based on a criterion.
2. The method of claim 1.
11. The criteria include one or more of a scheduled time between a load step and a load release of a memory operation, a scheduled time between a read operation and a write operation, or a predetermined period between a read operation or a write operation. The method of claim 10.
12. 1. A computer system for performing simulated writing, comprising: A physical layer circuit (PHY); a memory controller operatively coupled to the PHY and in communication with the PHY; The memory controller includes: in response to a scheduled memory operation, sending a simulated write operation to the PHY to increase circuit power without enabling an output of the PHY until the memory operation begins; responsive to the memory operation being completed, sending a simulated write operation to the PHY to reduce circuit power; 4. The method of claim 3, Computer system.
13. the simulated write instructs the PHY to increase circuit power to a level above an idle state that is lower than a power level state of the memory operation; 13. The computer system of claim 12.
14. a first simulated write increases circuit power of the PHY to a first power level above the idle state; 14. The computer system of claim 13.
15. A second simulated write increases a circuit power of the PHY to a second power level above the first power level.
15. The computer system of claim 14.
16. the memory controller sending the first simulated write before the second simulated write.
16. The computer system of claim 15.
17. the first simulated write and the second simulated write are transmitted before a memory operation.
17. The computer system of claim 16.
18. the memory controller sending the first simulated write after the second simulated write.
16. The computer system of claim 15.
19. the first simulated write and the second simulated write are transmitted upon completion of a memory operation.
20. The computer system of claim 18.
20. A memory controller operatively coupled to and in communication with a physical layer circuit (PHY), comprising: The memory controller includes: A processor is provided. The processor, in response to a scheduled memory operation, sending a simulated write operation to the PHY to increase circuit power without enabling an output of the PHY until the memory operation begins; responsive to the memory operation being completed, sending a simulated write operation to the PHY to reduce circuit power; [0023] Memory controller.
Citation Information
Patent Citations
Microcomputer, its memory contents changing system and memory contents changing method
JP2002024046A
Integrated Controller for Training of Memory Physical Layer Interface
JP2017523488A
Non-volatile memory (NVM) with adaptive write operations
US20140078829A1
Image processing device
US20210160425A1
Long-idle state system and method
US20210200298A1
