The Embedded MultiMediaCard (eMMC) is an internal data storage card, built using flash storage. Its tiny size and low price make it a popular choice for data storage in portable devices like smartphones, tablets, cameras, and laptops. Users can usually increase their storage capacity by adding removable Secure Digital (SD) cards.
However, despite its portable device credentials, eMMC is not limited to consumer devices. The same features that make it popular for phones and cameras, also make it popular for the tiny sensors and control devices that are the heart of the Internet of Things.
What is eMMC Flash Storage?
Developers build eMMC on top of the MultiMediaCard (MMC) standard for consumer-level storage. The eMMC integrated circuit (IC) uses a parallel connection to attach to the main circuit board of its device. The chip’s integrated controller takes over the right function from the CPU, which frees up the CPU to do other tasks. This architecture uses far less power than spinning disk and is considerably smaller than SSDs. (By the way, be aware that eMMC vs. eMCP are closely related products; eMCP is an embedded multi-chip package; both products are embedded solutions.)
- Size – MMC cards are about as big as a postage stamp, and are common in lightweight portable devices. MMCs can be removable. For example, For example, a digital camera would write photos to a removable MMC card. The user would use an MMC reader to transfer data between the card and a PC or the cloud. Another option is an operating system that reads MMC cards when the user inserts the card in the PC’s Secure Digital (SD) slot.
- Standards – JEDEC published the most recent eMMC standard in 2015. The 5.1 eMMC standard simplifies the card’s interface design by moving the flash controller into flash memory, containing both within a single integrated circuit (IC). This simplified design allows manufacturers to take advantage of cheaper production costs and smaller sizes. (See image below for the speeds of the various standards; note that each successive standard improves speed and efficiency.)
- NAND flash – eMMC builds on the MMC standard to create nonremovable NAND flash memory located on the motherboard. The bootable eMMC card houses a flash controller and NAND flash memory. It provides a decent amount of storage in a low-cost and lightweight form – typically 32GBs or 64GBs eMMC memory, and 128GB is available. Built-in controllers enable eMMC as bootable internal storage that takes the place of more expensive and heavier SSDs.
eMMC vs. SSD
eMMC cards stand at a mid-price point and mid-performance level between HDDs and SSDs. The tiny size and lower price of the eMMCs make it very suitable for midrange laptops, smaller consumer devices, and of course tiny sensors and control devices.
Unlike some other types of flash memory, both eMMC and SSDs are bootable. This feature allows laptop makers to manufacture laptops with internal flash storage. Although HDD still comprise most of the laptop internal storage, SSDs and eMMC’s allow manufacturers to offer higher performance laptops that are also lighter and thinner.
The majority of SSDs are highest performance and highest cost, although eMMC speeds are commensurate with SATA SSDs.
Flash and SSD are closely intertwined, but not synonymous. Although SSDs overwhelmingly use flash memory, specifically NAND, they are not limited to flash. Before 2009, most SSDs used DRAM volatile memory. NAND with its integrated circuits and persistent memory has largely replaced it.
However, SSD and flash are so closely connected that is fair enough to talk about SSD as flash storage, and compare it to other types of flash devices. Thumb drives and SD cards are designed to be inexpensive and simple to manufacture. USB flash drive memory chips exist on a printed circuit board. The flash drive also has a USB interface and basic controller. SD cards are also simple, and contain a flash memory chip on a circuit board and an SD controller.
SSDs are far more advanced than these types of flash memory devices, which is why it is popular in enterprise data centers. Instead of directing reads and writes to a single chip, parallel read/writes distribute operations across multiple NAND flash chips for much higher performance.
Wear-leveling is another advanced feature. The firmware selectively balances data writes across flash memory chips to prevent wearing down individual chips. SSDs also support new features like TRIM, which protects the SSD from overcapacity by physically deleting file-based data from the chip when the user deletes the file, which increases drive performance. Unlike hard drives, which simply overwrite, the SSD must erase a block for writing to it. By automatically deleting the data from a chip, the sector is empty and immediately ready for a new write.
In addition, the SSD interface is usually faster than available interfaces for flash memory devices. eMMC rivals SATA interface speeds, but in general SSD interfaces are faster. SSDs’ popularity also spurs development around fast interfaces, form factors like M.2, filesystems, and protocols such as NVM Express (NVMe).
eMMC Speed Performance
The 5.1 eMMC storage standard delivers transfer speeds up to 400MB/s. That’s roughly equivalent to SATA SSD highest transfer speeds – and clearly fast enough for enterprise applications. Each new generation has seen an eMMC upgrade in terms of performance.
However, performance is not measured simply by transfer speeds, but also by the number of memory gates. The more gates, the faster the storage device can process data. eMMC storage typically has fewer memory gates than standard SSDs. This more limited transfer performance reduces eMMC workload performance, making it a limited replacement for primary memory. However, it can create additional virtual memory when it acts as a page store.
Each new generation of eMMC offers major improvements in speed and performance.
Expanding into the Enterprise: IoT and eMMC
eMMC will continue to work well in portable consumer devices. Yet despite its popularity as a lower-priced alternative to SSDs, eMMC cards have a very promising future: they are crucial to the advance of the Internet of Things.
IoT manufacturers are building eMMCs into the tiny sensors and control devices built into refrigerators, streets, cars, warehouses: almost any type of physical device or structure that can generate usage data. These sensors receive data and transfer it to a central data storage repository.
The sensors are not made to store data long-term; that’s the data repository’s job. But the sensors do need persistent memory to make successful data transfers. And eMMCs with their minuscule size and flash performance enables those sensors to collect vast information at the enterprise level: the sciences, smart houses, autonomous vehicles, transportation industries, manufacturing, retail, and more.
HDDs and even SSDs are far too large to fit on small sensor sizes and related products such as consumer wearable devices. eMMCs with their tiny integrated circuits serve this market far better with their minuscule footprint. And with current transfer rates of about 400MB/s, data rates are more than enough for transfer speed requirements.
Smart cars are also a growing user of eMMC, particularly with intelligent onboard navigation and entertainment systems. There are even specialized eMMCs that are ruggedized for demanding environmental stressors in vehicle and industrial usages.
Another advantage for using eMMC for IoT is cost. The price of millions and billions of flash memory modules spread across millions of sensors – well, they add up. eMMC is much less expensive than buying sensor devices with memory slots and SD cards.
The primary requirements for IOT device storage fit right into eMMC territory: small size, acceptable performance, and low cost thanks to a simplified manufacturing process. This combination enables eMMC to work not only with portable consumer devices and IOT sensors, but also with meters, wearable devices, and robots.