Vik


FMS 2014: HGST Announces FlashMAX III PCIe SSDs

FMS 2014: HGST Announces FlashMAX III PCIe SSDs

Continuing with our Flash Memory Summit coverage, HGST announced their FlashMAX III enterprise SSD, which is the first fruit of HGST’s Virident acquistion and continues Virident’s FlashMAX brand. The FlashMAX III will come in half-height, half-length form factor and will be available in capacities of 1100GB, 1650GB and 2200GB. The controller is an FPGA-based 32-channel design with a PCIe 3.0 x8 interface, but there is no NVMe support since the FlashMAX III builds on the same architecture as the previous generation FlashMAX II. 

HGST FlashMAX III Specifications
Capacity 1100GB 1650GB 2200GB
Form Factor Half-Height, Half-Length (HH-HL)
Interface PCIe 3.0 x8
Controller 32-channel FPGA based
NAND Micron 20nm 64Gbit MLC
Sequential Read 2.7GB/s 2.0GB/s 2.7GB/s
Sequential Write 1.4GB/s 1.0GB/s 1.4GB/s
4KB Random Read 549K IOPS 409K IOPS 531K IOPS
4KB Random Write 53K IOPS 30K IOPS 59K IOPS
4KB 70/30 Random Read/Write 195K IOPS 145K IOPS 200K IOPS
Write Latency < 30 µsec
Max Power 25 watts
Endurance 2 DWPD
Warranty Five years

The maximum throughput seems a bit low for a design that uses up eight PCIe 3.0 lanes since 2.7GB/s should be achievable with just four PCIe 3.0 lanes. Obviously performance scaling is not that simple but for example Samsung’s XS1715 (which we will be reviewing soon!) is rated at up to 3.0GB/s while only consuming four PCIe 3.0 lanes. Using less PCIe lanes allows for more drives to be delpoyed as the amount of PCIe lanes is always rather limited.

The 1650GB model is even slower due to the fact that it utilizes less NAND channels because it is a middle capacity. Basically, the 1100GB and 2200GB models have the same number of NAND packages, with the 2200GB model having twice as much NAND per package; the 1650GB model uses the higher capacity packages but doesn’t fully populate the board. HGST told us that they are just testing the water to see if there is demand for something in between 1100GB and 2200GB.

The FlashMAX III also supports Virident Flash-management with Adaptive Shceduling (vFAS), which is a fancy name for Virident’s storage driver. vFAS presents the FlashMAX as a single volume block device to the OS, meaning that no additional storage protocols or controllers are needed, whereas some drives just use a RAID controller or need software RAID solutions to be configured into an array. Additionally vFAS handles NAND management by doing wear-leveling, garbage collection, data path protection, NAND-level parity, ECC, and more. 

The FlashMAX III is currently being qualified by select OEMs and will ship later in this quarter. 

FMS 2014: HGST Announces FlashMAX III PCIe SSDs

FMS 2014: HGST Announces FlashMAX III PCIe SSDs

Continuing with our Flash Memory Summit coverage, HGST announced their FlashMAX III enterprise SSD, which is the first fruit of HGST’s Virident acquistion and continues Virident’s FlashMAX brand. The FlashMAX III will come in half-height, half-length form factor and will be available in capacities of 1100GB, 1650GB and 2200GB. The controller is an FPGA-based 32-channel design with a PCIe 3.0 x8 interface, but there is no NVMe support since the FlashMAX III builds on the same architecture as the previous generation FlashMAX II. 

HGST FlashMAX III Specifications
Capacity 1100GB 1650GB 2200GB
Form Factor Half-Height, Half-Length (HH-HL)
Interface PCIe 3.0 x8
Controller 32-channel FPGA based
NAND Micron 20nm 64Gbit MLC
Sequential Read 2.7GB/s 2.0GB/s 2.7GB/s
Sequential Write 1.4GB/s 1.0GB/s 1.4GB/s
4KB Random Read 549K IOPS 409K IOPS 531K IOPS
4KB Random Write 53K IOPS 30K IOPS 59K IOPS
4KB 70/30 Random Read/Write 195K IOPS 145K IOPS 200K IOPS
Write Latency < 30 µsec
Max Power 25 watts
Endurance 2 DWPD
Warranty Five years

The maximum throughput seems a bit low for a design that uses up eight PCIe 3.0 lanes since 2.7GB/s should be achievable with just four PCIe 3.0 lanes. Obviously performance scaling is not that simple but for example Samsung’s XS1715 (which we will be reviewing soon!) is rated at up to 3.0GB/s while only consuming four PCIe 3.0 lanes. Using less PCIe lanes allows for more drives to be delpoyed as the amount of PCIe lanes is always rather limited.

The 1650GB model is even slower due to the fact that it utilizes less NAND channels because it is a middle capacity. Basically, the 1100GB and 2200GB models have the same number of NAND packages, with the 2200GB model having twice as much NAND per package; the 1650GB model uses the higher capacity packages but doesn’t fully populate the board. HGST told us that they are just testing the water to see if there is demand for something in between 1100GB and 2200GB.

The FlashMAX III also supports Virident Flash-management with Adaptive Shceduling (vFAS), which is a fancy name for Virident’s storage driver. vFAS presents the FlashMAX as a single volume block device to the OS, meaning that no additional storage protocols or controllers are needed, whereas some drives just use a RAID controller or need software RAID solutions to be configured into an array. Additionally vFAS handles NAND management by doing wear-leveling, garbage collection, data path protection, NAND-level parity, ECC, and more. 

The FlashMAX III is currently being qualified by select OEMs and will ship later in this quarter. 

FMS 2014: SanDisk ULLtraDIMM to Ship in Supermicro's Servers

FMS 2014: SanDisk ULLtraDIMM to Ship in Supermicro’s Servers

We are running a bit late with our Flash Memory Summit coverage as I did not get back from the US until last Friday, but I still wanted to cover the most interesting tidbits of the show. ULLtraDIMM (Ultra Low Latency DIMM) was initially launched by SMART Storage a year ago but SanDisk acquired the company shortly after, which made ULLtraDIMM a part of SanDisk’s product portfolio.

The ULLtraDIMM was developed in partnership with Diablo Technologies and it is an enterprise SSD that connects to the DDR3 interface instead of the traditional SATA/SAS and PCIe interfaces. IBM was the first to partner with the two to ship the ULLtraDIMM in servers, but at this year’s show SanDisk announced that Supermicro will be joining as the second partner to use ULLtraDIMM SSDs. More specifically Supermicro will be shipping ULLtraDIMM in its Green SuperServer and SuperStorage platforms and availability is scheduled for Q4 this year. 

SanDisk ULLtraDIMM Specifications
Capacities 200GB & 400GB
Controller 2x Marvell 88SS9187
NAND SanDisk 19nm MLC
Sequential Read 1,000MB/s
Sequential Write 760MB/s
4KB Random Read 150K IOPS
4KB Random Write 65K IOPS
Read Latency 150 µsec
Write Latency < 5 µsec
Endurance 10/25 DWPD (random/sequential)
Warranty Five years

We have not covered the ULLtraDIMM before, so I figured I would provide a quick overview of the product as well. Hardware wise the ULLtraDIMM consists of two Marvell 88SS9187 SATA 6Gbps controllers, which are configured in an array using a custom chip with a Diablo Technologies label, which I presume is also the secret behind DDR3 compatibility. ULLtraDIMM supports F.R.A.M.E. (Flexible Redundant Array of Memory Elements) that utilizes parity to protect against page/block/die level failures, which is SanDisk’s answer to SandForce’s RAISE and Micron’s RAIN. Power loss protection is supported as well and is provided by an array of capacitors. 

The benefit of using a DDR3 interface instead of SATA/SAS or PCIe is lower latency because the SSDs sit closer to the CPU. The memory interface has also been designed with parallelism in mind and can thus take greater advantage of multiple drives without sacrificing performance or latency. SanDisk claims write latency of less then five microseconds, which is lower than what even PCIe SSDs offer (e.g. Intel SSD DC P3700 is rated at 20µs).

Unfortunately there are no third party benchmarks for the ULLtraDIMM (update: there actually are benchmarks) so it is hard to say how it really stacks up against PCIe SSDs, but the concept is definitely intriguing. In the end, NAND flash is memory and putting it on the DDR3 interface is logical, even though NAND is not as fast as DRAM. NVMe is designed to make PCIe more flash friendly but there are still some intensive workloads that should benefit from the lower latency of the DDR3 interface. Hopefully we will be able to get a review sample soon, so we can put ULLtraDIMM through our own tests and see how it really compares with the competition.

FMS 2014: SanDisk ULLtraDIMM to Ship in Supermicro's Servers

FMS 2014: SanDisk ULLtraDIMM to Ship in Supermicro’s Servers

We are running a bit late with our Flash Memory Summit coverage as I did not get back from the US until last Friday, but I still wanted to cover the most interesting tidbits of the show. ULLtraDIMM (Ultra Low Latency DIMM) was initially launched by SMART Storage a year ago but SanDisk acquired the company shortly after, which made ULLtraDIMM a part of SanDisk’s product portfolio.

The ULLtraDIMM was developed in partnership with Diablo Technologies and it is an enterprise SSD that connects to the DDR3 interface instead of the traditional SATA/SAS and PCIe interfaces. IBM was the first to partner with the two to ship the ULLtraDIMM in servers, but at this year’s show SanDisk announced that Supermicro will be joining as the second partner to use ULLtraDIMM SSDs. More specifically Supermicro will be shipping ULLtraDIMM in its Green SuperServer and SuperStorage platforms and availability is scheduled for Q4 this year. 

SanDisk ULLtraDIMM Specifications
Capacities 200GB & 400GB
Controller 2x Marvell 88SS9187
NAND SanDisk 19nm MLC
Sequential Read 1,000MB/s
Sequential Write 760MB/s
4KB Random Read 150K IOPS
4KB Random Write 65K IOPS
Read Latency 150 µsec
Write Latency < 5 µsec
Endurance 10/25 DWPD (random/sequential)
Warranty Five years

We have not covered the ULLtraDIMM before, so I figured I would provide a quick overview of the product as well. Hardware wise the ULLtraDIMM consists of two Marvell 88SS9187 SATA 6Gbps controllers, which are configured in an array using a custom chip with a Diablo Technologies label, which I presume is also the secret behind DDR3 compatibility. ULLtraDIMM supports F.R.A.M.E. (Flexible Redundant Array of Memory Elements) that utilizes parity to protect against page/block/die level failures, which is SanDisk’s answer to SandForce’s RAISE and Micron’s RAIN. Power loss protection is supported as well and is provided by an array of capacitors. 

The benefit of using a DDR3 interface instead of SATA/SAS or PCIe is lower latency because the SSDs sit closer to the CPU. The memory interface has also been designed with parallelism in mind and can thus take greater advantage of multiple drives without sacrificing performance or latency. SanDisk claims write latency of less then five microseconds, which is lower than what even PCIe SSDs offer (e.g. Intel SSD DC P3700 is rated at 20µs).

Unfortunately there are no third party benchmarks for the ULLtraDIMM (update: there actually are benchmarks) so it is hard to say how it really stacks up against PCIe SSDs, but the concept is definitely intriguing. In the end, NAND flash is memory and putting it on the DDR3 interface is logical, even though NAND is not as fast as DRAM. NVMe is designed to make PCIe more flash friendly but there are still some intensive workloads that should benefit from the lower latency of the DDR3 interface. Hopefully we will be able to get a review sample soon, so we can put ULLtraDIMM through our own tests and see how it really compares with the competition.

Browser Face-Off: Chrome 37 Beta Battery Life Revisited

Browser Face-Off: Chrome 37 Beta Battery Life Revisited

Last week we posted our Browser Face-Off: Battery Life Explored 2014, where the battery run down times of Firefox 31, IE11 Desktop, IE11 Modern, Chrome 36, and Chrome 37 beta were tested on Windows. We used GUI automation to open browsers, tabs, and visit websites to simulate a real user in a light reading pattern. The article answered a lot of questions about popular browser battery life on Windows, but it raised additional questions as well.

Chrome 36 tested with the best battery life, but was the only browser that did not render correctly at 3200×1800 due to lack of HiDPI support. In the Chrome 37 beta, HiDPI support improved rendering but also took a 25% dive in battery life tying it for last place. However, the Chrome 37 beta includes more changes than just HiDPI support (along with some debugging code), so was the battery life penalty from the now-native 3200×1800 rendering or was it something else? After a few more days of testing at 1600×900 with 100% DPI scaling, we can narrow in on an answer.

Chrome Battery Life @ 1600x900

When both Chrome 36 and Chrome 37 beta natively render at 1600×900 there is less than 3% difference in battery life. Two tests of each browser were performed and the results averaged. The variation between runs was only 1%. Looking at our previous numbers of Chome 36 and 37 beta on the HiDPI setting of 3200×1800 and 200% scaling, the situation is entirely different.

Chrome Battery Life @ 3200x1800

I’ve added an asterisk here (and clarified the same text on the original article) to indicate Chrome 36 isn’t actually rendering at 3200×1800, but rather at 1600×900 and relying on Windows DPI Virtualization to scale up to 3200×1800.

Looking at the numbers, there’s some good news and some bad news. The good news is that Chrome 37’s new features likely won’t hurt the battery life of current users. If you are using Chrome now, you are probably not using a HiDPI display due to the existing blurry rendering. For these users, the pending Chrome 37 upgrade has no significant impact on battery life. The bad news is that if you have been avoiding Chrome due to its HiDPI rending issues, Chrome 37 resolves those issues but also appears to provide worse battery efficiency compared to Internet Explorer. On our XPS 15 that equated to about an hour less of mobility.

Given that this is the first version of Chrome to properly support HiDPI, it’s entirely possible – even likely – that there are many opportunities to further optimize the algorithms and hopefully return battery life at least close to Chome 36 levels. A slight dip in battery life is expected as it takes more work to render a 3200×1800 image compared to a 1600×900 image, but a 20% drop seems rather extreme. We’ll have to see what future updates bring, but hopefully by noting the discrepancy it will encourage developers to better tune performance.