Crossbar’s RRAM to boast terabytes of storage, faster write speeds than NAND

Crossbar's RRAM to boast terabytes of storage, faster write speeds than NAND

Hardware makers often sing the praises of their latest and greatest flash memory, but the folks at Crossbar are ready to show them up with resistive RAM (RRAM) that they’ve been quietly working on. Compared to NAND, RRAM comes in at half the size and boasts 20 times faster write speeds (140MB/s), reads data at 17MB per second, guzzles 20 times less power and has 10 times more endurance. Since RRAM is non-volatile memory, it can keep data even when it’s powered off, á la NAND. As if that weren’t enough, 3D stacking construction allows for several terabytes of storage, endowing one 200 x 200mm 200mm2 chip with one terabyte.

Unlike many tech breakthroughs however — we’re looking at you, graphene — this one is just about ready to find its way into finished products. Crossbar has manufactured RRAM within a standard chip factory, and claims that it can be churned out easily with existing production infrastructure. According to the firm, it’s in the fine-tuning process and plans to introduce the tech into the world of embedded SoCs. Sure, the outfit is the exclusive holder of some RRAM patents, but it aims to license its know-how to system-on-a-chip creators.

Update: Thanks to those readers who spotted our error on the silicon area — it’s now been fixed.

Filed under:

Comments

Via: VentureBeat

Source: Crossbar

Samsung develops ‘world’s fastest’ embedded memory, first with eMMC 5.0

DNP Samsung develops superfast embedded memory, first with eMMC 50

The evolution of mobile memory chips keeps moving faster, if Samsung’s progress is any indication. Not eight months since churning out those speedy eMMC 4.5 chips, and the company’s next version of NAND is already here. Indeed, the South Korean company says it’s now in mass production of what it deems as the world’s fastest embedded memory. The new eMMC PRO chip is based on 64GB 10nm class NAND flash technology and would be the first to support the eMMC 5.0 standard. The chips will be available in the usual 16, 32 and 64GB iterations and feature an interface speed of 400MB/s. The 32 and 64GB densities in particular boast random read and write speeds of 7,000 IOPS (inputs/outputs per second) and sequential read and write speeds of 250MB/s and 90MB/s respectively. What do all of those numbers mean? Well, they should translate to much better performance when it comes to multitasking, browsing, file transfers, HD video capture, gaming and just general computing. Combined with the firm’s upcoming Exynos 5 Octa 5420 SoC, and we’re champing at the bit to see them implemented in Samsung’s next generation of mobile devices.

Filed under:

Comments

Innodisk outs embedded SATA nanoSSD, nets 480MB per second from one chip

Innodisk unveils tiny yet speedy embedded SATA nanoSSD

While single-chip SSDs are clearly known quantities, they usually run at a much more leisurely pace than their larger counterparts. Innodisk doesn’t think size and speed have to be contradictory — it just unveiled an embedded version of its nanoSSD that performs almost as well as its much bigger counterparts. The µSSD-based SATA chip has a tiny footprint (0.63 x 0.79 inches) and draws just 1W of peak power, but can still read at up to 480MB/s and write at 175MB/s. As such, it’s one of the few SSDs that can theoretically stuff desktop-class storage into a smartphone or tablet. Whether or not it will is another matter. Innodisk hasn’t named customers for the nanoSSD so far, which leaves us guessing just where or when we’ll see the drive in a finished product.

Filed under: , , ,

Comments

Source: Innodisk

Fusion-io brings Fusion ioScale SSD to small, speedy server clusters

Fusionio brings 32TB Fusion ioScale SSDs to sma server clusters

Fusion-io has made a name for its Fusion ioDrive solid-state drives by selling them to the largest of enterprises — the sort that crave thousands of servers. Not everyone wants that level of computing muscle, though, which is why the pro-grade storage firm is now selling the Fusion ioScale to a much wider audience. Cloud service hosts and other, smaller companies just have to buy a (relatively) paltry 100 or more of the PCI Express-based drives, which include both slim 1.6TB and full-size, 3.2TB versions. Neither will be cheap for datacenters when prices start at $3.89 per gigabyte, although Fusion-io is vowing better deals for those buying in buik. We also suspect that the time saved by moving to fast flash storage could be worthwhile in itself.

Continue reading Fusion-io brings Fusion ioScale SSD to small, speedy server clusters

Filed under:

Comments

Source: Fusion-io

Macronix plans to heat up flash memory to keep it from burning out

Macronix plans to heat up flash memory to keep it from burning out

Despite the looming threat of being replaced by phase-change memory, contemporary memory modules aren’t quite ready to be shown the door — engineers at Macronix have found a way to revive spent NAND flash cells. Most flash modules fail after being written to and erased about 10,000 times, but Macronix found that the tired memory could be restored by baking it for extended periods of time. The team funneled the time consuming and cumbersome solution into a more practical package: a redesigned memory chip that packs onboard heaters. The new modules are designed to periodically heat focused groups of memory cells to 800 °C (1,472 °F) for a few milliseconds, effectively “healing” worn cells.

Researchers found that heated chips could tolerate more than 100 million write/erase cycles and erased faster at higher temperatures. The team said the power drain of the heaters shouldn’t effect battery life, either — chips don’t have to be heated often, and when they do, it can be done while prospective devices are recharging. Macronix will be presenting the technology at the IEEE International Electron Devices Meeting next week, but project deputy director Hans-Ting Lue wouldn’t say when the company plans on taking the technology to market. Lue was willing to speculate on what might become of it, however. “This may evolve into a ‘thermally assisted’ mode of operation that gives both better performance — such as the faster erasing — and better endurance flash memory.” Faster, more reliable, super-heated memory. Sounds fine by us.

[Image credit: Emily Cooper, IEEE]

Filed under:

Comments

Via: PhysOrg

Source: IEEE

Samsung creates F2FS file system for NAND flash storage, submits it to the Linux kernel

Samsung creates F2FS file system for NAND flash storage, submits it to the Linux kernel

Has the lack of NAND flash storage-optimized file systems been bugging you? Then you’ve got something in common with Samsung, which has developed F2FS (or “Flash-Friendly File-System”) for the memory of choice for mobile devices and its specific “internal geometry.” It’s based on a log-structured method, but tackles problems associated with older file systems intended mainly for retro, spinning-disk storage. The company isn’t keeping its hard work behind lock-and-patent either — it’s gone open-source and submitted the file system to the Linux kernel, meaning you could see it implemented in Android hardware of the future. It’s nice to see Sammy contributing code for the greater good, and if you’ve got the skills to understand it, a low-down of F2FS is available at the source below.

Filed under: , , , ,

Samsung creates F2FS file system for NAND flash storage, submits it to the Linux kernel originally appeared on Engadget on Mon, 08 Oct 2012 11:43:00 EDT. Please see our terms for use of feeds.

Permalink Phoronix  |  sourceThe Linux-Kernel Archive  | Email this | Comments

Nintendo Wii U expandable memory explained, TiVO hardware not built in

The Nintendo Wii U may only have 8 to 32GB of internal memory, but the company’s taking a decidedly open approach to expanding that memory: add your own. “You can plug in a full-on three terabyte hard drive if you want. I’ll love you as a digital consumer,” Nintendo America CEO and president Reggie Fils-Aime told us during a post-press conference investor Q&A. Much like Sony’s PlayStation 3, the Wii U allows for expansion of memory via USB, whether that memory be Flash or otherwise. Fils-Aime explained that, with the continuously dropping price of memory, there was little reason to offer pricing for the Wii U — available in $300 and $350 — tied to an evolving hardware model.

“The reason we did it that way is that the cost of that type of storage memory is plummeting. What we didn’t want to do is tie a profit model to something that’s gonna rapidly decline over time. We’ll let the consumer buy as much as they want, as cheaply as they want,” Fils-Aime said.

Fils-Aime also confirmed that the Wii U doesn’t have TiVO hardware built into it — functionality will require an existing TiVO box. The same goes for DVR functionality, which Fils-Aime said Nintendo doesn’t want consumers to bear the expense of.

Filed under:

Nintendo Wii U expandable memory explained, TiVO hardware not built in originally appeared on Engadget on Thu, 13 Sep 2012 12:55:00 EDT. Please see our terms for use of feeds.

Permalink   |   | Email this | Comments

Fujitsu and Acacia resolve patent disputes with settlement, keep it out of the courts

Fujitsu and Acacia resolve patent disputes with settlement, keep it out of the courtsFujitsu’s bank balance may be a little lighter today, since Acacia Research Corp. has reported that subsidiaries of both companies have signed a settlement deal over patent disputes. As usual, Acacia is keeping tight-lipped about exactly what the patents cover, but a little digging on our part has revealed they are related to flash memory and RAM technologies. The agreement resolves lawsuits in the works at district courts in Texas and California, which is probably a good thing. After all, these cases can get pretty messy when they go to court.

Continue reading Fujitsu and Acacia resolve patent disputes with settlement, keep it out of the courts

Filed under: , ,

Fujitsu and Acacia resolve patent disputes with settlement, keep it out of the courts originally appeared on Engadget on Mon, 27 Aug 2012 10:33:00 EDT. Please see our terms for use of feeds.

Permalink   |  sourceRopes & Gray (1), (2)  | Email this | Comments

Samsung starts mass-producing 4x faster mobile flash memory, kickstarts our phones and tablets

Samsung starts massproducing 4x faster mobile flash memory, kickstarts our phones and tablets

Samsung isn’t content to leave fast NAND flash memory to traditional solid-state drives. Its Pro Class 1500 promises a big jolt to the performance of frequently pokey smartphone and tablet storage. By how much? That name is a clue — it reaches 1,500 IOPS (inputs/outputs per second) when writing data, which along with 3,500 IOPS data reads is about four times faster than any previous embedded flash chip Samsung has tested. In the real world, that leads to as much as 140MB/s when reading data and 50MB/s for writes. The speed comes after Samsung has thrown virtually every trick in the book at its new chips, including a dense 20-nanometer manufacturing process, quick toggle DDR 2.0 memory with its own controller and a new JEDEC memory standard with 200MB/s of bandwidth to spare. Samsung hasn’t named customers for the 16GB, 32GB and 64GB parts that are rolling out of the factories, although we’d do well to remember that a flourishing phone business doesn’t guarantee that the only major customer is Samsung itself: even in the face of legal challenges, Samsung still has at least one noteworthy client that tends to snap up much of its flash supply.

Continue reading Samsung starts mass-producing 4x faster mobile flash memory, kickstarts our phones and tablets

Filed under: , ,

Samsung starts mass-producing 4x faster mobile flash memory, kickstarts our phones and tablets originally appeared on Engadget on Thu, 02 Aug 2012 01:18:00 EDT. Please see our terms for use of feeds.

Permalink   |  sourceSamsung Tomorrow  | Email this | Comments

Lexar XQD memory cards announced, will duel with Sony for Nikon D4’s attention

Lexar XQD memory cards announced, will need to duel with Sony for Nikon D4's attention

Nikon’s D4 is currently the only belle at the XQD ball, and until now, Sony was its only memory suitor. However, Lexar’s just arrived with a new line of cards to compete for the D4’s affections. The Fremont-based company says it collaborated with Nikon to build the PCI Express-based memory, and that they’ll hit the market sometime in the third quarter of 2012 — with no mention of specs or prices so far. Unfortunately, that means we don’t know how Lexar’s offerings will stack up against Sony’s H-series or S-series cards already on the market. That said, while you wait for more details about Lexar’s new cards, feel free to check out the lovely data the D4 will be putting on ’em when they arrive.

Filed under: ,

Lexar XQD memory cards announced, will duel with Sony for Nikon D4’s attention originally appeared on Engadget on Sun, 15 Jul 2012 02:04:00 EDT. Please see our terms for use of feeds.

Permalink Nikon Rumors  |  sourceLexar  | Email this | Comments