We’ve barely entered 2012 and already OCZ wants to flaunt the next in its series of enterprise PCIe SSDs. The R5 sports an entirely new ‘Kilimanjaro’ controller platform (shown in the reference design above), developed in cahoots with Marvell and incorporated into each and every flash module that you might wish to add to the base card. These scalable controllers communicate directly with the host system, removing the need for an extra SATA RAID chip and thereby promising greater speeds — especially as you pile on more modules. We won’t get full specs until CES, but in the meantime OCZ has hinted at a doubling of the SandForce-based R4′s performance, which could take us into the three million IOP realm. So long as the company also tackles the question of reliability on this new type of drive, then it’ll likely be an easy sell. Check out the source link for more.
You ask for more graphics power, well Asus ROG’s Andre Yang shows off the power of a quad-CrossFireX Radeon HD 7970 setup, which captures not one, not two but five “unofficial” world records.
In an article on the overclocking world rankings HWBOT, it was reported that Asus ROG’s Andre Yang posted to Facebook a news update showing off new World Records for 3DMark and Unigine Heaven. HWBOT does point out the results are not confirmed to be legit just yet. It is believed that the results will be made official after the official launch of the Radeon HD 7970 on January 9th.
Yang used four Radeon HD 7970′s and an Intel Core i7-3960X on an Asus Rampage IV Extreme X79 motherboard to achieve the five “unofficial” world records. With the help of overclocking on liquid nitrogen, he performs the benchmarks with CPU clock speed @ 5.6 GHz and GPU clock speed up to 1,600 MHz.
They popped out at local retailers by a small number since last November, but these motherboards sporting Intel’s X79 chipset are finally starting to make a decent dent in the marketplace and, after showing off a pair of options at IDF, MSI is back with a model for the serious builders out there. The Big Bang-XPower II comes packing the features you’d expect of any gamer-friendly MoBo, including four-way SLI and CrossFire support, PCIe 3.0, eight DDR3 slots and plenty of extra power connectors for hooking up your high-powered parts. And the company is taking care of overclockers too with military grade components and features like Direct OC, which lets you tweak your CPU’s frequency with the push of a button. Yes, overclocking can be that cool too.
Tablets might be all sexy curves and Gorilla Glass, but it’s what’s on the inside that counts, right? Efficient DRAM mightn’t set your heart alight, but Elpida Memory did just start shipping its next generation 30nm “Wide IO Mobile RAM.” The firm claims it uses 50 percent less power compared with equivalent DDR2 sticks, letting you caress your beloved device for longer. This economy is thanks to it purring along at just 200MHz, which is even more impressive when you consider it brags a 12.8 GB/s data rate per chip. This pumped, yet frugal, performance comes courtesy of using x512-bit data width — some ten times larger than that of existing DRAMs. The party doesn’t stop there, with the Elpida also debuting its LPDDR3 chip, brushing aside LPDDR2 with twice the data rates at 6.4GB/s a slice, and a 25 percent smaller power-drain in tow. Mass production should commence in 2012, check the source link for the full break-down.
Intel is preparing a wave of new server processors for the first half of 2012: 20 new CPUs in Q1 and another 20 new models in Q2, according to Digitimes.
There will be seven 8-core CPUs in the first quarter, including the Xeon E5-2690, two 6-core processors (Xeon E5-1660 and 1650) as well as three quad-core processors including the E5-2609 and dual-core processor E5-2637.
For Q2, Intel has apparently scheduled eleven Ivy Bridge H2-series CPUs and seven additional Xeon processors, including the Xeon E5-2470. There will also be several low-power models, including the eight-core Xeon E5-2650L and six-core E5-2630L, both of which will debut in Q1, as well as the Xeon E5-2450L and six-core E5-2430L in Q2.
With the HD 7970 set to release this month and HD 7990 & HD 7950 shortly after, we get our first leaked information on the HD 7770. It is the first upgrade to this series since the HD 5770 with the HD 6770 being a re-branded HD 5770.
ChipHell forum member bigpao007 has leaked some pictures and 3DMark results of the upcoming HD 7770. The HD 7770 is set between the current HD 6800 and HD 6700 series in AMD’s graphics platform and is a more “budget” friendly GPU compared to the HD 7900 series. The GPU uses the new 28nm Cape Verde Southern Islands featuring the Graphics Core Next architecture. Read more…
Digitimes reports that prices for wafers built on mature node processes have been dropped by 10 to 15 percent because of declining production costs. Of course, it is not just a selfless move; the intention is to convince customers to order more chip and build up inventory.
That, however, may not be such an easy argument, given the general uncertain business outlook and macroeconomic challenges. While the publication said that chip inventories are at “safe levels”, it is unlikely that chip customers will be rushing into premature orders if they don’t know how the market will develop in the new year.
Digitimes noted that chip inventories were at “excessive levels between the end of the second quarter and the beginning of the third quarter,” which was causedby a “combination of negative macroeconomic factors such as weak consumer confidence in the U.S. and the European crisis.”
Earlier on, reports surfaced around the Web that certain Gigabyte X79-based motherboards were having hardware issues that, in some cases, could lead to failure. Gigabyte reached out to us to refute that claim, clarifying that the issue was limited to its firmware, which could be easily fixed via a BIOS update.
To further prove its claims that all is well across its X79 line, consisting of the G1.Assassin 2, GA-X79-UD7, GA-X79-UD5 and GA-X79-UD3, Gigabyte passed along results demonstrating an Intel Core i7-3930K overclocked with a 57x multiplier running on a GA-X79-UD3 with Kingston HyperX Genesis PC3-19200 2 GB x 4 (@ 2376 MT/s) and a Corsair AX1200 PSU.
Overclocker “Hicookie”, equipped with the recently-issued F7 BIOS and liquid nitrogen cooling, was able to push to a frequency of 5643.3 MHz, producing record scores for the Super Pi 1M and 32M and PiFast benchmarks. Check out Hicookie’s most recent records on HWBot for the numbers and the video below of the overclock in action.
According to a blog post by Jean Luc Pelloie, the company’s Fellow Director of SOI Technology, 20 nm may represent an inflection point in which it will be necessary to transition from a metal-oxide semiconductor field-effect transistor (MOSFET) to Fin-Shaped Field Effect Transistors (FinFET) or 3D transistors, which Intel refers to as tri-gate designs that are set to debut with the company’s 22 nm Ivy Bridge product generation.
Pelloie explains that it is “increasingly difficult to control the vertical electric field between the gate and the substrate while maintaining the channel depletion below threshold and then minimize the leakage current” when shrinking transistors. FinFETs are considered a solution to solve this problem. The engineer wrote that FiNFETs could be used on either bulk or SOI wafers. However, ARM still has work to do and especially investigate the scalability of this technology. “However, 3D devices are clearly on the road for sub-20nm nodes … and FinFET’s time may finally be here,” Pelloie wrote.
FinFET history goes back to a December 2000 paper published in IEEE Transactions on Electron Devices entitled “FinFET – A Self-Aligned Double Gate MOSFET Scalable to 20 nm“. Intel was first to move to tri-gates and take this technology into mass-production, but was accused of having switched to 3D transistors only because it was not able to scale its SRAMs, which the company traditionally uses to unveil a manufacturing process, from 32 nm to 22 nm. Intel’s tri-gate technology was announced back in May of this year. Intel hopes that tri-gate technology will keep up with Moore’s Law and especially drive low-voltage and low-power features.
With the paper launch of the Radeon HD 7970 last week, we are starting to get more leaked information about the other models hitting the market under the new “Tahiti” platform. In the information leaked by xtreview, The HD 7950 will have 1792 stream processors, 28 GCN compute units, which is down 256 stream processors from the HD 7970. It will utilize a 384-bit memory bus and 3GB GDDR5 just like the HD 7970. The HD 7950 looks to utilize 112 texture units and 32 full color ROPs. It has support for a maximum of 6 displays. What is still unknown at this time is the core clock speed, memory clock speed and price point. With the HD 7970 set to release around $550, I would expect the HD 7950 to fall closer to the $400 price mark at release.
AMD is expected to release its flagship dual-GPU graphics card based on the “Tahiti” platform in 2012. We are learning under the codename “New Zealand”, AMD is working on the monster HD 7990, which will utilize two HD 7970′s and 6GB of total graphics memory (3 GB per GPU system). If AMD will indeed go for the HD 7970, this could mean that the card will include 62 compute units for a total of 4096 stream processors, 256 texture units and 64 full color ROPs.
One thing that will benefit the HD 7990 is AMD’s new ZeroCore technology. ZeroCore technology completely powers down other GPUs, other than the primary one, to zero state, when the system is not running graphics-heavy applications. This means at idle, desktop or video playback state, the HD 7990′s power draw will be nearly equal to that of the HD 7970, which through early testing was impressive.
Recent Comments