Anandtech News

**FEED** · 11-10-16, 07:25 AM

It has been many years since we last had a review of a PSU from Zalman, even though they currently are offering dozens of models. In this review we are having a look at the newest and most powerful addition to their EBT series, the ZM1200-EBT 1200W unit.

More...

**FEED** · 11-10-16, 09:24 AM

BenQ this week introduced its new display aimed specifically at professional gamers. The new monitor belongs to the ZOWIE family of BenQ’s products designed for e-Sports and features Full-HD (1920x1080) resolution as well as a refresh rate as high as 240 Hz. In addition to support for ultra-high refresh rates, the display also has other features developed specifically for gamers.
The BenQ ZOWIE XL2540 monitor has a 24.5” TN panel with a typical resolution for this size (1920×1080), support for 16.7 million (6-bit + FRC) colors as well as a typical contrast ratio for mainstream screens (1000:1). The monitor stands out when it comes to its brightness - 400 cd/m2 (400 nits) is a bit higher compared to what we usually see on TN-based devices and a refresh rate of 240 Hz to ensure as fast gameplay as possible. It is noteworthy that despite positioning, the ZOWIE ZL2540 does not support any dynamic refresh rate technology (either G-Sync, FreeSync or Adaptive-Sync).

When it comes to connectivity, the ZOWIE XL2540 excels many products on the market as it can use DisplayPort 1.2, HDMI and DVI-DL (naturally, DVI does not support 240 Hz refresh rate) to connect to host PCs (nowadays many monitors are only equipped with DP and HDMI inputs). Also, it has a three-port USB hub. The display also has a built-in 55 W power supply.

BenQ ZOWIE XL2540
		ZOWIE XL2540
Panel		24.5" TN
Native Resolution		1920 × 1080
Maximum Refresh Rate		240 Hz
Response Time		1 ms (gray-to-gray)
Brightness		400 cd/m²
Contrast		1000:1
Viewing Angles		160°/160° horizontal/vertical
Inputs		1 × DP 1.2 1 × HDMI 2.0 1 × DVI-D DL
USB Hub		3-port USB hub
Audio		audio in/out ports
Power Consumption	Typical	unknown
Maximum	55 W

In addition to ultra-high refresh rates as well as improved brightness, the ZOWIE XL2540 supports several features specifically for e-Sports professionals: the Black eQualizer that increases brightness of dark areas without oversaturating the bright areas, an option to quickly increase color vibrancies, a special external controller to rapidly activate different settings and profiles as well as a light-shielding hood (which BenQ calls a way to help gamers to focus on their games).

The manufacturer does not announce pricing as well as availability timeframe for its ZOWIE XL2540 just yet. The only thing that BenQ tells us at the moment is that the display will be demonstrated at trade shows from late November. Since BenQ does not produce its own panels, the ETA of the ZOWIE XL2540 depends completely on the company’s suppliers. Moreover, when the panels are available, the display will have to compete against similar products by other makers, such as the ASUS PG258Q (which has G-Sync) which has similar specifications, perhaps indicating it is the same panel.
Related Reading:

More...

**FEED** · 11-10-16, 11:18 AM

This week AOC has introduced its new AGON-branded curved display for gamers that boasts a 200 Hz refresh rate. The ultra-wide AG352QCX monitor has an MVA panel and a pretty high contrast ratio and a WFHD resolution. In addition, the AG352QCX supports AMD’s FreeSync and VESA’s Adaptive-Sync technologies.
The AOC AGON AG352QCX is based on a 35” MVA panel with 2560×1080 resolution and 21:9 aspect ratio. General specifications of the display are fairly typical: 300 nits brightness, 2000:1 contrast ratio, 178°/178° viewing angles, 16.7 million colors and a 4 ms response time (grey-to-grey) and so on. Meanwhile, the key selling points of the monitor are its 2000R curvature (which is rare, typically we see 1600R/3000R), support for a 200 Hz refresh rate as well as support for AMD’s FreeSync and VESA’s Adaptive-Sync technologies. The FreeSync works in a range between 30 and 200 Hz, thus supporting low framerate compensation feature (LFC) and offering smooth gameplay across different genres.

In a bid to appeal to gamers with different PCs, the AGON AG352QCX supports five types of input technologies, including DisplayPort 1.2a, HDMI 2.0, MHL, DVI and D-Sub connectors. To take advantage of the full 200 Hz refresh rate and range via FreeSync, the DP 1.2a input should be used. In other cases, only 50-146 Hz scanning frequencies are supported.

AOC's AGON Curved Display with a 200 Hz Refresh Rate
	AGON AG352QCX
Panel	35" MVA
Native Resolution	2560 × 1080
Refresh Rate Range	30-200 Hz (DP) 50-146 Hz (Other)
Response Time	4 ms
Brightness	300 cd/m²
Contrast	2000:1
Viewing Angles	178°/178° horizontal/vertical
Curvature	2000R
Pixel Pitch	0.32 mm
Inputs	1 × DP 1.2 1 × HDMI 2.0 1 × DVI 1 × D-Sub
USB Hub	2-port USB 3.0 hub, one port supports fast charging
Audio	5 W × 2 audio in/out ports
Power Consumption	Up to 60 W

What is a bit surprising is that the AGON AG352QCX also has two built-in 5 W speakers, which is not common for gaming monitors. Typically the justification here is that who like to play games usually use standalone speakers or headphones, however perhaps AOC is targeting a wider market.

While AOC’s AG352QCX display is listed at the company’s website, the manufacturer does not reveal when the product is set to be available as well as its price. The listing itself naturally indicates that the specifications of the monitor have been finalized and are not going to change significantly. However, it does not indicate mass production or actual shipments. We will probably see it pop up at the CES trade show in January.
Gallery: AOC Launches AG352QCX: 35-Inch 200 Hz Curved MVA Display with Adaptive-Sync

Related Reading:

More...

**FEED** · 11-10-16, 09:02 PM

Today NVIDIA announced their fiscal earnings for the third quarter of their 2017 fiscal year (and yes, that’s not a typo). NVIDIA has cracked the $2 billion mark for quarterly revenue, with earnings of $2.004 billion this quarter. This is up an impressive 54% compared to the same quarter last year. This record revenue follows a previous record for Q2, and NVIDIA is forecasting revenue for next quarter of $2.1 billion, plus or minus two percent. With this rate of growth, they will easily surpass their best yearly earnings ever which was $5.01 billion announced in January 2016. They are on pace for close to $7 billion for the year, which is a dramatic increase at a time when the core PC market is in decline. Gross margin for this quarter was up 2.7% year-over-year to 59.0%, and operating income was up 161% to $639 million. Net income for the quarter was $542 million, up 120% over last year, and earnings per share came in at $0.83, up 89% year-over-year.

NVIDIA Q3 2017 Financial Results (GAAP)
	Q3'2017	Q2'2017	Q3'2016	Q/Q	Y/Y
Revenue (in millions USD)	$2004	$1428	$1305	+40.3%	+53.6%
Gross Margin	59.0%	57.9%	56.3%	+1.1%	+2.7%
Operating Income (in millions USD)	$639	$317	$245	+102%	+161%
Net Income	$542	$261	$246	+108%	+120%
EPS	$0.83	$0.41	$0.44	+102%	+89%

NVIDIA has done a decent job diversifying itself, but it’s core market is still consumer GPU sales. With the launch of Pascal, NVIDIA has maintained a very healthy performance lead, and that has certainly translated into sales. Gaming related sales for this quarter came in at a staggering $1.244 billion, up 63.5% compared to Q3 2016. NVIDIA gained more revenue in the gaming segment this quarter than all of AMD’s Computing and Graphics segment earned in their last quarter. Clearly there was some pent-up demand for the FinFET based GPUs after being on 28 nm for so long. Professional Visualization had a good quarter as well, with revenues of $207 million. That is a 9% increase year-over-year, although they did have a 3% drop in revenue compared to their previous quarter.
With the rise in GPU compute in the datacenter, NVIDIA has aggressively pursued this market, and it is starting to make some real inroads here as well. For Q3 2017, NVIDIA had revenue of $240 million in their Datacenter group, which is up from $82 million a year ago, or a 193% increase in revenue year-over-year. Datacenter revenue is now the second largest segment for NVIDIA, and only a year ago it was half of the professional GPU revenue.
Automotive continues to be a solid performer for NVIDIA too, and they’ve released quite a few SoCs specifically for this market. NVIDIA announced that it has an agreement to supply Tesla Motors the DRIVE PX 2 platform to power a new AutoPilot system in the Model S, X, and upcoming 3. With a maximum TDP of 250 Watts, DRIVE PX 2 casts off the mobile SoC constraints of the previous design, despite still having NVIDIA’s Denver CPU paired with Cortex-A57. With the continued investment in this sector, NVIDIA has seen strong growth here, and revenue for this quarter was $127 million, which is up from $79 million a year ago, or a gain of 60%.
Finally, OEM and IP had earnings of $186 million for the quarter, which is down slightly from the $193 million they earned a year ago, but with the recent announcement that they will power the Nintendo Switch, perhaps they will see some growth here in the future.

NVIDIA Quarterly Revenue Comparison (GAAP)
In millions	Q3'2017	Q2'2017	Q3'2016	Q/Q	Y/Y
Gaming	$1244	$781	$761	+59.3%	+63.5%
Professional Visualization	$207	$214	$190	-3.3%	+8.9%
Datacenter	$240	$151	$82	+58.9%	+192.7%
Automotive	$127	$119	$79	+6.7%	+60.8%
OEM & IP	$186	$163	$193	+14.1%	-3.6%

This entire year has been very good for NVIDIA, but this quarter in particular has seen some very strong growth. The best news for NVIDIA and their investors is that they are not just seeing growth in their diversified markets, although they are doing very well there, but the growth in their core market was incredibly strong as well.
Source: NVIDIA Investor Relations

More...

**FEED** · 11-11-16, 06:46 AM

Early this morning HTC took the wraps off of a new phone that will be exclusive to Sprint, the Bolt. Not to be confused with the old HTC Thunderbolt, the Bolt is a somewhat less than flagship phone that roughly follows the same path as last year’s HTC One A9, offering another take on a high-end phone design with a slightly different configuration. Coupled with its change in hardware, the Bolt also serves as a network showcase of sorts for Sprint and their latest-generation LTE Plus network.

	HTC One M9	HTC 10	HTC Bolt
SoC	Snapdragon 810 4x Cortex-A57 @ 2Ghz 4x Cortex-A53 @ 1.5GHz Adreno 430 (TSMC 20SoC)	Snapdragon 820 2x Kryo @ 2.15GHz 2x Kryo @ 1.6GHz Adreno 530 (Samsung 14LPP)	Snapdragon 810 4x Cortex-A57 @ 2Ghz 4x Cortex-A53 @ 1.5GHz Adreno 430 (TSMC 20SoC)
RAM	3GB LPDDR4	4GB LPDDR4	3GB LPDDR4
NAND	32GB NAND + microSD	32/64GB NAND + microSD	32 NAND + microSD
Display	5” 1080p Super LCD3	5.2” 1440p Super LCD5	5.5” 1440p Super LCD3
Network	2G / 3G / 4G LTE (Category 6/9 LTE)	2G / 3G / 4G LTE (Category 6/9 LTE)	2G / 3G / 4G LTE (Category 6/9 LTE)
Dimensions	144.6 x 69.7 x 9.61mm, 157g	145.9 x 71.9 x 3-9mm, 161g	153.6 x 77.3 x 3.7-8.1mm, 174g
Camera	20MP Rear Facing f/2.2, 1.12µm, 1/2.4" (Toshiba T4KA7)	12MP Rear Facing w/ OIS and laser AF, f/1.8, 1.55µm, 1/2.3" (Sony IMX377)	16MP Rear Facing w/ OIS and PDAF, f/1.7
4MP Front Facing, f/2.0, 2µm (OmniVision OV4688)	5MP Front Facing w/ OIS, f/1.8, 1.34µm (Samsung S5K4E6)	8MP Front Facing
Battery	2840 mAh (10.93 Whr)	3000 mAh (11.55 Whr)	3200 mAh (12.32 Whr)
OS	Android 5 w/ HTC Sense 7	Android 6 w/ HTC Sense	Android 7 w/ HTC Sense
Connectivity	1x1 802.11a/b/g/n/ac, BT 4.1, (BCM4356), USB2.0, GPS/GLONASS, NFC	2x2 802.11a/b/g/n/ac, BT 4.2, (BCM4359) USB-C, USB3.1, GPS/GLONASS (US, JP) GPS/GLONASS/Beidou (EU, Asia) NFC	802.11a/b/g/n/ac, BT 4.1, NFC, USB-C (2.0) GPS/GLONASS
Fingerprint Sensor	N/A	Capacitive	Capacitive
SIM	NanoSIM	NanoSIM	NanoSIM
Launch Price	$649	$699	$600

Starting as always with the specs, while it’s obvious from the announcement that the Bolt is not meant to be a true competitor to the 10, I must admit the choice of hardware surprised me. HTC has opted to go with Qualcomm’s struggling (and at this point relatively old) Snapdragon 810 SoC, a staple of 2015 flagship phones that combined 4 ARM Cortex-A57 cores with 4 ARM Cortex-A53 cores and built on TSMC’s 20nm process. The fact that this phone is meant to be below the HTC 10 rules out Snapdragon 820, but that HTC didn’t use one of Qualcomm’s better received Snapdragon 600 SoCs is unexpected. It also gives us a good idea of what to expect for phone performance, and performance stability.

In terms of physical design, the Bolt is not much different than the 10, and I imagine that was intentional. With a 5.5-inch 2560x1440 display the phone is larger than the HTC 10, but it retains the former’s general curved design and button placement. This gives HTC’s latest phones a common style to them, which for HTC helps them stand apart from the likes of Samsung and the other Android handset vendors. The display itself is one of HTC’s Super LCD3 panels, the same generation as the HTC One M9.
The big news for HTC and Sprint here is of course the cellular connectivity. Backed by the Snapdragon 810’s integrated X10 mode, the Bolt is meant to showcase Sprint’s LTE Plus network. LTE Plus specifically takes advantage of carrier aggregation, with Sprint offering LTE on 800MHz, 1900MHz, and their recently recovered 2.5GHz spectrum. In the case of the Bolt, this allows for aggregating up to 3 20MHz channels over those bands. And though it’s beyond the scope of a news article, I’m curious as to how well 2.5GHz is working in practice for Sprint, given the greater hurdles that come from its reduced ability to penetrate buildings.
Otherwise many of the phone’s features and specifications are consistent with other high-end Android phones, including a fingerprint sensor and USB Type-C connectivity (albeit only 2.0 speeds). Meanwhile, relative to the HTC 10, HTC has gone with a higher resolution rear camera here, utilizing a 16MP sensor versus the HTC 10’s 12MP. Though specific sensor information isn’t available, the overall camera module is on the higher end, incorporating OIS and phase-detection auto focus. Further in the Bolt’s favor, the phone is IP57 rated. This is a rating we don’t see too often, and relative to the more common IP67 means that it has the same degree of water resistance as some other phones, but lower dust resistance, merely being protected against dust, but not completely sealed against it.
Powering the phone is a size-appropriate 3200mAh battery, which means we’re looking at a 7% capacity increase versus the battery found in the 5.2-inch HTC 10. And as this is a Snapdragon 810-based phone, it supports Qualcomm’s Quick Charge 2.0 rapid charging technology. However on the other hand, inexplicably missing is a 3.5mm headphone jack; despite the fact that this phone is larger than HTC’s other phones, the company has gone the Apple route and removed it. Instead HTC offers a set of USB-C headphones, otherwise you’ll need to track down a USB-C audio adapter. On the plus side, however, the phone does feature a microSD card slot, for users who find the single 32GB configuration to be too small.
Finally, the Bolt is the first HTC phone that will ship with Android 7. As usual, HTC is applying their Sense UI to Android, which has itself been upgraded to keep up with the latest release of Android.
Wrapping things up, as a Sprint exclusive, Sprint will of course be handling distribution. The carrier is listing the phone at $600 ($25 x 24 months), and while the Bolt is not being positioned as a flagship-level phone, the pricing unfortunately isn’t going to be too far off. As we noted in yesterday’s buyer’s guide, you can already find the HTC 10 for $549, which makes the Bolt the de facto more expensive phone, a somewhat awkward position, all things considered.
Gallery: HTC Bolt

More...

**FEED** · 11-11-16, 08:33 AM

ASUS this month finally began to sell its high-end ROG GT51 desktop in the U.S. The system comes equipped with a factory-overclocked Intel Core i7 processor, two GeForce GTX graphics cards in SLI mode as well as two NVMe PCIe SSDs in RAID 0 to maximize performance. The fully configured ROG GT51 system costs $4999 and marks ASUS’ entry on the market of high-performance desktop PCs.
ASUSTeK Computer was established in 1989 with the purpose to develop and build computer motherboards. The company quickly became famous for its mainboards and by the early 2000s it was the world’s No. 1 producer of mainboards and a leading maker of PC components in general. Starting from 2002~2003, ASUS began to diversify its business and started to make products like notebooks and various peripherals, and in 2006 the company established its Republic of Gamers (ROG) sub-brand for computer components designed specifically for gamers. Eventually, ASUS came up with high-end ROG laptops as well as gaming displays and even became the largest supplier of the aforementioned product categories (at least, according to its own statements). However, despite the fact that ROG-branded motherboards have been used by performance enthusiasts and a variety of boutique PC makers for their builds for a while, ASUS itself did not actively engage into desktop PC market until circa 2014, when it introduced its ROG G20 SFF PC and ROG GR8 console based on custom form factor designs. Over the last couple of years, ASUS has showcased various region specific builds, but the GT51 is the company’s first gaming desktop build to challenge high-end machines from well-known brands. It took the company quite a while to finalize the design and it is finally available for purchase.

The ASUS ROG GT51CA is based on Intel’s Core i7-6700K CPU that can be overclocked in one click to 4.60 GHz as well as a proprietary Intel Z170-based motherboard. The processor is cooled using an ROG-branded closed-loop liquid-cooling system, which fan also acts like an exhaust fan for the whole PC. The system comes equipped with two ASUS GeForce GTX 1080 graphics cards running in SLI mode, 64 GB of DDR4-2800 memory (4×16 GB), two Samsung SM951 512 GB M.2 SSDs operating in RAID 0 mode and with a 1 TB HDD (which has 7200 RPM spindle speed). Since we are talking about a serious gaming PC that has to compete against products from companies like Alienware and Origin PC, the ROG GT51CA can be further customized by their owners, but not built-to-order by retailers. For example, the end users may install faster memory, up to five 3.5”/2.5” HDDs or SSDs, different M.2 drives and so on.

Specifications of ASUS ROG GT51
	ASUS ROG GT51CA
CPU	Intel Core i7-6700K at 4.6 GHz
PCH	Intel Z170
Motherboard	ASUS SKU-Specific Custom
Graphics	Up to two NVIDIA GeForce GTX 1080
Cooling	ASUS closed-loop liquid cooling with 120 mm radiator
Memory	Four DIMM slots, 64 GB of DDR4-2800
M.2	Two Samsung SM951 512 GB in RAID 0 Two M.2 (PCIe 3.0 x4) slots for up to two M.2 2280 SSDs
SATA connectors	6
Storage bays	5×3.5"/2.5" 1×5.25"
Wireless	2×2 802.11ac + BT 4.0 Optional NFC
Ethernet	Gigabit Ethernet (Intel I219-V ?)
Display Outputs	HDMI/DP/DVI - according to GPU specs
Audio	Five 7.1-channel audio connectors (Realtek ALC1150?) a mini-jack for headphones a mini-jack for microphone Optical S/PDIF
USB	8 × USB 3.0 Type-A (5 Gbps) 1 × USB 3.1 Type-A (10 Gbps) 1 × USB 3.1 Type-C (10 Gbps) 2 × USB 2.0
Dimensions	20" (H) x 16" (W) x 20" (D) 50.8 cm (H) x 40.6 cm (W) x 50.8 cm (D)
PSU	700 W PSU by ASUS
Visual Aesthetics	Custom lighting: - Four lighting zones - Eight color modes - Motherboard lighting
Price	$4999
NB!	Exact specifications may vary as ASUS and partners may offer different configurations.

As for connectivity, the ASUS ROG GT51CA has gigabit ethernet, 802.11ac 8x2 Wi-Fi, multiple USB 3.0 Type-A ports (including two on the front), one USB 3.1 Type-A (10 Gbps) and one USB 3.1 Type-C (10 Gbps) port (both on the front). The audio of the PC clearly deserves a mention because ASUS has installed an audiophile-grade ESS Sabre headphone DAC and amplifier to improve audio quality to the front audio outputs. Meanwhile, for some reason the company decided not to equip its top-of-the-range system with one of its Xonar discrete audio cards and the ROG GT51CA still relies on a 7.1-channel integrated audio solution (which is typically an improved ALC1150 audio solution from ASUS).

Yet another exclusive capability of the ROG GT51CA is the ROG Band, which can be used to access a hidden hard drive space as well as the overclocking mode.

Like systems from boutique PC makers, the ASUS ROG GT51CA has custom dynamic LED lighting effects (eight million colors, four zones) and comes with a mechanical keyboard and a gaming grade mouse with matching design. Unfortunately, at this time ASUS does not offer aerography for its machine.
The ASUS ROG GT51CA premium desktop is available from Amazon, B&H Photo and some other retailers.
Gallery: ASUS ROG GT51 Now Available: Core i7-6700K at 4.6 GHz, GTX 1080 SLI, 64 GB DDR4, $4999

Related Reading:

More...

**FEED** · 11-14-16, 08:03 AM

The market for network-attached storage units has expanded significantly over the last few years. The rapid growth in public cloud storage (Dropbox, OneDrive, Google Drive and the like) has tempered the expansion a bit amongst consumers who are not very tech-savvy. However, the benefits provided by a NAS in the local network are undeniable, particularly when complemented with public cloud services. Enterprise users obviously need NAS units with different performance and feature requirements. Our previous NAS reviews have focused more on the performance aspect. With feature set and ease of use becoming important across all market segments, we believe that a qualitative evaluation of the different commercial NAS operating systems is needed to educate consumers on the options available. In this piece, we will take a look at the core features of some of the popular commercial off-the-shelf (COTS) NAS operating systems.

More...

**FEED** · 11-14-16, 08:03 AM

Kicking off this week is SC, the annual ACM/IEEE sponsored supercomputing conference in the United States. At this show last year, AMD announced the Boltzmann Initiative, their ambitious plan to overhaul their HPC software stack for GPUs. Recognizing that much of NVIDIA’s early and continued success has been due to the quality of the CUDA software ecosystem, AMD set out to create an ecosystem that could rival and even interact with CUDA, in hopes of closing the software gap between the two companies.

Now at SC16, the company is both updating participants on the current state of Boltzmann, along with providing the latest software update to the project. Now going under the name the Radeon Open Compute Platform – ROCm – AMD shipped the initial 1.0 version of the platform back in April of this year. That initial release however was, relative to the complete scope of Boltzmann, only the tip of the iceberg. A good deal of what AMD ultimately wanted to do with Boltzmann was not ready (or at least still in beta) with earlier ROCm releases, and it’s not a coincidence that ROCm 1.3, being announced today at the show, is much closer to completing the Boltzmann Initiative.

So where do things stand for AMD one year later with ROCm 1.3? Both the Heterogeneous Compute Compiler (HCC) and the Heterogeneous-compute Interface for Portability (HIP) have made significant progress. ROCm 1.3 introduces the shipping version of the native compiler, which as AMD previously discussed is based around LLVM. Of all of the parts of the Boltzmann plan, the native compiler is perhaps the most important part, as it and the associated driver/APIs/interfaces are the key to making HPC software work on AMD’s platform.

ROCm 1.3 also introduces several new features to the platform as a whole. Of particular note here, AMD has enabled support for 16bit floating point and integer formats within the platform. As regular readers may recall, AMD has offered native support for 16bit formats since GCN 1.3 (Tonga/Fiji), so this functionality is finally being exposed on ROCm. It should be noted however that this is not the same as packed 16bit formats, which allow for AMD’s 32bit ALUs to process two 16bit instructions at once, gaining FLOPS versus using 32bit instructions. Instead, the immediate purpose of this is to allow developers to leverage the space-saving aspects of 16bit data formats, reducing memory and register pressure by using smaller data types when lower precision is suitable, and cutting down on the number of instructions as well by avoiding unnecessary conversions.

Meanwhile, though AMD won’t comment on it, with the launch of the Playstation 4 Pro, we now know that AMD has an architecture supporting packed 16bit formats. Consequently, it’s widely expected that the next iteration of AMD’s dGPUs, Vega (incorporating GFX IP 9), will support these packed formats. In which case by exposing 16bit data types and instructions now, AMD is laying the groundwork for what will be following in 2017.
Moving on, AMD has now added “OpenCL 1.2+” support to ROCm. AMD’s history with OpenCL has been long and somewhat tortured – AMD’s bet on OpenCL has not been well rewarded – so this one caught my eye. The oddly named “1.2+” support reflects the fact that AMD isn’t implementing a full OpenCL 2.0 stack on top of ROCm, but rather they’re targeting existing OpenCL 1.x users. The end result is that ROCm’s OpenCL support is based around the 1.2 stack, but that it supports a select subset of the OpenCL 2.0 kernel language. The OpenCL runtime itself radically changed with 2.0, due to the introduction of SPIR-V, so how AMD is going about supporting OpenCL is important. They are essentially offering OpenCL 1.x users a souped up version of OpenCL 1.x, backporting several 2.0 features while retaining the 1.2 runtime. It’s a logical choice to offer a path forward to their existing OpenCL customers, and consistent with AMD’s previous sentiment on OpenCL. AMD’s future is in HCC and C++, not OpenCL.

AMD has made good progress on the HIP front as well. HIP itself as it 1.0, and to demonstrate their progress, AMD is touting how easy it was to port the CAFFE deep learning framework from CUDA to HIP. Using the HIPify Tools for automatic conversion, AMD was able to translate 99.6% of the code automatically. This is admittedly less than 100% – AMD has been upfront before that they may always be playing catch-up and won’t necessarily be able to replicate NVIDIA driver features 1-for-1 – but it’s significant progress. Ultimately AMD was able to finish the feature-complete port in less than 4 days for a 55K line project. And words can’t stress how important this is for AMD; they need CUDA interoperability to be able to break into what has become a $240mil/quarter business for NVIDIA, so successful ports with HIP are critical to becoming a viable alternative. Further adding to AMD’s success with CAFFE, the HIPified version of the framework is already faster than the OpenCL version, never mind the fact that the OpenCL version had to be created practically from scratch.
Meanwhile ROCm 1.3 also includes some new hardware functionality as well. AMD Polaris hardware is now supported, including both the Radeon RX 400 series and the Radeon Pro WX series. In the grand scheme of things AMD’s Polaris parts aren’t all that great for HPC due to a lack of total throughput – interestingly, I’m told the Radeon R9 Nano has been the big winner here due to its combination of throughput and best perf-per-watt for a Fiji-based part – and adding Polaris is more about opening the door to developers. Prior to this only Fiji and Hawaii GPUs were supported, so with the addition of Polaris support developers can develop and test against their local machines, and then deploy for production use on Hawaii/Fiji HPC gear.
AMD is also using SC16 to further outline their platform support plans. Along with x86 support – including of course, Zen – AMD is also going to be supporting ARMv8 AArch64, and IBM POWER8. The former is an interesting choice since ARM in the server space has to date been mostly about microservers and other forms of high density computing, while adding POWER8 support is a logical step to take in order be able to play in the same hardware space that NVIDIA does as part of OpenPOWER’s “anyone but Intel” consortium. The addition of POWER8 support also offers an interesting sign of where AMD might go in the future; along with POWER8, they are also supporting the OpenCAPI interface that future POWER CPUs will use. Could this mean that AMD will offer HPC-class GPUs with more than just PCIe support in the future?

Wrapping things up, while ROCm 1.3 brings AMD much closer to fulfilling the original vision of the Boltzmann Initiative, the company still has some work to do and is offering a brief look at their future software plans. Now that AMD has the core platform sorted out, they’re going to be increasing their investment in major libraries and frameworks used by many HPC applications. This includes fundmental math libraries like BLAS and FFT, but also major frameworks like AMBER and the aforementioned CAFFE. Along with presenting a sane, developer-friendly language, one of CUDA’s other great strengths has been libraries/frameworks, so this is where AMD will be catching up next. And while AMD isn’t laying out any specific dates on the hardware or software side, the way the stars are aligning right now, I’m not going to be too surprised if they’ll be ready to finally fully tackle NVIDIA with the 1-2-3 punch of ROCm, optimized software libraries, and Vega GPU hardware by the time we’re reporting on SC17 at this time next year.

Gallery: ROCm 1.3 Press Deck

More...

**FEED** · 11-14-16, 04:03 PM

Over the past week AKiTiO has formally introduced its new external graphics enclosure using a Thunderbolt 3 connection. As with other external graphics enclosures, the AKiTiO Node allows a user to easily upgrade the graphics sub-system of notebook, AIO or SFF PCs that have a Thunderbolt 3 port by using a discrete graphics card. One of the key elements to eGFX is price and look, and AKiTiO has placed the Node at $299, but itcan fit graphics cards with massive coolers.
Enthusiasts started to experiment with external graphics over Thunderbolt 2 interface several years ago after multiple attempts of companies like ATI/AMD, ASUS and some others to develop commercial external GPU technologies had failed. Back in 2014, plugging a video card to a MacBook Pro using Thunderbolt 2 was costly: users had to buy a chassis for up to $1000 (there were cheaper methods, but they did not look solid), a PSU and a graphics board. The homebrew method worked, but had a number of limitations when it came to bandwidth, hot-plugging (more importantly, unplugging), driver support and some others, not to mention the aesthetics of such eGFX setups. Given the trend towards miniaturization of mobile and desktop PCs, the idea of external graphics was far too alluring to skip, which is why Intel decided to support it with its Thunderbolt 3 technology introduced in 2014. While technically TB3 is PCIe 3.0 x4 over a cable, to properly enable operation of external GPUs, Intel and its allies had to ensure support of eGFX hardware though operating systems, GPU drivers, firmware and so on. Earlier this year all pieces of the puzzle finally came together and the first “official” TB3 eGFX chassis were introduced by Razer and PowerColor.

While both Razer Core and PowerColor Devil Box do the job and even feature a couple of extras, they are rather expensive as they add more than just GPU features - $499 and $379, respectively. This week, AKiTiO, which has been offering external Thunderbolt enclosures for years, introduced its eGFX box that costs $299.99.

AKiTiO Node (NODE-T3IA-AKTU) eGFX Chassis Specifications
Max Video Card Size		Double-Wide, 12.2" Long (312 × 170 × 44 mm)
Max Video Card Power		300 W (?)
Connectivity		1 × Thunderbolt 3 (>40 Gbps via active cable)
Chassis Size		5.71 × 16.85 × 8.94 inches (145 × 428 × 227 mm)
Internal PSU		400 W SFX
System Requirements		Thunderbolt 3 eGFX Certified PC Thunderbolt 3 w/Active Cable (included) Windows 10
Shipping Date		December 2016
Price		$299

The AKiTiO Node is a 42.8 × 14.5 × 22.7 cm (16.85 × 5.71 × 8.94 inches) box made of stainless steel that is compatible with a variety of double-wide full-length/full-height AMD and NVIDIA graphics cards (see the list here). The Node comes with an integrated 400 W SFX PSU (which means that it can be upgraded) as well as one 120 mm fan. As for dimensions, the Node seems to be a little more spacious inside than its rivals, which is good for cooling and compatibility with custom video cards (such as those from ASUS or EVGA). Despite this, GPUs with hybrid cooling (integrated liquid cooling) are not supported, such as AMD Radeon R9 Fury X.

Comparison of Thunderbolt 3 eGFX Chassis
		AKiTiO Node	BizonTech BizonBox 3	PowerColor Devil Box	Razer Core
Chassis Dimensions	Length	42.8 cm 16.85 in	36 cm 14.17 in	40 cm 15.748 in	34 cm 13.38 in
Height	22.7 cm 8.94 in	20.5 cm 8.07 in	24.2 cm 9.52 in	21.84 cm 8.6 in
Width	14.5 cm 5.71 in	8 cm 3.5 in	17.2 cm 6.77 in	10.5 cm 4.13 in
Max Dimension of Compatible Graphics Card	Length	31.2 cm 12.2 in
Height (PCB+Cables)	17 cm 6.7 in	over 14" over 5.51"	14 cm 5.51 in	15.2 cm 5.98 in
Width	4.4 cm 1.73 in		5 1.96 in	4.4 cm 1.73 in
Maximum GPU Power		300 W (?)		375 W
PSU	Wattage	400 W	200W/400W	500 W
Form-Factor	SFX	external	internal proprietary
Cooling Fans		1 × 120 mm	2 × 60 mm	unknown	3 × 80 mm
Connectivity	Thunderbolt	1 × TB3	2 × TB3	1 × TB3
Ethernet			1 × GbE
USB	-		4 × USB 3.0
SATA	-		1×SATA 6Gb/s	-
DisplayPort	-	1 × DP 1.2	-
Availability		December 2016		October 2016	April 2016
Price		$299	$649	$379	$499

Unlike competing products, the AKiTiO Node is a pure external GPU box and nothing more. It does not have extra USB 3.0 ports (to connect a VR headset, for example), a GbE controller nor a SATA connector. This naturally cuts down its costs and allows Akito to sell the eGFX enclosure at $299 price-point, but additional USB ports will be greatly missed by owners of ultra-thin laptops as well as SFF PCs.

AKiTiO plans to start selling the Node (NODE-T3IA-AKTU) in December.
Gallery: AKiTiO Introduces Node: Thunderbolt 3 eGFX Box for $299

More...

Alienware 13 R3 (2016)
		i5-6300 HQ dkcwe01h	i5-6300 HQ dkcwe02h	i7-6700 HQ dkcwe03h	i7-6700 HQ dkcwe04h
CPU		Intel Core i5-6300HQ 4C/4T 2.30/3.20 GHz 6 MB LLC HD Graphics 530 45 W		Intel Core i7-6700HQ 4C/8T 2.60/3.50 GHz 6 MB LLC HD Graphics 530 45 W
dGPU		NVIDIA GeForce GTX 1060 6 GB of GDDR5 at 8 GT/s 1280 SPs 80 texture units 48 ROPs 192-bit memory bus
DRAM	Installed	8 GB DDR4-2133 (8 GB × 1)	16 GB DDR4-2133 (8 GB × 2)	8 GB DDR4-2400 (8 GB × 1)	8 GB DDR4-2400 (8 GB × 1)
Upgradeable	32 GB DDR4-2133 (16 GB × 2)	32 GB DDR4-2400 (16 GB × 2)	32 GB DDR4-2400 (16 GB × 2)	32 GB DDR4-2400 (16 GB × 2)
Display	Type	TN	IPS	OLED	OLED
Resolution	1366×768	1920×1080	2560×1440
Brightness	200 cd/m²	300 cd/m²	400 cd/m²
Contrast		unknown	100,000:1
Touch	-	+
Storage (2x M.2)	Installed	180 GB SATA	256 GB PCIe		512 GB PCIe
Upgradeable	1 TB PCIe + 1 TB PCIe SSD
Battery		76 Wh
Wireless Connectivity		Rivet Networks Killer 1435 802.11ac 2×2 WiFi + Bluetooth 4.1
Ports		Thunderbolt 3 × 1 (Type-C) USB 3.1 × 1 (Type-C) USB 3.0 × 2 (Type A) GbE × 1 (Killer E2400) mDP 1.2 × 1 HDMI 2.0 × 1 3.5 mm audio × 2 Graphics Amplifier × 1
Camera		Windows Hello Tobii Eye Tracking
Dimensions		390 × 259 × 23.3 mm 15.35 × 10.2 × 0.92 inches
Weight		2 kg 5 lbs
Price		$1200	$1500	$1800	$2100

Thread: Anandtech News

Thread Tools

Search Thread

Anandtech: The Zalman ZM1200-EBT 1200W Power Supply Review

Anandtech: BenQ Announces the ZOWIE XL2540: 24-Inch 240 Hz Full-HD Display 'for e-Spo

Anandtech: AOC Launches the AG352QCX: 35-Inch 200 Hz 2560×1080 Curved Display with Ad

Anandtech: Alienware 13 R3: Quad-Core CPU, GeForce GTX 1060, QHD OLED, VR Ready

Anandtech: NVIDIA Announces Record Q3 FY 2017 Results

Anandtech: HTC Announces the HTC Bolt - A Sprint-Exclusive Smartphone for LTE Plus

Anandtech: ASUS ROG GT51 Desktop: Overclocked i7-6700K, GTX 1080 SLI, 64 GB DDR4, $49

Anandtech: Commercial NAS Operating Systems - A Comprehensive Overview of Core Featur

Anandtech: AMD @ SC16: Radeon Open Compute Platform (ROCm) 1.3 Released, Boltzmann Co

Anandtech: AKiTiO Introduces Node: Thunderbolt 3 eGFX Box for $299

Thread Information

Users Browsing this Thread

Tags for this Thread

Bookmarks

Bookmarks

Posting Permissions

Anandtech News