Selecting Storage for ECUs: Differences between eMMC and UFS and Adoption Guidelines

The role of in-vehicle storage

In an in-vehicle system, storage is not simply a component for storing data. It is deeply involved in the behavior and reliability of the entire system, from the high-speed reading of POWER SUPPLIES data that enables IVI and various ECUs to operate immediately after power-on, to the recording of drive recorders, event logs generated while driving, and sensor data, and even software updates via OTA.

In particular, for automotive applications, the boot time after starting the engine is directly linked to the user experience, and high rewrite endurance (TBW*) is required to be able to withstand frequent and continuous writing, such as for logs related to drive recorders and ADAS (Advanced Driver-Assistance Systems).In addition, OTA updates require the safe writing of large volumes of data in a short time, so write performance and data protection functions in the event of POWER SUPPLIES outage are also important requirements when selecting storage.

Furthermore, automotive equipment requires stable operation over a wide temperature range, such as-40 to 85°C, and is expected to be supplied and used for a long period of 10 years or more, so a reliability design that differs from that of consumer devices is essential.

As such, in-vehicle storage requires consideration not only of "capacity" and "price," but also of multiple aspects such as startup characteristics, Access patterns, write durability, reliability, etc. To meet these requirements, it is important to understand the design concepts and differences in standards of the storage itself and then select a method that is suitable for the application.

In the next chapter, we will summarize the characteristics of JEDEC, the standard for in-vehicle embedded storage, and the widely adopted eMMC and UFS.

• TBW (Terabytes Written): An index showing the cumulative amount of data that can be written to a storage device before it reaches the end of its lifespan. It is primarily defined based on the number of program/erase cycles (P/E) performed on NAND flash memory, and is used as a guideline for evaluating write durability and long-term reliability.

What is JEDEC? Essential common rules for selecting in-vehicle storage

JEDEC (Joint Electron Device Engineering Council) is an industry organization that develops international standards, primarily for semiconductor memory. Embedded storage devices such as eMMC and UFS are developed based on specifications set by JEDEC.

One of the reasons why JEDEC standards are important for automotive applications is compatibility between manufacturers and long-term supply. Automotive ECUs often have a product lifecycle of 10 years or more, so having a standard that is not dependent on the proprietary specifications of a specific manufacturer leads to stable procurement and design continuity.

Furthermore, the JEDEC standard goes beyond simple interface specifications to also define basic behavior related to initialization, error handling, and reliability. This makes it possible to proceed with design based on a common understanding of important automotive elements such as data protection during POWER SUPPLIES outages and behavior during abnormal conditions. For ECU developers, being able to treat basic storage operations as prerequisites is a major advantage.

In this way, JEDEC serves as a "common rule" for selecting in-vehicle storage. Among the JEDEC standards, eMMC and UFS are currently the most widely adopted embedded storage.

In the next chapter, we will look at the design concepts behind each method and how they affect the requirements for automotive ECUs.

eMMC – A compelling balance of simplicity and cost

eMMC (embedded Multi Media Card) is an embedded storage device that integrates NAND flash and a controller into a single package. Its major advantages are that host control is relatively simple, and it is easy to reduce the burden of connecting to an SoC and implementing software. For these reasons, it has been widely adopted for many years, even in automotive applications.

eMMC Block Diagram

The communication method is half-duplex with a parallel interface, and is designed to allow only transmission or reception at a time. For this reason, parallel processing performance is limited compared to UFS, which will be discussed later, but it provides sufficient performance for ECUs with specific applications. The JEDEC eMMC 5.x generation introduced the HS400 mode*, which enables high-speed transfers, supporting transfer speeds of up to 400MB/s.

Additionally, the Command Queuing (CQ)* function allows for limited processing of multiple commands, and random Access performance has also been improved compared to previous generations.In terms of security, it has a tamper-resistant RPMB* area, making it suitable for storing secure boot information and authentication data.In addition, it has a dedicated boot partition that is easy to use when starting up the ECU, which is another factor that makes it easy to use in automotive design.

Due to these characteristics, eMMC is widely used in body ECUs, whose main purposes are saving settings and relatively lightweight log recording, and in telematics units (TCUs), which primarily have communication functions. eMMC remains an effective option for systems that prioritize cost, ease of design, and stable operation over high performance.

• HS400 mode: A high-speed transfer mode introduced in eMMC 5.x. Supports transfer speeds of up to 400MB/s using the DDR method, contributing to shorter data read times when the ECU starts up.
• Command Queuing (CQ): A function that manages multiple Access requests as a queue to improve processing efficiency. With eMMC, it improves random Access performance within the range of half-duplex communication.
• RPMB (Replay Protected Memory Block): A secure dedicated area with an authentication mechanism that prevents tampering and replay attacks and is used to store data related to secure boot.

UFS - Next-generation system with fast boot speeds and excellent parallel processing

UFS (Universal Flash Storage) is an embedded storage standard designed for high performance and parallel processing capabilities. It has been widely adopted for smartphones, but in recent years it has been increasingly adopted in areas where high-speed response and large-volume data processing are required, such as in-vehicle IVI and ADAS-related ECUs.

UFS Block Diagram

A major feature of UFS is its use of serial full-duplex communication using MIPI M-PHY* and UniPro*. This allows for simultaneous read and write processing, achieving high parallelism. Compared to eMMC, which uses half-duplex communication, the performance difference becomes more noticeable in situations where there is a lot of random Access or multiple tasks occur simultaneously.

​Host Performance Booster (HPB)* is a feature that contributes to improving random Access performance. HPB utilizes the host's DYNAMIC RAMS to cache frequently Access information, enabling it to be read quickly, thereby speeding up application startup and map data reading. It also includes Write Booster*, which temporarily speeds up the writing of large amounts of data, effectively shortening the write time during OTA updates.

In terms of transfer performance, UFS has improved significantly with each generation, with the UFS 4.x generation theoretically supporting transfer speeds of 4GB/s. This reduces the likelihood of storage becoming a bottleneck for the entire system, allowing for more room in the design of increasingly sophisticated automotive ECUs. Additionally, power state management and data integrity protection mechanisms have been strengthened, achieving both performance and reliability.

• MIPI M-PHY: A high-speed serial physical layer interface used in UFS. It achieves high bandwidth and low power consumption with a small number of pins, making it suitable for highly reliable systems including automotive applications.
• UniPro (Unified Protocol): A transfer protocol that runs on MIPI M-PHY. It features full-duplex communication and high parallelism, supporting high-speed, highly efficient data transfer in UFS.
• HPB (Host Performance Booster): A UFS function that uses host-side DYNAMIC RAMS to cache address translation information and improve random read performance. This is effective in speeding up app launches and map data reads.
• Write Booster: A UFS function that uses a portion of the drive as a high-speed buffer during writing to reduce write latency. This helps reduce the time it takes to write large amounts of data, such as during OTA updates.

eMMC and UFS – Impact on ECU performance requirements

We've looked at the features of eMMC and UFS so far, but how do the differences between the two affect the actual behavior of the ECU? In automotive systems, storage performance is not just measured numerically; it is also reflected in boot time, update time, and user experience. The features of eMMC and UFS are summarized in the table below.

• Garbage collection (GC): An internal process that organizes invalid data and secures reusable space. Since NAND flash cannot be overwritten, GC is performed automatically to maintain write performance and stable operation.
• Wear leveling: A control that distributes writing and erasure evenly across the entire NAND flash memory to prevent concentration in specific areas. This evens out cell wear and extends the overall lifespan (endurance) of the storage device.

Below, we will explain the differences between the two methods based on typical ECU operating scenarios.

Startup time

Because eMMC uses half-duplex communication, the numerous random read operations that occur during startup tend to serialize Access, resulting in a certain amount of waiting time during startup. On the other hand, UFS has full-duplex communication and high parallel processing capabilities, allowing it to efficiently process multiple read requests. As a result, in systems such as IVI that require screen display and operation immediately after startup, UFS tends to achieve a perceived faster startup.

Flashing process and OTA update

While eMMC can handle medium-sized continuous writes without any issues, writing large amounts of data in one go, such as with OTA updates, tends to take longer. UFS's Write Booster feature makes it easier to reduce write delays, leading to shorter update times. For ECUs with frequent OTA updates, this difference translates into shorter system downtime and improved user convenience.

Random Access Performance

In IVI, where map data and applications are frequently read, random read performance is directly linked to operational response. While eMMC offers some improvement through command queuing, UFS makes it possible to utilize cache via the HPB, and the difference is particularly noticeable in systems with a lot of random Access.

Reliability and Data Protection

Both have basic reliability features based on JEDEC standards, but the newer generation of UFS provides stronger data integrity protection and error handling when POWER SUPPLIES is lost. Meanwhile, eMMC also utilizes RPMB and a boot partition to ensure sufficient reliability for secure booting and for saving configuration information.

Cost and System Design

eMMC has advantages in terms of component cost and design difficulty, making it a cost-effective option for ECUs with clear applications and limited performance requirements. In contrast, UFS is more expensive and more difficult to design, but it improves the quality of experience through its high speed and parallelism, and allows for a more flexible design with an eye toward future functional expansion.

As such, the difference between eMMC and UFS is not simply a difference in transfer speed, but also affects the overall quality of the ECU, including boot time, OTA update time, operation response, and design margin. After clarifying the ECU's role and Access patterns, determining which characteristics should be prioritized will lead to the selection of the optimal storage.

Choose between eMMC and UFS depending on your needs

Both eMMC and UFS are embedded storage devices with proven track records in automotive ECUs, but the optimal choice will vary greatly depending on the ECU's role and Access patterns. Here, we will organize approaches to selecting storage devices for each typical application.

ECUs whose main purpose is to store setting information and record lightweight logs

eMMC is a promising option. Body ECUs and some TCUs often have relatively simple Access patterns and do not require extreme startup times or random Access performance. For these applications, eMMC is easy to design and offers an excellent cost balance, ensuring sufficient performance and reliability.

ECU whose startup speed and operation response are directly linked to the user experience

The advantages of UFS are clear. In systems such as IVI and digital clusters, where screen display and operation are required immediately after POWER SUPPLIES-on, UFS's full-duplex communication and high parallelism can be expected to shorten startup times. The characteristics of UFS also work effectively when there is a lot of random Access to applications and map data.

ECUs that frequently perform OTA updates

Write performance and update time are important factors in determining whether a drive will perform an update. While eMMC also allows for OTA updates, the processing time tends to be long when writing large amounts of data at once. With UFS, features such as Write Booster make it easier to shorten update times, which reduces system downtime and improves user convenience.

ADAS-related ECUs and gateways that handle large amounts of data from cameras and sensors

The high bandwidth and high parallelism of UFS are major advantages. In applications where there is a lot of random Access involved in writing log data and recognition results, and analysis processing, eMMC can become a bottleneck, so choosing UFS can give the entire system more capacity.

As such, when selecting storage, it is important not only to compare simple performance, but also to clarify what the ECU will handle and how. Selecting the method that best suits the ECU's requirements, based on such concepts as eMMC if cost and ease of design are important, and UFS if fast boot time, future expandability, and user experience are important, will lead to a stable system design.

Summary

Both eMMC and UFS are embedded storage based on the JEDEC standard, but they are suited to different applications. eMMC offers a simple configuration and excellent cost balance, making it suitable for Access with relatively simple access. On the other hand, UFS offers high speed and high parallel processing performance, making it effective for ECUs where boot time and operation response are important.
When selecting storage for an automotive ECU, it is important to consider boot time, Access patterns, OTA update frequency, write endurance (TBW), and future expandability. Selecting between eMMC and UFS depending on the application will lead to optimal system design.
We also offer consultations on selecting storage methods, capacity, and generation based on ECU usage, Access characteristics, and reliability requirements. If you have any concerns about selecting storage, please feel free to Inquiry.