Thread: Anandtech News

Corsair is a well-known manufacturer of PC components, including DRAM, chassis, power supplies, USB storage, fans, SSDs, gaming peripherals (keyboards, mice, headsets) and cooling, among others.  Today we are looking at some of their mid-to-high range memory from their Vengeance Pro range, designed to cater for extreme system builders with DDR3-2400 CAS 10 speeds.  This also happens to be the memory we have been using for Haswell motherboard reviews.

More...

The drive for faster and more efficient interfaces, particularly those over cables, is driven by a large number of consumer devices producing and displaying higher definition content.  This only ever continues to grow, and thus the ecosystem around data transfer has to catch up – one prong of this voyage is that of the Intel Thunderbolt controller.  Now in its second generation format, Thunderbolt 2 offers up to 20 Gbps of PCIe and DisplayPort data across the controller.  The most famous device to be equipped with the controller comes from Apple in the new Mac Pro, however PC partners are also joining in the game.  The sword today is being wielded by GIGABYTE, and their Z87X-UD7 TH motherboard.

We saw this motherboard back at Computex, still in its design phase and waiting for Intel to launch their TB2 controller to OEMs.  The UD7 moniker on GIGABYTE motherboards is only for the high end channel segment of their lineup, and the specification list shows:

• Dual port Thunderbolt 2 (20 Gbps combined max throughput)
• PLX8747 chip for 4-way SLI and Crossfire, with PCIe x16 bypass
• Bundled 802.11ac WiFi PCIe x1 card
• Dual Intel Gigabit Ethernet
• Ultra Durable 5 Plus (60A IR3550 ICs and 60A chokes in the power delivery)
• OC Touch/OC Ignition buttons
• Nine fan headers
• Realtek ALC898 Audio + Sound Blaster X-Fi MB3 software + Headphone Amplifier
• Ten USB 3.0 ports, ten SATA 6 Gbps ports
• High gold content connectors to reduce resistance and increase longevity
• Dual HDMI video outputs

Putting this all into perspective and pricing: a TB2 controller costs around $40, a PLX chip is somewhere between $20-$40 depending on the deal, an 802.11ac module is found on $150 ITX motherboards so cannot be that expensive, IR3550 ICs and Chokes are ~$4 each which is expensive for power delivery, and features like OC Touch/OC Ignition have associated R&D costs of their own.  This is alongside high gold content connectors, USB 3 and SATA 6 controllers, dual Intel NICs and support for nine fan headers.

What should be interesting is the PCIe layout of this motherboard: here we have a system using a PLX chip, promoting eight or sixteen PCIe lanes from the CPU to 32.  The TB2 controller needs PCIe 2.0 x4, which could either come from the PLX 8747 chip or the chipset/PCH.  Due to the wording on the Gigabyte website, it would seem that they come from the PCH, using switches such that when the Thunderbolt ports are used, some of the PCIe lanes are not available. 
There is also something intriguing regarding FlexIO.  FlexIO is the system that distributes PCIe 2.0 lanes from the chipset.  Normally there are eighteen lanes available, four reserved for SATA 6 Gbps ports, four reserved for USB 3.0 ports and six reserved for normal PCIe x1.  That leaves four left, split into pairs – the first pair can be configured as 2x USB 3.0 or 2x PCIe x1, and the second pair can be configured as 2x PCIe x1 or 2x SATA 6 Gbps, or a combination (one of each) therein, although the total must not be more than eight PCIe x1 lanes.  In the specification sheet, GIGABYTE is only using two USB 3.0 ports from the PCH, choosing to add two Renesas uPD720210 USB 3.0 hubs.  That means that two USB 3.0 ports (or the equivalent of two PCIe x1 lanes) are not being used, or GIGABYTE has found a way to fool the chipset into thinking they are USB ports when in fact they are something else.  This is definitely worth delving into, but Occam’s Razor may apply: they might simply not be used.
GIGABYTE are aiming for the prosumer with this model: it differs from gamers (who have the G1 range) and overclockers (the Z87X-OC and Z87X-OC Force) and goes straight for a solid work system/workstation build alongside competition from other motherboard manufacturers.

Pricing and release date are as-yet unannounced, though I would not be surprised if we come nearer the $350-$400 mark by the end of the year. 
Edit: The board is available today from Newegg for $430, which is around the OC Force pricing.
Gallery: Thunderbolt 2 certified on GIGABYTE Z87X-UD7 TH

More...

The retail mSATA SSD market doesn't have too many players. Most OEMs, such as Samsung, Toshiba and SanDisk, only sell their mSATA SSDs straight to PC OEMs. Out of the big guys, only Intel and Crucial/Micron are in the retail game but fortunately there are a few smaller OEMs that sell retail mSATA SSDs as well. One of them is Mushkin and today we'll be looking at their Atlas lineup.

More...

Within the last year, there have been several major acquisitions in the storage industry. Western Digital acquired STEC, Virident and Velobit; SanDisk acquired SMART and now Toshiba is taking over OCZ's assets. With two weeks of 2013 left, there's still time for one more: Avago, a company which supplies a variety of semiconductors, is acquiring LSI for $6.6 billion in cash. 
Avago may not be a familiar name to most of you but there is a good chance that you own a device with an Avago chip in it. iPhone 5 along with multiple other Apple devices uses an LTE duplexer module and power amplifier from Avago but these are chips that don't usually get much attention. In terms of the storage industry, Avago's focus is on the enterprise side and they make Fibre Channel transceivers and Fiber Optics for instance.
For Avago, the motive behind the acquisition is LSI's IP and know-how. With LSI's product portfolio Avago will be able to expand their offerings in the storage industry and move to more vertically integrated business model. As for LSI, I've heard that part of the motivation was shareholders' dissatisfaction with the company and especially its growth opportunities. Avago opens plenty of doors since they have expertise in multiple fields of semiconductors, which may be useful in the future as the storage industry becomes more and more complex.
I'll be speaking with LSI tomorrow to find out more about the deal but I'm guessing LSI will, at least in the near future, continue to operate normally. 

More...

Squeezing in one final product announcement before the end of the year, today AMD is announcing the latest addition to the Radeon R7 family and the counterpart to the existing R7 260X, the R7 260. The R7 260 will be the near-obligatory lower tier part for the R7 260 series, featuring a lower performance Bonaire part in order to further bring down the price of Bonaire and fill in the gap between 260X and 250.
Looking quickly at the specs, whereas R7 260X is a fully enabled Bonaire part, R7 260 will be a cut down version of Bonaire, making it the first time AMD has shipped Bonaire in a less than full configuration. AMD has cut off two CUs, bringing the SP count down to 768 SPs and the texture unit count down to 48. Meanwhile clockspeeds have also been reined in slightly, with 260 topping out at 1GHz for the GPU clock and 6GHz for the memory clock. This puts theoretical performance at around 80% of 260X in GPU bound scenarios, and 92% in memory bound scenarios.
Unfortunately, but as to be expected, memory capacity will also be taking a hit to meet AMD’s price requirements. For the 260 AMD and will be outfitting the card with 1GB of VRAM versus 2GB on the 260X. The 260X was just fast enough to benefit from 2GB, so we expect that the 260 will be in something of a twilight zone in that respect, being a good fit for 1GB in some scenarios and capable of putting 2GB to good use in other scenarios.

AMD GPU Specification Comparison
 	AMD Radeon R9 270	AMD Radeon R9 260X	AMD Radeon R9 260	AMD Radeon HD 7770
Stream Processors	1280	896	768	640
Texture Units	80	56	48	40
ROPs	32	16	16	16
Core Clock	900MHz	?	?	1000MHz
Boost Clock	925MHz	1100MHz	1000MHz	N/A
Memory Clock	5.6GHz GDDR5	6.5GHz GDDR5	6GHz GDDR5	4.5GHz GDDR5
Memory Bus Width	256-bit	128-bit	128-bit	128-bit
VRAM	2GB	2GB	1GB	1GB
FP64	1/16	1/16	1/16	1/16
TrueAudio	N	Y	Y	N
Transistor Count	2.8B	2.08B	2.08B	1.5B
Typical Board Power	150W	115W	95W	100W
Manufacturing Process	TSMC 28nm	TSMC 28nm	TSMC 28nm	TSMC 28nm
Architecture	GCN 1.0	GCN 1.1	GCN 1.1	GCN 1.0
GPU	Pitcairn	Bonaire	Bonaire	Cape Verde
Launch Date	11/13/13	10/11/13	01/14/14 (Est)	02/15/12
Launch Price	$179	$139	$109	$159

Going hand-in-hand with lower performance will be lower power consumption. AMD’s official specifications state that power consumption for 260 is 95W, down from 115W on 260X. This means 260 will still require 1 PCIe power connector – Bonaire’s simply too big for sub-75W operation in a desktop SKU – but it will still be felt in very low power builds where every watt matters.
Meanwhile in an unusual move, AMD is announcing the 260 now, but won’t be launching it until “mid-January”, almost exactly a month from now. AMD doesn’t typically announce mainstream/value cards ahead of time, and admittedly we’re at a bit of a loss as to why they’re doing this now given the approaching holidays. Cards at this end of the spectrum don’t have the same pent-up demand as on the high-end, so AMD doesn’t have to contend with the rumor mill and price/positioning challenges as they do at the high-end as exemplified by the 290 series.
In any case, while the 260 won’t be available until next month AMD has already announced the product’s MSRP, placing it at $109. This is $30 less than the 260X and about $20 more than the R7 250, closing that relatively significant gap in AMD’s 200 series product lineup. At this price the 260 finally replaces the 7770, which has been retailing around $110 for the last few months. This is one of the few cases where AMD will be outright replacing a GCN 1.0 part with a GCN 1.1 part, with 260 bringing with it TrueAudio and power management functionality that the Cape Verde based 7770 did not feature.
Elsewhere, at $109 the expected competition will be NVIDIA’s GTX 650 and GTX 650 Ti, both of which regularly hover near that price point depending on what rebates are involved. As a reminder the bulk of AMD’s discrete video card business is below $150, so while cards like the 260 don’t attract the same level of glory or attention as high-end cards, from a volume standpoint they’re AMD’s bread and butter and are positioned accordingly.
Finally, as is usually the case for parts in this price range this is a pure “virtual” SKU from the start, so while AMD has a reference solution for testing and validation (pictured above), partners will be doing custom designs right away. We don’t have any further product details in-hand yet, but we’d expect all of these to be dual slot open air coolers, similar to the reference design and other cards in this power range.

More...

The final kit in our current run of DDR3 on Intel reviews falls at the feet of a kit that blends a high rated speed with density.  When it comes to high MHz numbers, we typically see 4 GB modules as the standard, due to a higher density kit being more difficult to push in frequency.  ADATA sampled us two of their DDR3-2800 C12 kits in 2x8 GB form, representing perhaps one of the final hurdles before DDR4 reaches the market, as long as your wallets are deep enough.

More...

Last week we held an awesome Ask the Experts Q&A with ARM's Peter Greenhalgh, lead architect for the Cortex A53. Peter did a great job answering questions in the comments, but for those of you who missed them we're compiling all of the Q&A here for you to go through. 
 
On Friday, December 20th at 12:00PM ET, Peter will be joining me for a live Google Hangouts chat. We'll be posting more details on that later this week. For now, enjoy the responses!

More...

The Humble Jumbo Bundle is now sadly over, but while good things come to an end we can always look forward to the next big thing. For the next two weeks, Humble is offering up a new bundle that includes six games for PC, OS X, Linux, and Android -- yes, all six titles (with more to come next Tuesday) are available for all four platforms (with Steam versions if you donate more than $1).
Following the standard rules, there are four games you get with any donation: Jack Lumber (77% iOS), Gemini Rue (82% PC), Little Inferno (82% iOS), and AaaaaAAaaaAAAaaAAAAaAAAAA!!! for the Awesome (also included is the Android counterpart, AaaaaAAaaaAAAaaAAAAaAAAAA!!! (Force = Mass x Acceleration) - 90% iOS). Pay more than the average (currently $4.04) and you’ll also get Hero Academy (77% iOS) and Anomaly 2 (86% iOS). The iOS versions are likely quite similar to the Android releases, but it appears the PC releases were generally not as highly favored by reviewers.
There's another Humble Weekly Bundle going on that's worth mentioning, though there's only about a day remaining. It includes six games from ACE Team, ATLUS Games, and Tripwire Interactive, with four core games and two extras if you donate $6 or more. The core games are Zeno Clash (77% PC) and Zeno Clash II (61% PC) for Windows and/or Steam, and Killing Floor (72% PC) and Dwarfs!? (77% PC) for Windows, OS X, Linux, and/or Steam. The bonus titles are Rock of Ages (74% PC) and Rising Storm (82% PC), both for Windows/Steam.
Child's Play Charity remains as one of the donation options for both bundles, with the annual Child's Play purse growing every year. American Red Cross and the EFF are also options. And since it's the season of giving, try not to be too much of a Scrooge. ;-)
Gallery: Humble Bundle for PC and Android 8 and More

More...

Back in October alongside the release of the first ShadowPlay beta, NVIDIA announced that ShadowPlay would be getting support for an additional video encoding mode: Twitch.tv video uploading. At the time the Twitch functionality was being delivered as an early announcement that wasn’t going to be ready for the first ShadowPlay beta, and now a bit less than 2 months since ShadowPlay was released, NVIDIA is adding the feature to their latest version of GeForce Experience/ShadowPlay.
GeForce Experience 1.8.1, released this week, is a point update for GeForce Experience that adds support for Twitch uploading. Both conceptually and technically Twitch uploading is a being implemented as function of NVIDIA’s existing ShadowPlay hardware, leveraging the same fast capture APIs and hardware encode functionality, replacing normal file encoding with uploading to the Twitch service. Hardware video encoding is still a rare feature for Twitch-capable capture utilities, so as with ShadowPlay NVIDIA is looking to exploit their edge here in capturing and hardware encoding to offer a client without the significant overhead of inefficient frame captures and software video encoding.

Since it is a function of ShadowPlay, GeForce Experience’s Twitch support has the same basic limitations of ShadowPlay itself, primarily that capturing is limited to full screen applications. So GeForce Experience can’t (currently) replace scenarios where windowed or borderless windowed mode capturing is necessary. That said NVIDIA has added overlay support specifically for Twitch, allowing the operator camera to be overlaid onto the video output without breaking the encode chain.
Also, as with ShadowPlay GFE’s Twitch support comes at three different quality/bitrate settings. High, Medium, and Low correspond to 3.5Mbps 2Mbps, and 0.75Mbps respectively.

Shadowplay Twitch.tv Bitrates
High Quality	3.5Mbps
Medium Quality	2Mbps
Low Quality	0.75Mbps

NVIDIA has also posted a quick demonstration video to showcase the functionality and quality of their Twitch upload support.







Watch live video from EatGamer on TwitchTV On a final note, with basic Twitch support now in ShadowPlay, NVIDIA has mentioned what their next goals are for their utility. “Desktop capture for ShadowPlay, and enhanced microphone control” are mentioned as the next milestones, with desktop capture in particular being of interest since it would eliminate the full screen game requirement under the current version of ShadowPlay. That said we do know that NVIDIA has to use different capture APIs for that versus a full screen application, so it will be interesting to see just what further impact (if any) desktop capture may have.

More...

As I hinted at in our Ask the Experts post with ARM's Peter Greenhalgh, a good turnout from all of you in the comments might help secure some giveaways from ARM. You guys held up your part of the bargain, now it's our turn. ARM is giving away two international Samsung Galaxy Note 3 smartphones (N9005) to two lucky AnandTech readers. Brian reviewed the US version of the Galaxy Note 3 here. The N9005 is an unlocked international variant of the Galaxy Note 3. This particular variant has all of the LTE bands you'd want for europe/international use, but not for use on any US operators (I've listed band support in the table below). That being said, HSPA+ support shouldn't be an issue. We're giving away 32GB black versions here. A full list of its specs are below:

 	Samsung Galaxy Note 3 (SM-N9005)
SoC	2.3 GHz Qualcomm Snapdragon 800 (MSM8974)
Display	5.7-inch Super AMOLED (1080p)
RAM	3 GB LPDDR3
WiFi	802.11a/b/g/n/ac (BCM4339) + BT 4.0
Cellular Bands	GSM/GPRS/EDGE: 850/900/1800/1900MHz HSPA+: 850/900/1900/2100MHz LTE: 800/850/900/1800/2100/2600MHz
Storage	32GB + microSDXC (up to 64 GB)
I/O	microUSB 3.0, MHL 2.0, IR LED (remote), NFC
OS	Android 4.3
Battery	3200 mAh, 3.8V, 12.1 Whr
Size / Mass	151.2 x 79.2 x 8.3mm, 168g
Camera	13 MP w/AF, LED (Rear Facing) – 1080p60, 720p120, 4k30 2 MP (Front Facing)

To enter the giveaway please leave a comment below (please only post one). As with all of our giveaways, this is only open to US residents with US mailing addresses. Unfortunately this is a legal requirement as each country/territory requires its own set of rules in order to be in compliance. Good luck!

More...