Samsung Demos Its First BGA SSD: 1500 MB/s Read Speed and Tiny Package

In the recent years SSDs in M.2 form-factor have greatly reduced the amount of space required for storage sub-systems inside modern PCs. However, as computers get even smaller, there is pressure for SSDs to further shrink as well. Solid-state drives in BGA packaging are considerably smaller than SSDs in M.2 form-factor and for a couple of years platform developers like Intel have been promoting such drives among PC makers. Intel said in 2014 that an M.2-2260 SSD could take 15% of space inside a 2-in-1 hybrid PC, whereas usage of a BGA SSD could save a great amount of space and could allow to increase battery size by around 10%. Moreover, in many cases usage of BGA storage can shrink Z-height of devices as well as improve thermal performance compared to M.2 modules, according to Intel.

As a result, the market for BGA form factor SSDs has been growing in size and importance at a quick pace. To that end, at the 2016 Samsung SSD Forum Japan, Samsung demonstrated its first SSD in a BGA (ball-grid array) packaging. Despite of tiny form-factor, Samsung’s PM971 SSD offers rather high performance thanks to PCIe interface and a special controller. Meanwhile the little drive will be targetted for use inside PCs, tablets, 2-in-1s and other types of small computers or embedded applications.

First BGA SSD From Samsung

Samsung’s PM971 SSD is based on Samsung’s new Photon controller as well as MLC V-NAND flash memory. It is impossible to say at this time whether the Photon controller of the PM971 SSD has anything to do with the controller used inside the 750 EVO drives (the controller is smaller than usual as it has only two processing cores), but it looks very likely.

Samsung plans to offer three versions of the SSD, featuring 128 GB, 256 GB, and 512 GB capacities. The PM971 SSD supports sequential read speed of up to 1500 MB/s as well as sequential write speed of up to 600 MB/s, suggesting that it relies on PCIe 3.0 interface. The PM971 SSD can perform 190K random read IOPS as well as up to 150K random write IOPS, according to information disclosed by Samsung.

Samsung PM971 SSD. Image by PC Watch.

At the event, Samsung did not reveal anything about internal architecture of its BGA SSD, which is rather interesting because the NAND controller has to extract maximum performance out of a limited number of NAND devices over a limited number of channels (provided, of course, that internal architecture relies on industry standards). Moreover, it is unknown whether the drive is compatible with the proposed M.2 specification for BGA SSDs. The only thing that is known about the SSD right now is that it is smaller than an SD card, according to PC Watch web-site.

Samsung is aiming for tablets and 2-in-1 hybrid PCs with its BGA SSDs, The company expects device manufacturers to adopt the product in the second half of 2016 or in the first half of 2017.

Standards for BGA SSDs Proposed

While we’re on the subject of BGA SSDs, this is a good time to touch upon the recent developments in the standardization of SSDs in this format. Even with the recent growth of the market, SSDs in BGA packaging are not something completely new. Intel, Toshiba, SanDisk, Microsemi, Silicon Motion and some others have all offered SSDs in BGA form-factors for years to various makers of embedded applications, whom in turn needed to save space or run their storage sub-systems in harsh environments. Early last year Toshiba rolled-out its first BGA SSDs with PCIe interface, and in September several members of PCI SIG (the organization, which develops PCIe interface, its derivatives and standards for PCIe-based devices) proposed a set of mechanical standards for BGA SSDs, which could open the doors for many manufacturers to enter the emerging market.

BGA SSDs with PCIe 3.0 or SATA interfaces will be a part of the PCI SIG’s M.2 specifications. At present select members of the organization (HP, Intel, Lenovo, Micron, SanDisk, Seagate and Toshiba) propose four types of soldered-down solid-state storage solutions: Type 1620, Type 2024, Type 2228 and Type 2828. M.2 types traditionally define width and length of the package in millimeters, so, the smallest BGA SSD will have measurements of 16 × 20 mm, whereas the largest BGA SSD will feature 28 × 28 mm packaging. SSDs in a BGA package may have Z-height up to 2 mm (measured with solder balls collapsed), but may be slimmer.

BGA SSDs that comply with the proposed M.2 types use the same signals as the M.2 socket 3 (so, they are compliant with both PCIe 3.0 and SATA 3.2 protocols and can utilize up to four PCIe lanes), but use 1.2V, 1.8V and 3.3V power rails. All the BGA SSDs contain the common core ball map of Type 1620 (which look pretty much the same as the core ball map of Toshiba’s BG-series SSDs) for data and power, but the larger drives also feature additional mechanical retention balls. Some BGA SSDs can be placed on M.2 modules in a bid to enable design flexibility for device makers (and upgradeability option for end-users), but only on the condition that the modules have voltage conversion circuitry to provide 1.8V or 1.2V as required.

A BGA SSD not only integrates NAND flash memory, but also a NAND controller, DRAM and all the things that could be needed for a fully-functional solid-state storage solution. From the software perspective, M.2 spec BGA SSDs are just solid-state drives with PCIe or SATA interfaces. And unlike MCPs (multi-chip packages) consisting of DRAM and NAND that companies like Micron and Samsung offer to makers of mobile devices, BGA SSDs are complete storage systems that can be attached directly to appropriate host interfaces.

The proposed BGA M.2 form-factors intend to unify packaging of single-chip SSDs and make such SSDs industry-standard devices available from various makers. It is unknown at this time whether Samsung’s PM971 will be compliant with the proposed specs, but it’s worth noting that the company is a member of PCI SIG and typically makes products in standard packages and form factors.

OWC Introduces SSD Upgrade for MacBook Pro and MacBook Air PCIe SSDs

Apple’s Retina MacBook Pro and all but the earliest MacBook Air models have relied solely on SSDs for internal storage, as Apple slimmed down the designs to the point that even a 1.8″ hard drive was too bulky. Rather than adopt the mSATA or later M.2 form factor, Apple’s SSDs have used custom form factors and pinouts. This has contributed to keeping the market for third-party upgrades very small. Only a few companies have produced SSDs in Apple-specific form factors, most notably Other World Computing (OWC) and Transcend. Transcend has generally used Silicon Motion controllers while OWC has used SandForce controllers, but until now their offerings have been limited to SATA-based SSDs.

Apple migrated their notebook SSDs to PCIe-based interfaces in 2013 and has been using drives supplied by Toshiba, SanDisk, and Samsung. OWC has finally devised a compatible replacement and released it as part of their Aura SSD product line. Like the Apple originals, the OWC Aura PCIe SSD uses the AHCI protocol; Apple so far only supports and uses NVMe on the Retina MacBook that doesn’t have a removable SSD. The requirement to use AHCI instead of NVMe limited OWC’s choices for SSD controller. While Apple is a big enough customer to convince Samsung to make the SM951 in a custom form factor, OWC is not. Marvell has shipped several AHCI-compatible PCIe SSD controllers, but their typical business model is to sell just the controller and leave it up to the customer to write their own firmware or license from a third party, either of which is a substantial up-front expense.

In order to keep costs under control, OWC has opted to not use a native PCIe SSD controller. Instead, the PCIe Aura SSD uses a Marvell 9230 SATA RAID controller and a pair of Silicon Motion SM2256 SATA SSD controllers. The Marvell 9230 has a PCIe 2.0 x2 host interface, so the PCIe Aura SSD has the potential to outperform SATA SSDs but won’t be able to approach the peak transfer rates of the recent Samsung SM951-based Apple originals. The Silicon Motion SM2256 controllers mean the PCIe Aura SSD is almost certainly using TLC flash, which is less expensive but also performs worse and draws more power than MLC flash. The PCIe Aura SSD’s RAID design unfortunately does not support passing through TRIM commands nor retrieving SMART information from the individual SSD controllers.

Based on OWC’s measurements of the first PCIe SSDs Apple used back in 2013, the Aura SSD’s peak performance is slightly better than the slowest 128GB SanDisk/Marvell drive, but without TRIM the Aura’s write performance advantage could easily disappear over time. That leaves the PCIe Aura SSD with capacity as its only strong selling point. The MacBook Air can be configured with up to 512GB of storage from Apple, but the Aura SSD can provide up to 960GB. Many Apple customers are put off by the steep price of build-to-order SSD upgrades: $200 to upgrade from 128GB to 256GB, another $300 to move up to 512GB, and another $500 to move up to 1TB for the MacBook Pro. At $347.99 for 480GB and $597.99 for 960GB, OWC’s Aura manages to be both much cheaper than Apple’s SSD upgrades and much more expensive than single-controller drives with a standard form factor.

The Aura SSD is sold either as a bare drive or an upgrade kit that includes the necessary screwdrivers to install the SSD and a USB 3.0 enclosure to facilitate data migration. The drive is expected to start shipping in late March.

Seagate Announces PCIe x16 SSD Capable Of 10GB/s

At the Open Compute Project Summit this week in San Jose, Seagate will show off a pair of upcoming enterprise NVMe SSDs with impressive throughput specifications. The drives will have PCIe x16 and x8 interfaces and provide maximum throughput of 10GB/s and 6.7GB/s respectively. Seagate has provided few details so far, but it’s safe to say those numbers are peak sequential read speeds.

The big question is what controllers are used in these drives. Most NVMe SSD controllers support at most 4 PCIe lanes, with the notable exceptions being PMC-Sierra’s 8-lane controllers. Seagate does have an internal development team with SandForce, but it’s highly unlikely they’ve been able to develop such a large controller so soon. And these new Seagate SSDs are probably not based on unannounced third-party controllers.

This means the 16-lane SSD from Seagate is almost certainly a multi-controller solution with an on-board PCIe switch, which is now common for top of the line enterprise PCIe SSDs. A 10GB/s read speed suggests that the 16-lane drive is most likely based on four of Seagate’s Nytro XM1440 M.2 SSDs, which advertise 2.5GB/s read speed for capacities of at least 800GB and use a Marvell controller. Seagate’s blog shows CAD renderings that seem consistent with a layout of four Nytro XM1440 M.2 drives on one card, but the requisite PCIe switch chip isn’t shown. (EDIT: Photographs of the board have surfaced showing that it does passively route the PCIe lanes from the x16 connector to the M.2 slots without a PCIe switch chip. This may prevent all four drives from being used on platforms like Intel’s Xeon E3 where the 16 PCIe lanes can only be divided into x8+x8 or x8+x4+x4 configurations.)

Seagate claims that the 10GB/s speed of the 16-lane drive is 4GB/s faster than any competing drive. If Seagate’s new drive is a single-controller solution then that’s a fair comparison and an impressive accomplishment, but there are already multi-drive products on the market offering RAID0 speeds well in excess of 6GB/s. HP’s Z Turbo Drive Quad Pro is a PCIe x16 card that provides connectivity and cooling for up to four M.2 PCIe SSDs. When ordered as part of a workstation it can be configured with four Samsung SM951 SSDs to provide an advertised 9GB/s sequential read speed, though when sold separately it only comes populated with two M.2 SSDs.

Meanwhile the 8-lane drive is probably not based on two 4-lane controllers, despite it being the most obvious solution. Most products based on a single controller from Intel, Samsung, Marvell or Phison with a PCIe 3.0 x4 link advertise maximum read speeds of 2.2–2.8GB/s, so providing 6.7GB/s from just two controllers would require 20% higher performance than any PCIe 3.0 x4 NVMe controller has attained. Instead, the PCIe x8 SSD Seagate is announcing is probably another four Marvell controller design that is limited in sequential speeds by the overhead of the drive’s PCIe switch and the upstream PCIe link. The 1M IOPS claimed for the 8-lane drive is slightly higher than four times the rating for a single XM1440, but some capacities of the 2.5″ XF1440 offer enough IOPS. The thermal constraints of the M.2 form factor compared to 2.5″ drives and add-in cards with large heatsinks account for the discrepancy in IOPS rating.  According to Seagate the 8-lane drive will offer some cost and power savings over the 16-lane drive, and it’s not hard to imagine that it could also allow servers to a larger total capacity for the same number of PCIe lanes.

Seagate’s blog shows a rendering of the 8-lane card with the same heatsink layout as their Nytro XP series flash accelerator cards that use a RAID of SandForce SATA controllers to provide up to 4GB/s sequential read speeds. There’s a good chance this is just a placeholder illustration, as Seagate says the 8-lane drive is still being finalized.

Both drives are intended for the Open Compute Project (OCP) hardware ecosystem founded by Facebook and now also supported by a variety of major companies in cloud computing, telecom, networking, and finance. The Open Compute Project focuses on datacenter hardware and infrastructure, with members contributing specifications and designs that are more detailed than industry standards like the ATX form factor. Seagate says their new drives will comply with OCP specifications, but the specific standards haven’t been identified. Potentially relevant standards include a specification for thermal monitoring of PCIe add-in cards and a specification for M.2 SSDs that sets standards for things like minimum performance, the conditions under which thermal throttling is permitted, maximum power consumption and mandatory eDrive encryption support.

Based on the assumption that both drives are rougly equivalent to four Nytro XM1440 drives plus a PCIe switch chip, peak power consumption will probably be at least 29W for the 8-lane drive and could be nearly 40W for the 16-lane drive.

Seagate describes the PCIe x16 drive as production-ready but the 8-lane drive is still being finalized. Samples of each have been made available to Seagate’s customers and the full product launch is planned for summer of 2016. Capacities have not been announced but are likely to start at 3.2TB or 3.84TB for the highest-performing models.