Evolution of the Solid-State Drive

Believe it or not, it has been 35 years since the very first solid-state drive (aka solid-state disk) hit the market. Like all SSDs, that model was designed to appear to a computer like a traditional rotating disk, while storing and retrieving data far faster than traditional hard drives could. Such devices are called “solid-state” because they contain no moving parts, only memory chips.

Over the years, the computer industry’s quest for faster, cheaper, higher-capacity SSDs has driven storage technology in ways no one could have foreseen in 1976, including the use of SSDs as the primary storage component in some consumer PCs.

In the next 15 slides, you’ll witness the evolution of the solid-state drive, from a bulky, obscenely expensive server accessory to a tiny consumer box (with hundreds of gigabytes of capacity) that anyone can buy for $50.

In 1976, Dataram introduced the world’s first solid-state drive, the Bulk Core. The product consisted of a rack-mount chassis–measuring 19 inches wide by 15.75 inches tall–that held up to eight individual memory boards, each packed with 256KB of RAM chips. In total, the Bulk Core system could provide a massive 2MB of storage for minicomputers such as the DEC PDP-11 and the Data General Nova. Data-access times ranged from 0.75 milliseconds to 2 milliseconds, depending on the controller board. (Today, SSDs regularly have 0.06ms access times.)

A Bulk Core setup, including a controller board and 256KB of storage, cost $9700 in 1977, which is equivalent to $36,317 today. At that pricing rate, a 1TB SSD (which costs about $1100 today) would have cost $152 billion.

Photos: Dataram Corporation

The STC 4305, which emulated the IBM 2305 drum-storage unit, represented a significant boost in the capacity of SSDs. A 4305 cabinet could hold up to 45MB of data, which it stored using charge-coupled devices, a novel approach at the time. Such a system, including the required dual controller cards, would set you back $400,000 in 1978 (about $1.5 million in today’s dollars). If that sounds like a lot for storing the equivalent of one moderately sized smartphone app, consider that this device was 52 percent cheaper than the drum-storage equivalent IBM sold.

Photos: Storage Technology Corporation

Magnetic bubble memory has properties similar to modern flash memory in that it doesn’t lose data when you shut off its power. However, the technology hit a wall in terms of capacity early on, and never gained widespread use.

Although bubble memory had been around since the mid-1960s, it wasn’t until Intel released a 1-megabit bubble memory chip, the 7110 in 1979, that consumer-level products started using the technology. In 1982, the chip appeared in a few portable computers such as the Grid Compass, as well as in an early Apple II SSD called the MPC Bubdisk, shown here. The Bubdisk held 128KB of data, and retailed for $895.

Photos: Vintage Technology Association, MPC

In 1982, Nolan Bushnell’s toy company Axlon began selling a line of RAM disks for home PCs such as the Apple II and Atari 800. The Ramdisk 320, intended for the Apple II, retailed for $1395 and held 320KB of data in a box the size of a Disk II drive. Since it stored data on conventional RAM chips that would lose the data if powered off, the 320 included a 3-hour rechargeable battery.

Photo: Axlon

RAM disks existed for all types of computers in the early 1980s, including those based on the aging S-100 bus standard (introduced in the Altair 8800 in 1975). This 1982 advertisement from a Byte magazine issue shows a 256KB “RAM DISC” card developed by SD Systems that sold for $800.

Photo: SD Systems

Axlon was one of many companies producing SSDs for personal computers. The 1983 PION Interstellar Drive worked with many models of home computers and held up to 1MB of storage. Its 256KB base configuration sold for $1095, while each additional 256KB card cost $595.

The Synetix 2202 plugged into an Apple II expansion slot and held up to 294KB for $529. Both products used volatile RAM chips that needed constant power to retain data.

Photos: PION, Synetix

In 1988, a small Alabama-based PC vendor called Digipro revealed a prototype of the world’s first solid-state drive to use flash memory, which the introduction of Intel’s NOR flash memory chips earlier in the year had made possible.

Called simply Flashdisk, Digipro’s plug-in board for IBM PC compatibles could hold up to 16MB of data. It shipped in January 1990 in 2MB, 4MB, 6MB, and 8MB capacities, with the high-end version selling for $5000. An Israeli flash company called M-Systems created its own flash-drive prototype in 1989, but didn’t commercialize it until 1995, making Flashdisk the first flash SSD to market.

Photo: Digipro

In the early 1990s, flash memory was still costly and rare, and it didn’t boast the same data-access times as dynamic RAM-based SSDs did. Such RAM-based SSDs were useful mostly in large server applications that demanded high-speed data access.

DEC, for instance, offered two lines of SSDs in the early 1990s. The EZ5x series of 5.25-inch SCSI-based drives shipped in capacities of 107MB ($13,999) to 428MB ($47,099). The faster ESE50 series offered capacities from 120MB ($40,000) to 1GB ($135,000). With prices like that, DEC’s largest SSD customer was probably Tony Stark.

Photos: Digital Equipment Corporation

The two dynamic RAM-based solid-state drives you see here, the NewerTech Dart Drive (which held up to 512MB) and the ATTO SiliconDrive II (which stored up to 2.6GB), both used the SCSI interface. Priced in the many thousands to tens of thousands of dollars, both were aimed at the high-end Sun workstation market. They offered blisteringly fast access times (0.02ms for the SiliconDrive II) compared to hard drives of the day.

Most RAM SSDs of the 1990s contained both a battery backup and a hard disk that would automatically store the RAM disk’s contents if power were lost.

Photos: NewerTech, ATTO

In 1995, Israeli firm M-Systems set the template for the modern flash-based SSD with its Fast Flash Disk (FFD-350) series, one of the first flash SSDs to ship in the 3.5-inch form factor that most hard drives used at the time. The first FFD-350 models used the SCSI interface and shipped in capacities from 16MB to 896MB. With prices typically in the tens of thousands of dollars per drive, these SSDs found use mostly in military and aeronautical applications that demanded rugged data storage.

Over the next decade, M-Systems continued to expand its FFD line, with higher capacities, faster access times, and different designs, as shown on the right.

Photos: M-Systems

In 2003, Transcend introduced a line of flash modules that emulated Parallel ATA IDE hard drives, which were common in consumer PCs at the time. Each module, much smaller than a traditional PATA hard drive, included a 40- or 44-pin PATA connector and shipped in 16MB to 512MB capacities. Larger capacities followed in later years.

Technology adapted from mass-market flash media cards (commonly used in digital cameras) made Transcend’s modules dramatically less expensive than their predecessors. With prices starting as low as $50, they were some of the first commonly available flash-based SSDs for consumers.

Photos: Transcend

In 2006 Samsung released one of the first mass-market flash SSDs, a 2.5-inch 32GB drive with a PATA interface ($699) designed as a drop-in replacement for laptop hard drives. SanDisk followed in 2007 with its own 2.5-inch 32GB drive, the SATA 5000.

Using wear-leveling technology, flash SSDs in 2006 were capable of many more rewrites than flash media cards at the time, bringing them closer to replacing mechanical hard drives for everyday use. This new generation of products triggered a consumer SSD market explosion that continues to this day.

Photos: Samsung, SanDisk

New SSD technologies are so fast that the SATA interface most hard drives use has become a bottleneck. So manufacturers are continuing the grand tradition of putting SSDs on plug-in cards, as exemplified by the two products shown here. The $1495 DDrive X1 plugs into a spare PCI Express slot and provides 4GB of high-speed DRAM storage. It also includes 4GB of flash for backup in case the power goes out.

The Fusion IoDrive Duo also ships in a PCI Express card format, but it utilizes a specially designed form of flash memory to achieve sustained read speeds of 1.5 GBps (that’s fast). Available in capacities from 128GB to 1.28TB, the IoDrive Duo carried a base price of $5950 at its launch in 2009.

Photos: Fusion, DDrive

Today’s consumer SSDs keep getting faster and cheaper thanks to new flash chips and higher-speed SATA interfaces. For example, the 160GB member of the Intel 320 SSD series, shown here, currently retails for about $320 and offers sustained read speeds of 270 MBps.

Manufacturers are also experimenting with new ways to package SSDs, as illustrated by the Viking Modular SATADIMM: It uses a spare 240-pin DRAM slot on your motherboard to host a tiny, 25GB to 400GB flash-based SSD. Only time will tell if this novel format–which still uses a SATA cable to transfer data–will gain widespread use.

Photos: Viking, Intel

Where will the future of solid-state drives take us? Expect to see more SSDs embedded directly on computer motherboards, in higher capacities, and with greater read/write speeds. SSDs will likely become both bigger and cheaper than traditional hard drives sometime in the next decade, rendering spinning platters obsolete.

We will also see new forms of SSD storage media, such as phase-change memory, which can potentially offer greater capacity, speed, and durability than current flash technology can. Here you can see a prototype PCME module called Onyx, which is under development at the University of California San Diego.

Wherever SSDs take us, at least we know they will take us there very quickly.

Photos: UCSD Non-Volatile Systems Laboratory

Can You Do Real Work With the 30-Year-Old IBM 5150?

Our intrepid reporter spends a week trying to write, browse the Web, edit photos, and even (shudder) tweet on IBM’s first PC.