Anandtech News

**FEED** · 12-06-17, 10:07 AM

Dell EMC on Tuesday introduced its new-generation rugged PowerEdge XR servers, designed to work in harsh environments of industrial sites or even conflict zones. The new machines are equipped with Intel’s latest Xeon Scalable (Skylake-SP) processors along with up to 512 GB DDR4 memory and up to 30 TB of solid-state storage. In addition, the new PowerEdge XR2 features a special common access card that provides an additional way of authentication to drive data encryption and can instantly render the machines useless to the enemy once removed.
As computing becomes pervasive, servers get needed everywhere, including severe environments, such as industrial sites, rural areas, conflict zones and others. Dell is among a few companies that offer ruggedized laptops, tablets, and other hardware, so the company was among the first to address such needs. Dell’s first-gen rugged servers were custom-built machines based on the company’s PowerEdge servers and were sold to select customers. Because of growing demand for such machines and because Dell realized that servers for harsh environments have to be designed for extreme conditions from scratch, the company introduced its first “official” rugged server called the PowerEdge R420xr back in 2014. Today, the company is launching its second-generation purpose-built rugged server (which really is the company’s third-gen rugged server platform) — the PowerEdge XR2.

Just like the predecessor, the Dell EMC PowerEdge XR2 comes in 1U 20” deep chassis that features shock and vibration resistance, optional dust filtration and uses components certified to work in low and high temperatures (from -5ºC to +55ºC) as well as altitudes of up to 15K feet* (think specialized DRAM modules, SSDs, other chips). Meanwhile, the new machine is completely different from the system launched over three years ago and brings huge performance improvements.
The PowerEdge XR2 is based on two Intel’s Xeon Gold (Skylake-SP) processors with up to 22 cores, 30 MB L3 cache and a 140 W TDP. Power consumption is a concern for ruggedized servers because of cooling, so Dell decided to stick to CPUs with a moderate TDP (after all, 44 cores is a lot). The primary CPU socket can be equipped with 10 DIMMs (4 channels at 2 DPC and 2 channels at 1 DPC), whereas the secondary CPU socket can support six DDR4 memory modules (6 channels at 1 DPC), for a potential 512 GB total DRAM per box (previously only 384 GB was supported by the R420xr). Such an unorthodox memory configuration may look a bit odd and physically Intel’s latest Xeon Scalable CPUs can support more memory, but for its rugged servers Dell intends to use certified DIMMs made for severe environments and their capacity is being limited to 32 GB per module today.

The new storage sub-system is something that Dell must be proud of: the PowerEdge XR2 supports eight hot-swappable SATA/SAS SSDs featuring a total capacity of up to 30 TB capacity (up from 6.4 TB on the previous-gen model). Optionally, Dell may equip the system with self-encrypting SSDs, but by default the machine will encrypt the drives itself and will require Dell’s common access card to access/decrypt them. Once such card is removed, the drives cannot be accessed by unauthorized personnel or enemy, which will come in handy in various conflict zones. Depending on the needs, the PowerEdge may be equipped with a variety of RAID controllers.
Two CPUs, multiple memory modules, and eight high-end SSDs consume a substantial amount of power. Dell equips the PowerEdge XR2 with a 550 W redundant PSU, which will feed the aforementioned components. Meanwhile, due to volume restrictions, TDP limits, and power constraints, the new machine will not support any accelerators, just like its predecessor. This may be a pity for oil and gas exploration applications many of which rely on NVIDIA’s Tesla accelerators, but typically oil and gas companies can afford to build custom hardware for their exploration needs.

Depending on exact configuration, the Dell PowerEdge XR2 machines can be equipped with 1Gb, 10Gb BASE-T, 10Gb SFP+, and 40Gb QSFP+ networking cards. The XR2 machines are IPMI compliant and can support Dell’s proprietary iDRAC9 remote management. As for operating systems, expect the machine to be compatible with Microsoft’s Windows Server as well as various Linux builds.

Dell PowerEdge XR2: General Specifications
	PowerEdge XR2 1U/20"
CPU	Two Intel Xeon SP processors Up to 140 W TDP (each): Intel Xeon Gold 6152 Intel Xeon Gold 6140 Intel Xeon Gold 6138 Intel Xeon Gold 6132 Intel Xeon Gold 6130 Intel Xeon Gold 6126 Intel Xeon Gold 5122 Intel Xeon Gold 5120 Intel Xeon Gold 5120T Intel Xeon Gold 5118 Intel Xeon Silver 4116 Intel Xeon Silver 4114 Intel Xeon Silver 4112 Intel Xeon Silver 4110 Intel Xeon Silver 4108 Intel Xeon Bronze 3106 Intel Xeon Bronze 3104
Chipset	unknown
RAM	Up to 512 GB DDR4-2667 RDIMMs with ECC CPU1: 10 DIMMs, 4 ch at 2 DPC, 2 ch at 1 DPC CPU2: 6 DIMMs, 6 channels at 1 DPC 8 GB, 16 GB, 32 GB modules supported
Storage Controllers	PERC H330 PERC H730p PERC HBA330
Storage	2.5" SATA/SAS: up to hotplug 8 drives, 30 TB capacity M.2 SATA: 2 drives for fast boot, OS Up to three drive form-factor storage devices.
Expansion	2 × PCIe x16 for NICs
Networking	Integrated Broadcom 5720 2 x 1GbLOM plus optional LOM Riser LOM Riser Options: Broadcom 5720 2×1Gb Broadcom 57416 2×10Gb Base-T Broadcom 57416 2×10Gb SFP+
I/O	Front ports: - D-Sub, eSATA, USB 2.0, - Dedicated iDRACDirect Micro-USB Rear ports: - D-Sub, RS232, 2×USB 3.0, - Dedicated iDRAC network port
Embedded Management	IPMI 2.0 compliant iDRAC9 with Lifecycle Controller (Express, Enterprise) Quick Sync 2 wireless module optional
Security	TPM 1.2/2.0 optional Cryptographically signed firmware Secure Boot System Lockdown Secure Erase Integrated Common Access Card Reader
Harsh Environment Testing	MIL-STD-810G (temp, shock, vibration, altitude, sand/dust) MIL-STD-461G for conductive/radiative immunity DNV-GL for temperature, humidity, vibration, EMC
PSU	550W redundant

Dell EMC will sell its PowerEdge XR2 machines directly as well as through resellers and OEMs who may adapt them for particular needs by loading software and perform other customizations. OEMs can also install their own bezels and load BIOS with their logotypes and/or features.
Gallery: Dell PowerEdge XR2

*Formally, the machines are compliant with MIL-STD-810G and MIL-STD461G requirements for temperature, shock, vibration, altitude, and conductive/radiative immunity, as well as DNV, IEC 60945 requirements for maritime navigation and radiocommunication equipment when it comes to temperature, humidity, vibration, and EMC.
Related Reading

More...

**FEED** · 12-06-17, 05:08 PM

Today as part of its media event, Qualcomm finally announced the much expected Snapdragon 845, successor to last year's very successful Snapdragon 835. The Snapdragon 845 is a large step in terms of SoC architectures as it's the first to employ ARM's DynamiQ CPU cluster organization. Quickly explained, DynamIQ enables the various different CPU cores within an SoC to be hosted within the same cluster and cache hierarchy, as opposed to having separate discrete clusters with no shared cache between them (with coherency instead happening over an interconnect such as ARM's CCI). This major transition is probably the largest to date that we've seen in modern mobile smartphone ARM consumer SoCs.

More...

**FEED** · 12-07-17, 01:32 PM

Phison has been sampling its entry-level PS5008-E8 NVMe SSD controller for well over a year now, and a number of its partners have demonstrated their drives featuring the chip, but no one has started to ship actual products thus far. As it appears, things are about to change as Lite-On has quietly added its new entry-level MU X PCIe 3.0 x2 SSDs based on Phison's controller to its lineup and promises to start selling them shortly. Interestingly, Lite-On will sell the MU X products in retail, which does not happen often with the brand.
Lite-On’s MU X M.2-2280 SSDs will be available in 128 GB and 256 GB configurations targeting inexpensive gaming PCs that need something faster than conventional SATA SSDs, but cannot afford more advanced PCIe 3.0 x4 drives. The Phison PS5008-E8 controller will work with Toshiba’s 256 Gb BiCS3 3D TLC NAND memory, which is gaining traction among SSD makers these days. Phison yet has to disclose all the details about its PS5008-E8, but what we do know is that the controller is compliant with the NVMe 1.2 specification, uses a PCIe 3.0 x2 interface, features four NAND channels, a DDR3/DDR3L DRAM buffer, Phison’s proprietary error correction algorithm called StrongECC, the company’s own RAID ECC algorithm to recover user data from uncorrectable errors, AES-256 encryption, and so on. The key selling point of the PS5008-E8 is not extreme performance, but affordability and reliability.

Indeed, the Lite-On MUX 256 GB offers up to 1600 MB/s sequential read speed as well as up to 850 MB/s sequential write speed (with pSLC enabled), which does not seem like a record-breaking performance, but which is in line with contemporary entry-level PCIe 3.0 x4 SSDs. When it comes to random performance, the MUX 256 GB is capable of 140K/145K random read/write IOPS, which is ahead of many PCIe 3.0 x4 drives. As for the 128 GB model, its write speed is speced at up to 1500 MB/s, but its read speed is 450 MB/s, which is comparable with modern SATA SSDs. As for reliability, Lite-On rates its MU X drives for 1.5 million hours MTBF and covers them with a three-year warranty.

Lite-On MU X Specifications
Capacity	128 GB	256 GB
Model Number	PP3-8D128	PP3-8D256
Controller	Phison PS5008-E8
NAND Flash	Toshiba's 256 Gb 64-layer BiCS3 3D TLC
Form-Factor, Interface	M.2-2280, PCIe 3.0 x2, NVMe 1.2
Sequential Read	1500 MB/s	1600 MB/s
Sequential Write	450 MB/s	850 MB/s
Random Read IOPS	91K IOPS	145K IOPS
Random Write IOPS	110K IOPS	140K IOPS
Pseudo-SLC Caching	Supported
DRAM Buffer	Yes, capacity unknown
TCG Opal Encryption	Unknown
Power Management	Unknown
Warranty	3 years
MTBF	1,500,000 hours
MSRP	?	?

Lite-On produces most of its drives for various PC and server OEMs, to address consumer markets, the company uses the Plextor brand. OEM orientation of Lite-On’s SSD business is a primary reason why the company uses SSD controllers from multiple developers (including Marvell, Silicon Motion, and Phison) — different clients have different requirements when it comes to performance and pricing. Some of Lite-On’s drives use in-house design, other are based on reference designs from respective developers of controllers. When it comes to the MU X, the drive uses a design developed by Phison, which has passed Lite-On’s quality checks.
Lite-On’s MU X drives will hit store shelves in the coming weeks. MSRPs of the drives are currently unknown, but since affordability is a key feature of the controller, expect the actual SSDs to be priced accordingly.
Buy Plextor M8Pe 256 GB SSD on Amazon.com
Related Reading:

More...

**FEED** · 12-07-17, 01:32 PM

This week, EVGA launched another model for their GeForce GTX 1070 Ti family: the FTW Ultra Silent, featuring the ACX 3.0 cooling system. Starting from its triple slot wide PCIe bracket, the card has basically more of everything from the lower-end SC Black Edition: another 8 pin PCIe connector, 10+2 power phases, and 235W power draw. As a mix between the ACX 3.0 equipped SC Black Edition and iCX FTW2, the FTW Ultra Silent is a little over 17mm thicker than the SC Black Edition due to a heftier heatsink. In turn, that excess cooling capability permits the FTW Ultra Silent a shot at living up to its name, while also allowing for overclocking headroom.

Advertised at 1607+/1683+ MHz clockspeeds, EVGA is referring to their Precision XOC overclocking utility, as well as the Precision XOC 'OC Scanner X' feature, in light of the GTX 1070 Ti’s requirement to ship at reference clocks. The OC Scanner X functionality detects artifacting and tests the card for what it believes to be the “optimal” overclock, and then applies it; for gamers who are not comfortable with overclocking, automatic overclocking brings GTX 1070 Ti’s a step closer to actual factory overclocks. Though like with all automatic overclock features, manual tweaking is likely more precise and tuned for a given end-user.
Meanwhile, the FTW Ultra Silent picks up a couple more high-end board features with 2 BIOS chip and a backplate, though it misses out on the king of all premium graphics card attributes – RGB LEDs. The FTW Ultra Silent comes with a white LED logo like the rest of the EVGA GTX 1070 Ti family, while the FTW2 remains the only one with RGB LED capability.

Selected EVGA GeForce GTX 1070 Ti Models
		EVGA GTX 1070 Ti FTW Ultra Silent	EVGA GTX 1070 Ti FTW2	EVGA GTX 1070 Ti SC Black Edition
Boost Clock		1607+ MHz
Base Clock		1683+ MHz
VRAM	Clock / Type	8Gbps GDDR5
Capacity	8 GB
Bus Width	256-bit
TDP		235W	235W	217W
Outputs		3x DP1.4, 1x HDMI2.0, 1x DVI-D
Power Connectors		2 x 8pin	2 x 8pin	1 x 8pin
Power Phases		10+2	10+2	5+1
Length		267mm	267mm	267mm
Bracket Width		3 Slot	2 Slot	2 Slot
Height		129mm	129mm	111.2mm
Cooler Type		Open Air (ACX 3.0)	Open Air (iCX)	Open Air (ACX 3.0)
Price		$499	$499	$469

The FTW Ultra Silent does share the $500 price with the FTW2, essentially at the price ceiling for GTX 1070 Ti, but trades off the FTW2 and iCX extras for a larger heatsink. And despite the wider three slot PCIe bracket in supporting that heatsink, the FTW Ultra Silent has the same display output configuration as the FTW2: 1 x DVI-D, 3x DisplayPort, & 1x HDMI (maximum 4 monitors supported). The FTW Ultra Silent also comes with the eponymous Ultra Silent fan profile, but details thereof were not specified.
At this time, the GTX 1070 Ti FTW Ultra Silent is only available at the EVGA store for $499.99, and appears to be in stock.
Buy EVGA GTX 1070 Ti FTW2 on Amazon.com
Buy EVGA GTX 1070Ti SC Hybrid on Amazon.com
Gallery: EVGA GTX 1070 Ti FTW Ultra Silent

More...

**FEED** · 12-07-17, 03:59 PM

Adding on to the recent slew of custom RX Vega announcements, last week Gigabyte announced the Radeon RX Vega 64 Gaming OC 8G, featuring their dual-fan Windforce cooling system. The 2.5 slot width card comes with a very mild factory overclock of 1560MHz boost and 1276MHz base, but otherwise brings the usual amenities of high-end custom designs: 12+1 power phases, backplate with copper slug and heatpipe, and zero fanspeed idle under their “3D Active Fan” branding, where fans turn off under lower temperatures or loads.

The rest of the Windforce 2X design is fairly standard for Gigabyte’s high-end models. Underneath the two 100mm alternate spinning fans, the aluminum heatsink’s five copper heatpipes make direct contact with the GPU, and presumably the HBM2 as well. On the rear, the metal backplate features a copper plate behind the GPU and a copper heatpipe, features that are not typically present in Gigabyte’s budget models. Altogether, the heatsink with rear copper plate and heatpipe complex is an interesting design choice to complement the dual fan design, and ideally would mitigate the lesser cooling potential versus three fans.

Specifications of Selected Radeon RX Vega 64 Cards
		Gigabyte Radeon RX Vega 64 Gaming OC Windforce	AMD Radeon RX Vega 64 Air	AMD Radeon RX Vega 64 Liquid
Base Clock		1276Mhz	1247MHz	1406MHz
Boost Clock		1560MHz	1546MHz	1677MHz
VRAM	Clock / Type	1.89Gbps HBM2	1.89Gbps HBM2	1.89Gbps HBM2
Capacity	8GB	8GB	8GB
Bus Width	2048-bit	2048-bit	2048-bit
Board Power		TBA	345W	295W
Dimensions		274 x 132 x 53 mm	272 x 112 x 40mm	272 x 112 x 40mm
Power Connectors		2 x 8pin	2 x 8pin	2 x 8pin
Outputs		3 x HDMI 2.0 3 x DP 1.4	1 x HDMI 2.0 3 x DP 1.4	1 x HDMI 2.0 3 x DP 1.4
Process		GloFo 14nm	GloFo 14nm	GloFo 14nm
Launch Price		TBA	$699 (Liquid Pack)	$499

The card also features a 3x HDMI & 3x DisplayPort output configuration, leaving it suitable for VR and 6-display setups. While the card has 2x 8pin PCIe power connectors, TBP was not disclosed, and the specifications simply recommend a 750W PSU.
While Gigabyte’s announcement states that they have released the product, the RX Vega 64 Gaming OC is yet to be found on sale. Gigabyte did comment that a custom Radeon RX Vega 56 featuring the same Windforce 2X cooler will be released in early December, presumably in time for the holiday season.
Gallery: Gigabyte Radeon RX Vega 64 Gaming OC Windforce 2X

Selected Upcoming Radeon RX Vega Custom Boards
Card	Clocks	Availability
ASUS RX Vega 64 ROG Strix OC	1405 MHz (Base) 1590 MHz (Boost)	TBA
ASUS RX Vega 56 ROG Strix OC	1156+ MHz (Base) 1471+ MHz (Boost)	TBA
GIGABYTE RX Vega 64 Gaming OC (Windforce 2X)	1278 MHz (Base) 1560 MHz (Boost)	December
GIGABYTE RX Vega 56 Gaming OC (Windforce 2X)	1156+ MHz (Base) 1471+ MHz (Boost)	December
PowerColor RX Vega 64 Red Devil	1417 MHz (Base) 1607 MHz (Boost)	December
PowerColor RX Vega 56 Red Devil	1156+ MHz (Base) 1526 MHz (Boost)	December
XFX Radeon RX Vega 64 Double Edition	1247 MHz (Base) 1546 MHz (Boost)	December or later
XFX Radeon RX Vega 56 Double Edition	1156 MHz (Base) 1471 MHz (Boost)	December or later

More...

**FEED** · 12-07-17, 06:11 PM

EKWB is launching a new liquid cooling kit designed to support AMD Radeon RX Vega based video cards, which the company is dubbing the EK Fluid Gaming A240R kit. The kit includes the EK-AC Radeon Vega full-cover block, the Supremacy AX CPU block, the Alustream SE 240 radiator, as well as the Vardar 120 series of fans. Though we have seen similar kits before from the Fluid gaming line, the EK-A240G, this is EK’s first kit engineered to support AMD Vega video cards and is an easier way to jump into custom liquid cooling without the hassle of finding all the parts needed.

The Fluid Gaming A240R kit’s star of the show is the EK-AC Radeon Vega full-cover water block. Designed with AMD, the water block directly cools the GPU, HBM2 memory, as well as the VRM with water flowing directly over those critical parts. The block itself is made out of aluminum while the outside is black and has the Radeon Vega name and symbol on it. Capping off its overall aesthetic is a matte black brushed aluminum backplate giving the whole thing a clean look. EK says the EK-AC Radeon Vega block will fit all reference Radeon Vega Front Edition, RX Vega 64, and RX Vega 56 graphics cards. As always, check the EK Cooling Configurator to ensure compatibility.

The kit also comes with the EK-Supremacy AC water block featuring a black brushed aluminum faceplate which should blend in with most other hardware, just as the EK-AC Radeon Vega block above. It is a universal water block that EK says fits all modern AMD and Intel CPUs. The cold-plate is made from die casted aluminum alloy and uses a micro-channel fin structure to get heat away from the CPU underneath. It comes with standard G 1/4” ports and comes with a pre-assembled tool-less mounting mechanism designed for easy installation.
The EK- Alustream SE 240 radiator is matte black in color matching the other parts in the kit and comes in at a slim 27.5mm thick. EKWB says it has dense aluminum fins and is optimized for maximum heat dissipation across the range of the included Vardar fans. Also included in the kit are all the fittings (EK compression fittings), tubing, and coolant (EK Cryofuel) to get started. Users do need to be aware there is aluminum in this loop and should take heed of this when expanding and introducing a mixed metal situation. EK does go over this HERE saying the purpose made coolant contains all the needed chemicals to protect the parts from corrosion and biological growth.

The EK Fluid Gaming A240R kit and EK-AC Radeon Vega full-cover water block are currently available for pre-order through the EK Fluid Gaming web-shop. Pre-orders will start shipping Friday, December 8^th.

EKWB Fluid Gaming A240R
Name	MSRP (inc. VAT)
EK Fluid Gaming A240R	249.95€ / $239.99
EK-AC Radeon Vega	89.95€ / $89.99

Related Reading:

More...

**FEED** · 12-07-17, 08:15 PM

Toshiba is announcing its first family of helium-filled hard drives today. It includes the industry’s first 14 TB HDD with nine platters featuring perpendicular magnetic recording (PMR) technology without shingled platters. The drives are primarily aimed at operators of cloud and exascale datacenters that need all the capacity they can get. It is a market segment that Toshiba has barely been able to address in the past. In addition, Toshiba also expects to sell its 12 TB and 14 TB HDDs through distributors once it ramps up production and satisfies the demand from key clients.
Toshiba’s First Helium-Filled HDD Platform

Toshiba is introducing its hermetically sealed helium-filled hard drives years after its rivals. However, the platform being used for the new MG07ACA HDDs is the most sophisticated in the industry today. It supports the most number of platters, features up to 18 heads, and also has the highest areal density. Toshiba’s flagship MGA070ACA 14 TB hard drive incorporates nine PMR platters with ~1.56 TB capacity each. By contrast, its competitors have eight 1.5 TB platters. In order to cram nine platters into an industry-standard 3.5” HDD, Toshiba had to use Showa Denko's new offerings that are 0.635 mm thick (down from the 0.8 mm platters in the previous generation drives), shrink the distance between the platters to 1.58 mm and use new heads from TDK featuring its 2^nd generation dual-stage actuator. Toshiba does not say whether it uses 9^th generation or 8^th generation PMR media from Showa Denko for the new HDDs. However, knowledge of the areal density makes the actual formal generation unimportant. Meanwhile, Toshiba’s helium-filled hard drive platform also includes a new cover design with laser welding, Nidec’s motors for helium-filled HDDs, hermetic connectors and other necessary components.
Since the MG07ACA are intended for datacenter racks containing hundreds of vibrating hard drives, they feature numerous enhancements when it comes to reliability and durability. These include top and bottom attached motors, RVFF as well as environmental sensors and so on. Because the new drives shrink the distance between platters to 1.58 mm, it is important to handle vibration effectively. Toshiba rates the new HDDs for 550 TB average annualized workload, 2.5 million hours MTBF and covers them with a standard five-year warranty.

In addition, the new storage devices support Toshiba’s persistent write cache (PWC) with power loss protection (PLP). This is very important for 4K sector drives that emulate 512B sectors. The PWC with PLP feature guards data in case of power loss while performing read-modify-write (RMW) operation to align the source write request with the physical sectors it has to modify. The fact that Toshiba supports RMW in its latest nearline HDDs is important because the tech enables the company to address its clients who run legacy systems with rather extreme capacity points.
12 TB and 14 TB HDDs

Toshiba’s MG07ACA family of 3.5” enterprise-class HDDs will have two models - a 12 TB SKU, and a 14 TB SKU. The flagship 14 TB one uses nine platters with the highest areal density, whereas the 12 TB variant has eight. Both drives feature a 7200 RPM spindle speed, a 256 MB cache buffer, and a SATA 6 Gbps interface. Due to the increased areal density, the flagship drive is able to advertise a 260 MB/s maximum sustained transfer rate, whereas the 12 TB model is slightly slower at around 250 MB/s. Toshiba says the drives can perform up to 167 read IOPS, and, up to 70 write IOPS (4 KB blocks) when subject to random accesses.
Helium has 1/7^th of air's density, and that reduces drag force acting on the spinning platters, while also lowering the fluid flow forces that affect the disks and the heads. This reduces the power consumption substantially, compared to Toshiba's previous-generation enterprise hard drives. The 14 TB variant does not exceed 7.6 W when performing read/write operations and drops to 4.6 W in idle. Operating power of Toshiba’s MG07ACA HDDs is in-line with Seagate’s helium-filled offerings, but is a bit higher when compared to helium HDDs from Western Digital/HGST. Meanwhile, the 4.6 W idle power consumption is the lowest among enterprise-grade high-capacity HDDs. In terms of power consumption efficiency at idle (large hard drives could spend plenty of time idling), the 14 TB MG07ACA is an indisputable champion consuming just 0.32 W per TB.

Brief Specifications of Toshiba's MG07ACA HDDs
Capacity		14 TB	12 TB
P/N	4K Native	MG07ACA14TE	MG07ACA12TE
512e	tbd	tbd
512e SIE	tbd	tbd
Platters		9	8
Heads		18	16
Recording Technology		Perpendicular Magnetic Recording (PMR) / Conventional
RPM		7200 RPM
Interface		SATA 6 Gbps
DRAM Cache		256 MB
Persistent Write Cache		Yes
Helium-Filling		Yes
Sequential Data Transfer Rate (host to/from drive)		260 MB/s	~250 MB/s
MTBF		2.5 million
Rated Annual Workload		550 TB
Acoustics (Seek)		34 dBA
Power Consumption	Random read/write	7.6 W	? W
Idle	4.6 W	? W
Warranty		5 Years

3360 TB Per Rack

Toshiba’s MG07ACA HDDs are designed for operators of cloud and exascale datacenters that demand maximum capacity. The new hard drives are drop-in compatible with existing datacenter infrastructure (because of PMR, SATA, and a relatively low power consumption) and can increase the total storage capacity for a single rack from 2440 TB to 3360 TB, when replacing 10 TB drives with 14 TB HDDs. An increase of storage capacity per rack (and per drive) by 40% can be a major benefit for companies that need to maximize their storage capacity per square meter and per watt, while meeting other TCO objectives.

Servers in a Rackspace datacenter

The most important advantage of Toshiba’s 14 TB MG07ACA hard drive over HGST’s Ultrastar Hs14 is the usage of PMR technology without shingled platters (SMR). Toshiba indicated that many cloud and exascale datacenters are yet to invest in SMR management at the software level, and are unable to use SMR drives effectively. SMR drives have a number of peculiarities when it comes to writing, which have been discussed before.
Q2 Availability

Toshiba plans to start sample shipments of the MG07ACA hard drives in the coming weeks. Since we are dealing with HDDs aimed primarily at cloud datacenters, it will take operators several months to evaluate the storage devices and qualify them for their applications. Some of Toshiba’s clients have apparently tried engineering samples of the MG07ACA HDDs already. Toshiba plans to start volume B2B shipments of the drives three to four months after sampling, sometimes in Q1 or Q2 2018. After the company meets demand of its initial customers, the company will make the 12 TB and 14 TB products available through distributors, which essentially means that they are going to hit Amazon eventually. Pricing is not current available, but since the MG07ACA hard drives belong to the ultra-high-end segment, expect them to be priced accordingly.

What’s Next?

Toshiba has introduced its first helium-filled enterprise HDDs four years after HGST and 1.5 years after Seagate. The launch of the company’s eight-platter 10 TB MG06ACA HDD earlier this year clearly demonstrated that Toshiba extended use of traditional air-filled HDD technology to its maximum. While deserving respect, it also means that the company did not have competitive hard drives for nearline applications for years as its rivals put out helium-filled 8 TB and 10 TB drives in the market. But, with its 1^st generation helium drive, the company is leapfrogging its competitors. So what’s next for Toshiba?

The nine-disk hermetically sealed platform has been in development for years and the 14 TB helium HDD entered Toshiba’s public roadmap in 2016. Along with the 14 TB model, the company included an even higher-capacity 16 TB helium HDD into the roadmap. Obviously, Toshiba does not want to talk about that one today, but it is logical to suspect this drive will feature nine 1.8 TB platters, unless Toshiba manages to squeeze 10 lower-capacity disks into the industry-standard 3.5” form-factor.
When it comes to other HDD manufacturers, we do know that in this industry companies tend to catch up with each other in terms of technologies they use to offer certain capacities. Therefore, the question is when Toshiba’s rivals introduce their nine-platter drives, not if they have such plans. In theory, Seagate and Western Digital could extend usage of 8^th generation PMR platters and launch 14 TB PMR HDDs with nine 1.56 TB disks (assuming that their G8 PMR platters can support such areal density). Alternatively, they could switch to 9^th generation 3.5” PMR platters featuring 1.8 TB capacity and then go with eight-platter 14 TB drives and nine-platter 16 TB models. In both scenarios, nearline PMR HDDs will gain capacity in the coming quarters.
Buy Toshiba X300 8 TB on Amazon.com
Related Reading:

More...

**FEED** · 12-07-17, 11:55 PM

Out of nowhere, NVIDIA has revealed the NVIDIA Titan V today at NIPS 2017, with CEO Jen-Hsun Huang flashing out the card on stage. A mere 7 months after Volta was announced with the Tesla V100 accelerator and the GV100 GPU inside it, NVIDIA continues its breakneck pace by releasing the GV100-powered Titan V, available for sale today. Or in other words, the 815 mm² behemoth die that is GV100 is now available to the broader public.

NVIDIA Compute Accelerator Specification Comparison
	Titan V	Tesla V100 (PCIe)	Tesla P100 (PCIe)	Titan Xp
CUDA Cores	5120	5120	3584	3840
Tensor Cores	640	640	N/A	N/A
Core Clock	1200MHz	?	?	1485MHz
Boost Clock(s)	1455MHz	1370MHz	1300MHz	1582MHz
Memory Clock	1.7Gbps HBM2	1.75Gbps HBM2	1.4Gbps HBM2	11.4Gbps GDDR5X
Memory Bus Width	3072-bit	4096-bit	4096-bit	384-bit
Memory Bandwidth	653GB/sec	900GB/sec	720GB/sec	547GB/sec
VRAM	12GB	16GB	16GB	12GB
L2 Cache	4.5MB	6MB	4MB	3MB
Half Precision	30 TFLOPS?	28 TFLOPS	18.7 TFLOPS	0.19 TFLOPS (1/64 rate)
Single Precision	15 TFLOPS	14 TFLOPS	9.3 TFLOPS	12.1 TFLOPS
Double Precision	7.5 TFLOPS? (1/2 rate)	7 TFLOPS (1/2 rate)	4.7 TFLOPS (1/2 rate)	0.38 TFLOPS (1/32 rate)
Tensor Performance (Deep Learning)	110 TFLOPS	112 TFLOPS	N/A	N/A
GPU	GV100 (815mm2)	GV100 (815mm2)	GP100 (610mm2)	GP102 (471mm2)
Transistor Count	21.1B	21.1B	15.3B	12B
TDP	250W	250W	250W	250W
Form Factor	PCIe	PCIe	PCIe	PCIe
Cooling	Active	Passive	Passive	Active
Manufacturing Process	TSMC 12nm FFN	TSMC 12nm FFN	TSMC 16nm FinFET	TSMC 16nm FinFET
Architecture	Volta	Volta	Pascal	Pascal
Launch Date	12/07/2017	Q3'17	Q4'16	04/07/2017
Price	$2999	~$10000	~$6000	$1299

For the card itself, it unsurprisingly features a vapor chamber cooler with copper heatsink and 16 power phases, all for the 250W TDP that has become standard with the single GPU Titan models. Output-wise, the Titan V brings 3 DisplayPorts and 1 HDMI connector. And as for card-to-card communication, the board does not seem to have SLI connections, instead appearing to have NVLink connections on the top of the PCB, though mostly covered by the shroud and so presumably disabled.
NVIDIA is unsurprisingly pushing this as a compute accelerator card, especially considering that Titan V features tensor cores. But there are those of us who know better than to assume people won’t drop $3000 to use the latest Titan card for gaming. To that end, no gaming performance information has been disclosed.

If the golden shroud didn’t already suggest so, the Titan V is also carving out a new eye-watering price point, dropping in at $2999 and on sale now at the NVIDIA store. As NVIDIA’s new flagship video card, this succeeds the $1200 Pascal Titan X announced around 16 months ago. This release also marks NVIDIA's first HBM2-equipped consumer/prosumer graphics card, matching the 12GB capacity of the Pascal Titan X.
Given that this was a surprise announcement without advance notice, watch this space for more details as the evening progresses.
Gallery: NVIDIA Titan V

More...

**FEED** · 12-08-17, 03:43 PM

Lite-On has added its new EPX-series enterprise-class SSDs to its lineup. Based around 3D TLC NAND memory and in the uncommon M.2-22110 form-factor, the new drives are aimed at read-intensive applications and come with a host of enterprise-class features to maximize performance, data integrity and overall reliability. At present, Lite-On is testing the EPX SSDs internally and will start their sample shipments in the coming weeks or months.
The Lite-On EPX SSDs use Toshiba’s 256 Gb 64-layer BiCS3 3D TLC NAND flash memory chips and are to be available in 960 GB as well as 1920 GB configurations. Lite-On picked up Marvell's proven 88SS1093 "Eldora" controller for the drive, possibly because it brings together high performance (it has three Cortex R5 cores, 8 NAND channels with 4 CE per channel), Marvell’s third-gen ECC technology based on the LDPC algorithm as well as various enterprise-class features.
When it comes to performance, the Lite-On EPX drives offer up to 1700 - 1800 MB/s sustained sequential read speed and around 670 - 800 MB/s sustained sequential write speed, depending on the model (see the table below for details). Enterprise drives do not use pSLC caching, so the stated write performance is available at all times, not only while the cache lasts. Meanwhile, since Lite-On aims the EPX family at read-intensive applications, their relatively low sustained sequential write performance is justified. As for random performance, Lite-On specs the drives for up to 37K – 40K sustained random read speed with QD=4 as well as 90K – 105K sustained random write speed with QD=4. Keep in mind that we are talking about sustained performance numbers that are available at all times, not peak performance figures that manufacturers usually publish for client SSDs. The power consumption of the Lite-On EPX SSDs will be up to 8.25 W, which in line with other products based on the same controller and memory.

One of the interesting performance-improving features of the LiteOn EPX is the Intelligent Read Retry (IRR) technology that relies on machine learning to lower read latencies. Normally, when an SSD controller fails to read data from a cell, it starts to retry reading using various threshold voltage settings, starting from the lowest setting to the highest setting in a sequential fashion. Lite-On’s IRR takes into account multiple parameters (such as the number of P/E cycles endured by the cell, temperature, location, etc.) and then applies threshold voltage that will most likely recover the data, thus shrinking latency. Lite-On formally introduced its IRR technology at FMS2017 and the EPX drives are the first to use it. Lite-On says that in the worst case scenario of read retry, IRR can retain 47% of the original read performance using machine learning compared to 9% of original read performance using traditional read retry.
Like many other enterprise-class SSDs, the Lite-On EPX comes with power loss protection capacitors and supports end-to-end data path protection compatible with the T10 Data Integrity Field (DIF) spec. In addition, the EPX drives from Lite-On feature internal RAID (Redundant Array of Independent Flash Elements, RAIFE) to protect data with parity and thus recover it in cases when regular ECC recovery does not work.

Specifications of Lite-On EPX NVMe SSDs
		960 GB		1920 TB
Model		EPX-KB 960		EPX-KB 1920
Form Factor		M.2-2280-22110
Controller		Marvell 88SS1093
NAND		Toshiba's 3D TLC NAND, 64-layers, 256 Gb
Interface		PCIe 3.0 x4
Protocol		NVMe 1.2
DRAM		Yes (capacity unknown)
Sustained Sequential Read		1700 MB/s		1800 MB/s
Sustained Sequential Write		670 MB/s		800 MB/s
Sustained 4KB Random Read	QD=4	37K		40K
QD=32	165K		180K
QD=256	175L		190K
Sustained 4KB Random Write	QD=4	90K		105K
QD=32	96K		110K
QD=256	120K		140K
Power Consumption	Idle	2.4 W
Active	8.25 W
UBER		1 error in 10¹⁷ bits transferred
End-to-End Protection		Supported, compatible with T10 Data Integrity Field (DIF)
Power Loss Protection		Supported
Temperature	Operating	0 to 70?C
Non-Operating	-40 to 85?C
MTBF		2 million hours
Launch Date		Q4 2017

At present, Lite-On is still polishing off the EPX SSDs internally and will ship it to customers among server OEMs in the coming weeks or months. The company will be among the first independent makers of SSDs to use Toshiba’s 256 Gb 64-layer BICS 3D TLC NAND for enterprise-grade SSDs, so it is not surprising that the company wants to ensure that the drives perform as expected.
Lite-On specs the EPX SSDs for two million hours MTBF and covers the drives with its standard three-year warranty. Pricing of the EPX SSDs will be determined by agreements with particular OEMs and will depend on volumes and some other factors.
Related Reading:

Source: Lite-On (via TechPowerUp)

More...

**FEED** · 12-08-17, 04:55 PM

In the mood for some free hardware? Well then you're in luck: our awesome community team has collected a few of AMD’s Ryzen CPUs and put together a giveaway for this weekend. Altogether we’re giving away one each of the six-core Ryzen 5 1600X, the four-core Ryzen 5 1500X, and the four-core Ryzen 3 1300X.

Ryzen Processor Giveaway Prizes
	Cores/ Threads	Base/ Turbo	XFR	L3	TDP
Ryzen 5 1600X	6/12	3.6/4.0	+100	16 MB	95 W
Ryzen 5 1500X	4/8	3.5/3.7	+200	16 MB	65 W
Ryzen 3 1300X	4/4	3.5/3.7	+200	8 MB	65 W

Needing no introduction, AMD’s Ryzen CPUs have reinvigorated the company over the past several months. Offering a massive leap in performance over AMD’s prior generation designs, AMD’s Zen CPU architecture is now much more competitive with Intel’s Core CPU architecture. Coupled with AMD’s aggressive pricing for their latest processors, it has significantly shaken up the desktop CPU landscape and has made 4+ core CPUs the de facto baseline for all but the most budget desktop builds. In short, the market is now awash in great CPU choices, a very welcome change from the half-decade before.
The giveaway is running through December 11th and is open to all US residents. You can find the full details for entering the giveaway over on the CPU section of our forums.

More...

Thread: Anandtech News

Thread Tools

Search Thread

Anandtech: Dell EMC Launches PowerEdge XR2 Rugged Server: 1U, 44 Cores, 512 GB RAM, 3

Anandtech: Qualcomm Announces Snapdragon 845 Mobile Platform: Tocks Next-Gen CPU Core

Anandtech: Lite-On Adds MU X SSD to Lineup: Phison PS5008-E8, 3D TLC, Up to 1600 MB/s

Anandtech: EVGA Releases GeForce GTX 1070 Ti FTW Ultra Silent: 3 Slot Bracket, ACX 3.

Anandtech: Gigabyte Unveils Radeon RX Vega 64 Gaming OC Windforce 2X, To Release RX V

Anandtech: EKWB Launches EK Fluid Gaming A240R Kit: Full-Cover Vega GPU Cooling & Sup

Anandtech: Toshiba Announces 14 TB PMR MG07ACA HDD: 9 Platters, Helium-Filled, 260 MB

Anandtech: NVIDIA Announces “NVIDIA Titan V" Video Card: GV100 for $3000, On Sale Now

Anandtech: Lite-On Announces Enterprise EPX M.2 NVMe SSDs: 3D TLC, Up to 1920 GB, 180

Anandtech: AMD Ryzen CPU Giveaway: Win A Ryzen 5 1600X, Ryzen 5 1500X, or Ryzen 3 130

Thread Information

Users Browsing this Thread

Tags for this Thread

Bookmarks

Bookmarks

Posting Permissions

TF2 Blog: RGL.gg Season 15 LAN Playoffs

Anandtech News

Anandtech News