Brad Chacos
(0 comments, 217 posts)
This user hasn't shared any profile information
Home page: http://www.maximumpc.com/articles/all/feed
Posts by Brad Chacos
Cooler Master Adds Fans To Hyper 412 Slim CPU Cooler, Intros Three New Thermal Pastes
0
Cooler Master has spread its wings into a lot of different product lines, but it's still best known for its namesake: stuff that keeps your PC running cool. To that effect, today the company announced an update to the design of its Hyper 412 Slim CPU cooler as well as three new thermal pastes.
The Hyper 412 Slim redesign introduces a pair of appropriately slim fans on either side of the heatsink. Why slim fans, you ask? Cooler Master says the new look will increase cooling performance while still allowing you to plop memory down around the cooler. The Hyper 412 Slim works best with LGA 2011 sockets and should be available next month for around $50.
Check out more about the Hyper 412 Slim on the Cooler Master website.
Of course, you need thermal paste to install a CPU cooler, so Cooler Master announced three new compounds to go with the redesigned Hyper 412 Slim: the gold-colored, IC Essential E1, the grey-colored IC Essential E2, and the white-colored IC Value V1, which offer varying levels of conductivity. All three will join the new-look Hyper 412 Slim on store shelves in June.
Samsung Researchers Design A Better Graphene Transistor
0
Intel's doing a bang-up job and shrinking transistors and packing them in tighter than ever before, but let's face it: it's going to be hard to scale silicon down much further. That eventual wall is why engineers are pumped about the potential of graphene, a substance with more than 200 times the electron mobility of silicon. (Read: better potential performance.) Coaxing graphene transistors into switching off current to create the 1 and 0 signals we know and love has been tricky, however. Now Samsung says it's developed a solution that does just that, without limiting graphene's electron mobility.
That last part is key: most of the previous solutions to graphene's electric current woes involved transforming the material into a semi-conductor, but doing so reduced its electron mobility -- thereby eliminating much of its performance gain over silicon.
Samsung Electronics' R&D arm, the Samsung Advanced Institute of Technology, created a special three-terminal Schottky barrier (dubbed the "Barristor" by Samsung) that can be raised or lowered in order to allow or impede the flow of electrical. The three-terminal device is basically a gate.
So how does Barristor stop current without mucking up electron mobility? From the abstract:
The key is an atomically sharp interface between graphene and hydrogenated silicon. Large modulation on the device current (on/off ratio of 10^5) is achieved by adjusting the gate voltage to control the graphene-silicon Schottky barrier.
Subscribers to the Science journal can check out the full text for an even more in-depth and jargon-filled explanation. It was published on the website yesterday.
Researchers Use Calculation Errors To Greatly Improve Processor Speed And Power Efficiency
0
A team of researchers from prominent institutions around the world claim that they've figured out how to make computer processors smaller, faster and more power efficient than ever before: by letting chips mess up once in a while. No, seriously. By allowing "inexact" chips to make a pre-calculated amount of errors rather than striving for absolute perfection, the researchers claim that drastic power reductions can be made -- and they already have a working prototype.
The researchers -- from Rice University, Singapore’s Nanyang Technological University, Switzerland’s Center for Electronics and Microtechnology, and the University of California, Berkeley -- use probability to account for the number of errors created by the inexact chip and place limitations on exactly which chip calculations can produce errors, to avoid mistakes during critical processes. Since an inexact chip isn't constantly triple-checking its work for pinpoint accuracy, it also receives a significant processing speed increase compared to traditional processors.
A couple of innovations help inexact chips consume less power than traditional chips. The variance allows the team to "prune" infrequently used sections of the chips to make the hardware smaller, while a technique called "confined voltage scaling" taps into the processing speed increase to reduce power consumption when it isn't needed.
“In the latest tests, we showed that pruning could cut energy demands 3.5 times with chips that deviated from the correct value by an average of 0.25 percent,” study co-author Avinash Lingamneni said in a Rice University press release. “When we factored in size and speed gains, these chips were 7.5 times more efficient than regular chips. Chips that got wrong answers with a larger deviation of about 8 percent were up to 15 times more efficient.”
But what good is a chip that screws up all the time? Plenty of good, as it turns out. While you obviously wouldn't want an inexact chip running, say, critical banking or security systems, the researchers say it could provide significant gains in areas that can have more tolerance for slight errors. Since the human body is hard-wired for basic error correction, the release says that initial markets may include hearing aids and devices that generate pictures.
"We used inexact adders to process images and found that relative errors up to 0.54 percent were almost indiscernible," project co-investigator Christian Enz reports. "Relative errors as high as 7.5 percent still produced discernible images.”
Hey, does that mean Photoshop might get cheaper in the future? In any case, the team hopes to have the first inexact chip-powered prototype hearing aids and educational tablets available sometime in 2013.
Comcast Plans To Raise Subscriber Bandwidth Caps, Add Additional Data Options
0
Comcast has fallen under fire recently for the way it handles its Xfinity app on the Xbox 360 console; basically, the company doesn't count the bandwidth against subscribers' monthly data cap. It sounds great for Comcast customers, but critics -- including Netflix's Reed Hastings -- say the practice is a violation of Net Neutrality. Perhaps to silence the screams for blood, Comcast announced today that it plans on increasing its data cap and trying out some new data management approaches.
The current Comcast policy entails a hard 250GB bandwidth cap. Top it too many times and Comcast could cut you off. That looks like it's changing in the future.
Comcast VP Cathy Avgiris described the possible "flexible data usage management approaches" on the company's blog today. One approach involves the data cap being raised to 300GB, with additional bandwidth available at the example rate of $10 for 50GB. (That's obviously subject to change.) The second trial approach is basically the same, but customers with more expensive Internet packages receive even higher bandwidth caps.
If you're not in a trial market, Comcast will "suspend enforcement of our current usage cap as we transition to a new data usage management approach, although we will continue to contact the very small number of excessive users about their usage."
Unfortunately, there's no word on when the trials and changes are actually going to take place.
EVGA Recalls Some GTX 670 Superclock Cards
0
Ruh-roh! Being an early adopter of technology often means putting up with headaches while a product's kinks get worked out, and it seems that's holding true for at least some early GTX 670 buyers. EVGA apparently forgot to quality test a small batch of GTX 670 Superclock cards and is recalling them as a result.
HardOCP's Kyle Bennett heard the first whispers and asked EVGA what was going on. Here's what the company told him:
EVGA has isolated this problem to an early batch of GTX 670 Superclock cards (P/N: 02G-P3-2672-KR) that were not properly screened during QA/QC procedure. We have already been working with our partners to retest this particular batch. In the meantime, our R&D has also done numerous tests, burn in and component quality verification to confirm that the EVGA GTX 670 Superclock is a well designed product.
The rep didn't mention whether there were specific defect concerns or a way to identify potentially untested cards. Several threads in EVGA's GTX 600-series subforum mention GTX 670s crashing shortly after installation; maybe the cause is related? In any case, those forum users report that the RMA process is painless, and EVGA told Bennett that they'd be replacing customers' recalled GTX 670 Superclocks with GTX 670 FTW cards to help make up for the headache.
Is your EVGA GTX 670 Superclock acting up? The company told HardOCP that customers should contact Jacob Freeman at jacobf@evga.com to get the RMA ball rolling.










