Heat dissipation analysis device, method, program, and material selection method

Abstract

This heat dissipation analysis device: simulates the maximum operating temperature of a high bandwidth memory by changing parameters of a simulation model of the high bandwidth memory that affect the heat dissipation of the high bandwidth memory; identifies, on the basis of the result of the simulation, a prediction formula that includes the parameters as variables and that predicts the maximum operating temperature of the high bandwidth memory; and uses the prediction formula to predict the maximum operating temperature of the high bandwidth memory when prescribed values are set for the parameters.

Description

Heat dissipation analysis apparatus, method, program, and material selection method

[0001] This disclosure relates to a heat dissipation analysis apparatus, a heat dissipation analysis method, a heat dissipation analysis program, and a material selection method.

[0002] Graphics Processing Units (GPUs) and Application Specific Integrated Circuits (ASICs) used in artificial intelligence (AI) and high-performance computing (HPC) incorporate high-bandwidth memory (HBMs) that enable high-speed data exchange. HBMs have a die-stacking structure of dynamic random access memory (DRAMs), and the DRAM dies are bonded together with non-conductive film (NCFs) (see Patent Document 1).

[0003] Japanese Patent Publication No. 2024-137897

[0004] While the number of memory die layers in HBMs has increased with each generation, the package thickness remains the same within the same generation, and newer HBMs have narrower gaps (thickness) between layers. Furthermore, because HBMs have high data transmission frequencies and generate significant heat, the number of dummy bumps has increased and the pitch (distance) between bumps has narrowed to improve heat dissipation. Given these narrow gaps and pitches, it is necessary to understand the physical properties of the NCF applied to the HBM, such as the requirement for high fluidity. In other words, it is desirable to apply an NCF with physical properties that match the heat dissipation characteristics of the HBM.

[0005] However, it was difficult to easily understand the heat dissipation characteristics of the HBM in accordance with the physical properties of the NCF and the structure of the HBM, such as dummy bumps for heat dissipation.

[0006] This disclosure has been made in view of the above points, and aims to provide a heat dissipation analysis device, method, and program that can easily analyze the heat dissipation characteristics of a broadband memory.

[0007] The heat dissipation analysis apparatus according to the first embodiment comprises: a simulation unit that simulates the maximum temperature during operation of the broadband memory by changing parameters that affect heat dissipation of the broadband memory in a simulation model of the broadband memory; a specification unit that identifies a prediction formula that includes the parameters as variables and predicts the maximum temperature during operation of the broadband memory based on the results of the simulation; and a prediction unit that uses the prediction formula to predict the maximum temperature during operation of the broadband memory when predetermined values ​​are set for the parameters.

[0008] The second aspect of the heat dissipation analysis method is a method in which a computer performs a process that includes changing a parameter in a simulation model of the broadband memory that affects the heat dissipation of the broadband memory to simulate the maximum temperature of the broadband memory during operation, identifying a prediction formula that includes the parameter as a variable and predicts the maximum temperature of the broadband memory during operation based on the results of the simulation, and using the prediction formula to predict the maximum temperature of the broadband memory during operation when a predetermined value is set for the parameter.

[0009] The heat dissipation analysis program according to the third embodiment is a program that causes a computer to perform the following processes: changing a parameter in a simulation model of a broadband memory that affects the heat dissipation of the broadband memory to simulate the maximum temperature of the broadband memory during operation; identifying a prediction formula that includes the parameter as a variable and predicts the maximum temperature of the broadband memory during operation based on the results of the simulation; and using the prediction formula to predict the maximum temperature of the broadband memory during operation when a predetermined value is set for the parameter.

[0010] According to the heat dissipation analysis apparatus, method, and program described herein, the heat dissipation characteristics of a broadband memory can be easily analyzed.

[0011] This is a block diagram showing the hardware configuration of the heat dissipation analysis device. This is a block diagram showing an example of the functional configuration of the heat dissipation analysis device. This is a schematic configuration diagram of the HBM. This is a plan view layout diagram of the HBM. This is a schematic configuration diagram of the HBM cross-section. This is a diagram showing an example of a simulation model of the HBM. This is a plan view layout diagram of the simulation model. This is a diagram showing an example of parameter values ​​for simulation. This is a diagram showing an example of simulation conditions. This is a diagram to explain the simplification of the simulation model. This is a diagram showing an example of a comparison between measured and predicted values ​​of the maximum temperature. This is a diagram showing an example of a prediction result. This is a diagram showing an example of a prediction result used for structure estimation. This is a diagram showing an example of a structure list. This is a diagram to explain structure estimation. This is a diagram to explain structure estimation. This is a diagram to explain structure estimation. This is a diagram showing an example of a prediction result used for structure estimation. This is a diagram to explain structure estimation. This is a flowchart showing an example of the heat dissipation analysis process.

[0012] An example of this embodiment will be described below with reference to the drawings. Note that identical or equivalent components and parts are given the same reference numerals in each drawing. Also, the dimensions and proportions in the drawings are exaggerated for illustrative purposes and may differ from actual proportions.

[0013] Figure 1 is a block diagram showing the hardware configuration of the heat dissipation analysis device 10 according to this embodiment. As shown in Figure 1, the heat dissipation analysis device 10 has a CPU (Central Processing Unit) 12, memory 14, storage device 16, input device 18, output device 20, storage medium reader 22, and communication I / F (Interface) 24. Each component is connected to the others so as to be able to communicate with each other via a bus 26.

[0014] The storage device 16 stores a heat dissipation analysis program for executing the heat dissipation analysis process described later. The CPU 12 is a central processing unit that executes various programs and controls each component. Specifically, the CPU 12 reads the program from the storage device 16 and executes the program using memory 14 as a working area. The CPU 12 controls each component and performs various calculations according to the program stored in the storage device 16.

[0015] Memory 14 is composed of RAM (Random Access Memory) and temporarily stores programs and data as a working area. Storage device 16 is composed of ROM (Read Only Memory), HDD (Hard Disk Drive), SSD (Solid State Drive), etc., and stores various programs including the operating system and various data.

[0016] The input device 18 is a device for performing various types of input, such as a keyboard or mouse. The output device 20 is a device for outputting various types of information, such as a display or printer. By using a touch panel display as the output device 20, it may also function as the input device 18.

[0017] The storage medium reader 22 reads data stored on various storage media such as CD (Compact Disc)-ROM, DVD (Digital Versatile Disc)-ROM, Blu-ray disc, and USB (Universal Serial Bus) memory, and writes data to the storage media. The communication I / F 24 is an interface for communication with other devices, and standards such as Ethernet®, FDDI, and Wi-Fi® are used.

[0018] Next, the functional configuration of the heat dissipation analysis device 10 according to this embodiment will be described. Figure 2 is a block diagram showing an example of the functional configuration of the heat dissipation analysis device 10. As shown in Figure 2, the heat dissipation analysis device 10 includes a simulation unit 32, a specific unit 34, a prediction unit 36, and an estimation unit 38 as its functional configuration. Each functional configuration is realized by the CPU 12 reading a heat dissipation analysis program stored in the storage device 16, expanding it into the memory 14, and executing it. In addition, a simulation model 40, a prediction formula DB 42, and a structure list 44 are stored in a predetermined storage area of ​​the heat dissipation analysis device 10.

[0019] Here, referring to Figure 3, the configuration of the package 100, which includes the HBM that forms the basis of the simulation model 40, will be described. In package 100, a logic die 106 and an HBM 108 are mounted on the package substrate 102 via an interposer 104. The HBM 108 includes multiple layers of stacked DRAM dies (hereinafter also referred to as "memory dies") 110, and the memory dies 110 and the HBM 108 and logic die 106 are electrically connected via TSVs (Through Silicon Vias) 112. In addition, in package 100, an SoC chip 114 is mounted on the package substrate 102 via the interposer 104. The logic die 106 and the SoC chip 114 are electrically connected via their respective PHY (physical layer) areas 116 and 118.

[0020] Figure 4 shows a plan view layout of the HBM108. As shown in Figure 4, each memory die 110 of the HBM108 includes an area where I / O bumps for data input / output are located and an area where dummy bumps for heat dissipation are located.

[0021] Figure 5 shows a cross-sectional view of the HBM108 along line A-A shown in Figure 4. As shown in Figure 5, the memory die 110 of the HBM108 includes the bottom memory die 110B, the top memory die 110T, and the middle memory die 110M between them. The bottom memory die 110B is larger in size than the middle memory die 110M and the top memory die 110T, and the difference is filled with a sealing material 120. The memory dies 110 are bonded together with NCF 124, which includes bumps or pads 122.

[0022] Figure 6 shows an example of the simulation model 40. The example in Figure 6 models the right half of the HBM 108. Figure 7 shows a plan view layout of the simulation model 40. In this embodiment, of the surface of the memory die 110 of the HBM 108, the part corresponding to the PHY area 116, which is the part where data is exchanged with the logic die 106, is set as the heat generation area 46 in the simulation model 40, as shown in Figure 7.

[0023] The simulation unit 32 simulates the maximum operating temperature of the HBM 108 by changing the parameters in the simulation model 40 that affect the heat dissipation of the HBM 108. In this embodiment, the maximum temperature is the highest temperature ultimately reached at each position of the HBM when the HBM 108 is always generating heat at its rated capacity.

[0024] Furthermore, in this embodiment, the parameters that affect the heat dissipation of the HBM108 are the sum of the gaps between the memory dies 110 (hereinafter also referred to as "die gaps"), the total area when dummy bumps for heat dissipation are projected onto the memory die plane (hereinafter also referred to as "sum of bump projected areas"), and the thermal conductivity of the NCF. The simulation unit 32 acquires parameter values, for example, as shown in Figure 8. In the example of Figure 8, the sum of the die gaps is set by the product of the number of layers of the memory die 110 and the die gaps. The sum of the bump projected areas is set based on the diameter of the bumps and the pitch between the bumps. In this case, the size of the memory die 110 is set in advance in the simulation model 40, and the sum of the bump projected areas is set by the product of the number of dummy bumps obtained from the size, the diameter and pitch of the bumps, and the area of ​​each bump obtained from the diameter of the bumps. With this setting method, the setting values ​​that exist as the actual configuration of the HBM108 can be set as parameter values.

[0025] The parameters are not limited to the three mentioned above, and may also include factors that affect the heat dissipation characteristics of the HBM108, such as the operating conditions of the HBM108 and the cooling conditions of the equipment to which the package 100 containing the HBM108 is applied.

[0026] Furthermore, the simulation unit 32 acquires simulation conditions, for example, as shown in Figure 9. In the example in Figure 9, the heat generation condition is specified as 15W for the heat generation area shown in Figure 7. As a heat dissipation condition, it is specified that all the heat generated by the logic die 106 is dissipated from the top surface, and the heat transfer coefficient is specified as the heat dissipation condition. As a heat dissipation condition, it is specified that the sides and bottom of the HBM 108 are insulated. In addition, the physical properties of each component constituting the HBM 108 are specified as simulation conditions. The physical properties of the components include at least specific heat, thermal conductivity, and other physical properties related to heat dissipation, and may also include other physical properties such as density.

[0027] The simulation unit 32 uses the simulation model 40 to analyze the heat dissipation characteristics during HBM108 operation for each pattern, which is a combination of the acquired parameter values ​​according to the acquired simulation conditions, for example by three-dimensional thermal analysis, and derives the maximum temperature for each pattern.

[0028] In this case, the simulation unit 32 simplifies the simulation model 40 to reduce the number of bumps while maintaining the sum of the bump projection areas specified as parameters. For example, as shown in Figure 10, the simulation unit 32 reduces the number of bumps by changing the actual bump shape, which is circular in plan view, to a rectangle, and by merging multiple bumps into one. In this way, by simplifying the shape and number of bumps while maintaining the sum of the bump projection areas, which have a large impact on heat dissipation characteristics, it is possible to reduce the computational load while suppressing a decrease in the accuracy of the simulation.

[0029] The identification unit 34 identifies a prediction formula that includes the above-mentioned parameters as variables and predicts the maximum temperature during operation of the HBM 108, based on the simulation results from the simulation unit 32.

[0030] Specifically, the identification unit 34 performs statistical processing such as multiple regression analysis using the maximum temperature for each parameter pattern, which is the simulation result, to identify the coefficients included in the prediction formula for the maximum temperature T shown in equation (1) below. T = f(x) + f(y) + f(z) (1)

[0031] In Equation (1), f(x), f(y), and f(z) are functions of variables corresponding to the sum of die gaps, the sum of bump projected areas, and the thermal conductivity of NCF, respectively. More specifically, the prediction equation may be as shown in Equation (2) below.

[0032] T = α １ + α ２ × x + α ３ × y + α ４ × z + (x - α ５ ) × ((y - α ６ ) × α ７ ) + (y - α ６ ) × ((y - α ６ ) × α ８ ) + (x - α ５ ) × ((z - α ９ ) × α １０ ) + (y - α ６ ) × ((z - α ９ ) × α １０ ) (2)

[0033] In Equation (2), x is the variable corresponding to the sum of die gaps, y is the variable corresponding to the sum of bump projected areas, z is the variable corresponding to the thermal conductivity of NCF, and α １ ~α １０ are coefficients. That is, the specifying unit 34 substitutes the values of the sum of die gaps, the sum of bump projected areas, and the thermal conductivity of NCF in each pattern of the simulation results, and the maximum temperature T derived for each pattern into Equation (2), and performs statistical processing such as multiple regression analysis to specify the coefficients α １ ~α １０ . The specifying unit 34 stores the prediction equation with the specified coefficients in the prediction equation DB 42.

[0034] Fig. 11 shows an example of the comparison between the calculated value obtained by the simulation of the maximum temperature and the predicted value obtained by the prediction formula. In this example, the adjusted R-squared value between the measured value and the predicted value is 0.9 or more, the PRESS R-squared value is 0.8 or more, the P-value is less than 0.05, and the VIF is 1 or more and less than 5. It can be determined that a good prediction formula has been identified. Further, by the statistical analysis of the prediction formula, when the influence degree for each parameter was determined, it was found that the sum of the die gaps and the sum of the bump projected areas, that is, the configuration of HBM108 has a higher influence degree on the maximum temperature than the thermal conductivity of NCF.

[0035] The prediction unit 36 predicts the maximum temperature during the operation of HBM108 when a predetermined value is set for the parameter, using the prediction formula specified by the specifying unit 34. Specifically, the prediction unit 36 receives, as prediction conditions, the values of the sum of the die gaps, the sum of the bump projected areas, and the thermal conductivity of NCF, respectively. Further, the prediction unit 36 reads out the prediction formula from the prediction formula DB42, substitutes the received prediction conditions into the prediction formula, and predicts the maximum temperature. The prediction unit 36 outputs the predicted maximum temperature as a prediction result.

[0036] Further, when the prediction unit 36 receives a prediction condition including the maximum temperature and the values of two parameters, the prediction unit 36 may substitute the received prediction condition into the prediction formula, predict the values of the remaining parameters, and output them as a prediction result.

[0037] Further, the prediction unit 36 may output, as a prediction result, a graph showing the relationship between each parameter and the maximum temperature. An example of the prediction result in this case is shown in Figs. 12 and 13. Fig. 12 shows the relationship between the sum of the die gaps and the sum of the bump projected areas and the maximum temperature when the thermal conductivity of NCF is fixed at 0.43 W / m·K, and Fig. 13 shows the relationship between the sum of the die gaps and the sum of the bump projected areas and the maximum temperature when the thermal conductivity of NCF is fixed at 2.15 W / m·K.

[0038] The estimation unit 38 estimates the configuration of the HBM 108 based on the prediction result by the prediction unit 36. Specifically, the estimation unit 38 estimates the number of stacked memory dies 110, bump diameter, bump pitch, etc. based on the values of the sum of die gaps and the sum of bump projected areas among the predicted or specified parameters. More specifically, the estimation unit 38 refers to a structure list 44 in which the values of each item are stored as the actual structure of the HBM 108 as shown in FIG. 14, and estimates a structure that satisfies the values of the predicted or specified sum of die gaps and the sum of bump projected areas. The estimation unit 38 outputs the estimation result.

[0039] A more specific example of the processing of the estimation unit 38 will be described. For example, as shown in FIG. 15, consider setting a reference value for the maximum temperature used in the structure estimation of the HBM 108 based on the NCFs of three cases. In this case, for example, the maximum temperature is predicted from a prediction formula using the values shown in FIG. 16 and predetermined assumption conditions. The assumption conditions here are, for example, values such as the number of I / O bumps, the diameter of I / O bumps, the diameter of pads, the diameter of dummy bumps, the pitch of dummy bumps, the number of stacked layers, etc. In the example of FIG. 16, the die gap is set to a value obtained by considering a crushing allowance of 0.2 μm from the thickness of the NCF. Also, for case 1 where there is no upper limit value for the thermal conductivity of the NCF, the thermal conductivity of the NCF is set to 1.5 W / m·K.

[0040] Looking at this prediction result, assume that the user determines that case 1 is NG because the maximum temperature is high despite setting a high thermal conductivity, and determines that cases 2 and 3 are OK. From this, assume that the reference value for the maximum temperature is set to 95°C based on the maximum temperatures of cases 2 and 3.

[0041] Let's assume that the following conditions for estimating the structure were set as requirements for the NCF: a thickness of 5.2 μm (die gap of 5 μm) and a thermal conductivity of 1.5 W / m·K. The estimation unit 38 identifies the number of layers corresponding to the estimation condition "die gap of 5 μm" from the structure list 44. Here, 16 layers (hereinafter referred to as "16H"; total die gap: 80 μm) and 12 layers (hereinafter referred to as "12H"; total die gap: 60 μm) are identified. The estimation unit 38 also identifies the structure of the dummy bump corresponding to 16H or 12H from the structure list 44. Here, let's assume that Case 1 and Case 2 shown in Figure 17 are identified.

[0042] Figure 18 shows the predicted relationship between the maximum temperature and the sum of the die gaps and the sum of the bump projection areas in this case. From these prediction results, the estimation unit 38 identifies cases in which the maximum temperature is less than the reference value of 95°C, as shown in Figure 19. Based on case 12H of case 1, which has the maximum temperature closest to the reference value among the OK cases, the estimation unit 38 estimates the structure of the HBM 108 to have 12 layers, a dummy bump diameter of 20 μm, and a pitch of 35 μm.

[0043] In this way, by estimating the structure of HBM108 from the predicted maximum temperature, it is possible to understand the structure of HBM108 even when only the required values ​​for NCF are specified. This allows for the selection and development of more appropriate NCF materials based on the structure of HBM108. Furthermore, as mentioned above, the structure of the dummy bumps has a stronger influence on the heat dissipation characteristics of HBM108 than the thermal conductivity of the NCF, so it may be appropriate to change to a structure with a higher heat dissipation effect than the estimated structure. The estimation results can serve as information for consideration in such cases.

[0044] Next, the operation of the heat dissipation analysis device 10 according to this embodiment will be described.

[0045] Figure 20 is a flowchart showing the flow of the heat dissipation analysis process executed by the CPU 12 of the heat dissipation analysis device 10. The CPU 12 reads the heat dissipation analysis program from the storage device 16, loads it into the memory 14, and executes it. As a result, the CPU 12 functions as one of the various functional components of the heat dissipation analysis device 10, and the heat dissipation analysis process shown in Figure 20 is executed. Note that the heat dissipation analysis process is an example of the heat dissipation analysis method of this disclosure.

[0046] In step S10, the simulation unit 32 acquires simulation conditions and simulation values ​​for each parameter. Next, in step S12, the simulation unit 32 uses the simulation model 40 to analyze the heat dissipation characteristics during HBM108 operation for each pattern of combinations of the acquired parameter values ​​according to the acquired simulation conditions, for example by three-dimensional thermal analysis, and derives the maximum temperature for each pattern.

[0047] Next, in step S14, the identification unit 34 performs statistical processing such as multiple regression analysis using the maximum temperature for each parameter pattern, which is the simulation result, to identify the coefficients included in the prediction formula for the maximum temperature T shown in equation (1). Then, the identification unit 34 stores the prediction formula with the identified coefficients in the prediction formula DB 42.

[0048] Next, in step S16, the prediction unit 36 ​​determines whether it has received prediction conditions that include three values ​​from among the maximum temperature, the sum of the die gaps, the sum of the bump projected areas, and the thermal conductivity of the NCF. If it has received the prediction conditions, it proceeds to step S18. If it has not received the prediction conditions after a predetermined time has elapsed, the heat dissipation analysis process ends.

[0049] In step S18, the prediction unit 36 ​​reads a prediction formula from the prediction formula DB 42, substitutes the received prediction conditions into the prediction formula, predicts the value of the maximum temperature or the remaining one parameter, and outputs the prediction result.

[0050] Next, in step S20, the estimation unit 38 determines whether it has received the estimation conditions, including the required values ​​for the thermal conductivity and maximum temperature of the NCF. If the estimation conditions have been received, the process proceeds to step S22. If the estimation conditions have not been received after a predetermined time has elapsed, the heat dissipation analysis process is terminated.

[0051] In step S22, the estimation unit 38 estimates the structure of the HBM 108 based on the estimation conditions, prediction results, and structure list 44, outputs the estimation results, and the heat dissipation analysis process ends.

[0052] As described above, the heat dissipation analysis device according to this embodiment simulates the maximum operating temperature of the HBM by changing the parameters that affect the heat dissipation of the HBM in the HBM simulation model. Furthermore, based on the simulation results, the heat dissipation analysis device identifies a prediction formula that includes the parameters as variables and predicts the maximum operating temperature of the HBM.Then, the heat dissipation analysis device uses the prediction formula to predict the maximum operating temperature of the HBM when predetermined values ​​are set for the parameters.This makes it possible to easily analyze the heat dissipation characteristics of the HBM.

[0053] <Modification> The material of the component to be used in the HBM 108 may be selected based on the prediction results predicted by the prediction unit 36 ​​of the above embodiment. For example, the prediction unit 36 ​​accepts the maximum temperature, which is a required value, and the sum of the die gap and the sum of the bump projection area based on the assumed configuration of the HBM 108 as prediction conditions. The prediction unit 36 ​​then substitutes these prediction conditions into the prediction formula to predict the thermal conductivity of the NCF that satisfies these prediction conditions.

[0054] Then, multiple NCF products are registered in the database in advance, along with their thermal conductivity, and the NCF product with the thermal conductivity closest to the predicted thermal conductivity is presented. The presented NCF product may be adopted as the NCF to be applied to HBM108, or the information on the presented NCF product may be used as reference information for developing an NCF product that satisfies the predicted thermal conductivity.

[0055] Furthermore, the heat dissipation analysis process, which the CPU reads and executes in the above embodiment, may also be executed by various processors other than the CPU. Examples of such processors include PLDs (Programmable Logic Devices) such as FPGAs (Field-Programmable Gate Arrays) whose circuit configuration can be changed after manufacturing, and dedicated electrical circuits that are processors with circuit configurations specifically designed to execute particular processes, such as ASICs. The heat dissipation analysis process may also be executed by one of these various processors, or by a combination of two or more processors of the same or different types (for example, multiple FPGAs, and a combination of a CPU and an FPGA). A GPU may also be used as a processor. More specifically, the hardware structure of these various processors is an electrical circuit that combines circuit elements such as semiconductor elements.

[0056] Furthermore, although the above embodiment describes a configuration in which the heat dissipation analysis program is pre-stored (installed) in a storage device, the invention is not limited to this configuration. The program may be provided in a form stored on a storage medium such as a CD-ROM, DVD-ROM, or USB memory. Alternatively, the program may be provided in a form that can be downloaded from an external device via a network.

[0057] The following additional information is disclosed regarding the above embodiment.

[0058] (Note 1) A heat dissipation analysis apparatus comprising: a simulation unit that simulates the maximum operating temperature of a broadband memory by changing parameters that affect the heat dissipation of the broadband memory in a simulation model of the broadband memory; a specification unit that identifies a prediction formula that includes the parameters as variables and predicts the maximum operating temperature of the broadband memory based on the results of the simulation; and a prediction unit that uses the prediction formula to predict the maximum operating temperature of the broadband memory when predetermined values ​​are set for the parameters.

[0059] (Note 2) The heat dissipation analysis apparatus according to Note 1, wherein the parameters include at least the sum of the distances between multiple stacked memory dies in the broadband memory, the total area of ​​the dummy bumps for heat dissipation, and the thermal conductivity of the insulating adhesive film used to bond the memory dies together.

[0060] (Appendix 3) The heat dissipation analysis apparatus according to Appendix 2, wherein the parameters further include at least one of the operating conditions of the broadband memory and the cooling conditions of the equipment to which the broadband memory is applied.

[0061] (Note 4) The heat dissipation analysis apparatus according to any one of Notes 1 to 3, wherein the simulation unit simulates the temperature that would ultimately be reached if the broadband memory were always generating heat at its rated temperature, with the maximum temperature being used for the simulation.

[0062] (Note 5) A heat dissipation analysis apparatus according to any one of Note 1 to Note 4, wherein in the simulation model, the portion where the broadband memory is connected to the logic die is set as the heat-generating area of ​​the broadband memory.

[0063] (Note 6) The heat dissipation analysis apparatus according to Note 5, wherein the simulation unit simulates the maximum temperature assuming that the sides and bottom of the broadband memory are insulated and that the heat generated in the heat-generating area is dissipated from the top surface of the broadband memory.

[0064] (Appendix 7) The heat dissipation analysis apparatus according to Appendix 2 or Appendix 3, wherein the simulation model is a model in which the number of dummy bumps is reduced while maintaining the total area of ​​the dummy bumps.

[0065] (Note 8) The heat dissipation analysis apparatus according to any one of Note 1 to Note 7, wherein the simulation unit sets the physical properties of each component constituting the broadband memory in the simulation model.

[0066] (Note 9) The heat dissipation analysis apparatus described in Note 8, which includes at least specific heat and thermal conductivity as the physical properties related to heat dissipation.

[0067] (Note 10) The physical properties related to heat dissipation are further determined by the heat dissipation analysis apparatus described in Note 9, including density.

[0068] (Appendix 11) A heat dissipation analysis apparatus according to any one of Appendix 1 to Appendix 10, which includes an estimation unit that estimates the structure of the broadband memory based on the prediction results from the prediction unit.

[0069] (Appendix 12) A material selection method for selecting the material of the component to be used in the broadband memory based on the prediction results from the heat dissipation analysis device described in any one of Appendix 1 to Appendix 11.

[0070] 10 Heat dissipation analysis device 12 CPU 14 Memory 16 Storage device 18 Input device 20 Output device 22 Storage medium reader 24 Communication I / F 26 Bus 32 Simulation unit 34 Identification unit 36 ​​Prediction unit 38 Estimation unit 40 Simulation model 42 Prediction formula DB 44 Structure list 46 Heat-generating area 100 Package 102 Package substrate 104 Interposer 106 Logic die 108 HBM 110 Memory die 116 PYS area 120 Encapsulation material 122 Bump or pad

Claims

1. A heat dissipation analysis apparatus comprising: a simulation unit that simulates the maximum operating temperature of a broadband memory by changing parameters that affect the heat dissipation of the broadband memory in a simulation model of the broadband memory; a specification unit that identifies a prediction formula that includes the parameters as variables and predicts the maximum operating temperature of the broadband memory based on the results of the simulation; and a prediction unit that uses the prediction formula to predict the maximum operating temperature of the broadband memory when predetermined values ​​are set for the parameters.

2. The heat dissipation analysis apparatus according to claim 1, wherein the parameters include at least the sum of the distances between multiple stacked memory dies in the broadband memory, the total area of ​​the dummy bumps for heat dissipation, and the thermal conductivity of the insulating adhesive film used to bond the memory dies together.

3. The heat dissipation analysis apparatus according to claim 2, wherein the parameters further include at least one of the operating conditions of the broadband memory and the cooling conditions of the equipment to which the broadband memory is applied.

4. The heat dissipation analysis apparatus according to any one of claims 1 to 3, wherein the simulation unit simulates the temperature that would ultimately be reached if the broadband memory were to always generate heat at its rated temperature, with the maximum temperature being used for the simulation.

5. The heat dissipation analysis apparatus according to any one of claims 1 to 3, wherein in the simulation model, the portion where the broadband memory is connected to the logic die is set as the heat-generating area of ​​the broadband memory.

6. The heat dissipation analysis apparatus according to claim 5, wherein the simulation unit simulates the maximum temperature assuming that the sides and bottom of the broadband memory are insulated and that the heat generated in the heat-generating area is dissipated from the top surface of the broadband memory.

7. The heat dissipation analysis apparatus according to claim 2 or 3, wherein the simulation model is a model in which the number of dummy bumps is reduced while maintaining the total area of ​​the dummy bumps.

8. The heat dissipation analysis apparatus according to any one of claims 1 to 3, wherein the simulation unit sets the physical properties of each component constituting the broadband memory in the simulation model.

9. The heat dissipation analysis apparatus according to claim 8, wherein the physical properties related to heat dissipation include at least specific heat and thermal conductivity.

10. The heat dissipation analysis apparatus according to claim 9, wherein the physical properties related to heat dissipation further include density.

11. A heat dissipation analysis apparatus according to any one of claims 1 to 3, comprising an estimation unit that estimates the structure of the broadband memory based on the prediction results from the prediction unit.

12. A material selection method for selecting a material for a component to be used in the broadband memory based on the prediction results obtained by the heat dissipation analysis device described in any one of claims 1 to 3.

13. A heat dissipation analysis method in which a computer performs the following steps:

1. In a simulation model of a broadband memory, 2. In a simulation model of the broadband memory, 3. In a simulation model of the broadband memory, 4. In a simulation model of the broadband memory, 5. In a simulation model of the broadband memory, 6. In a simulation model of the broadband memory, 17. In a simulation model of the broadband memory, 18. In a simulation model of the broadband memory, 18. In a simulation model of the broadband memory, 18. In a simulation model of the broadband memory, 199. In a simulation model of the broadband memory, 18. In a simulation model of the broadband memory, 19. In a simulation model of the broadband memory, 18. In a simulation model of the broadband memory, 19. In a 14. A heat dissipation analysis program for causing a computer to perform the following processes: 1) Simulating the maximum operating temperature of a broadband memory by changing a parameter in a simulation model of the broadband memory that affects the heat dissipation of the broadband memory; 2) Based on the results of the simulation, identifying a prediction formula that includes the parameter as a variable and predicts the maximum operating temperature of the broadband memory; and 3) Using the prediction formula, predicting the maximum operating temperature of the broadband memory when a predetermined value is set for the parameter.

Images