TweakTown NewsRefine News by Category:
AMD has released its Richland line of 35W mobile APUs this morning. The company says that it has been shipping the processors to PC makers since January and that devices running the new APU's are available now.
Richland is may be the new APU in town, but not much of it is different than its predecessor Trinity. The same Piledriver based CPU cores are still there as well as the Radeon HD 6900 based GPU. The new APU is even fabricated on the same 32nm process. The difference lies within more efficient transistors, which provide better power management and better utilizes the existing architecture.
Like Trinity, Richland manages power states via a 32-bit microcontroller. Both have a network of thermal sensors built-in that report back to the MCU, whereas most of these go unused in Trinity, Richland takes full advantage of the sensor network. This allows the CPU and GPU clocks to be changed based on the APU's temperature.
Richland utilizes better algorithms to more precisely tune those clocks based on temperature as well. AMD claims that these improvements allow Richland to consume 17% less power than Trinity and 38% less when playing 720p video.
Intel has a bit of a problem on its hands. It's probably not as terrible as the Sandy Bridge SATA controller issue, but it could end up being another costly fix for the chip giant. According to Fudzilla, the problem cannot be fixed by a mere software update. Rather, the chipset will require a new revision to take care of the issue.
The bug isn't too major. If a USB 3.0 port is being used when the system is put to sleep (S3 state), the device will need to be reconnected upon waking the machine. It's nothing more than a slight hassle, but this is the second chipset glitch experienced by Intel in less than a year.
Motherboard vendors are saying that June is still the planned introduction for Haswell, so it looks like the first batch of motherboards could possibly be coming with the buggy chipset. We'll know more as the proposed launch period nears.
There's some bad news coming out of the cracks of the Internet today, with Hardware.info reporting that they have talked with a "reliable source" who says that the next-generation processor from Intel named Haswell, is having issues with USB 3.0.
The site claims they've taken a look at a document that Intel reportedly sent out to inform system builders of a bug surrounding USB 3.0 and Haswell - the bug specifically has issues with waking from sleep and using a USB 3.0 device. If the system exits S3 level sleep while a user is accessing data on a USB 3.0 drive, their session with that data will end.
This would mean if you're working on a document open before you put your PC to sleep, or it went to sleep after a period of time, when you come back that document could be just blank. Intel is reportedly aware of this issue and are said to be trying to make partners accept the flaw before purchasing the affected CPUs.
A group of overclockers at Canada's biggest LAN party have managed to push an AMD FX 8350 all the way up to 8027MHz. The feat was achieved using extreme voltage and cooing and was part of a demonstration about the importance of insulation and testing the hardware step by step.
The feat was achieved with an ASUS Crosshair V Formula-Z, Corsair Dominator GTX3 memory, Corsair AX850i PSU, and Ryba LN2 pot. Achieving a max speed of 8027.93MHz required 1.968 volts through the core and a multiplier of 31.5 and bus clock of 254.86MHz. Of course, only 2 cores and threads were active for the run.
This is the sixteenth best score achieved on this CPU if you go by Hwbot's listings.
We've already seen the insane performance that NVIDIA's upcoming Tegra 4 should provide us, but Qualcomm aren't just sitting on their hands. Rather, their hands are up in defensive position, swinging at NVIDIA before these chips have even made it into consumer devices.
Qualcomm's Senior Vice President of Product Management, Raj Talluri, told The Verge that Qualcomm is more focused on shipping products than refuting competitors' benchmarks. The company believes that their upcoming Snapdragon 800 easily beats NVIDIA's Tegra 4. Talluri also said that while NVIDIA may be offering something impressive, Qualcomm's Snapdragon 800 is so much more integrated, where the LTE modem is built directly into the silicon die.
The Snapdragon 800 also includes the ability to both encode and decode 4K content, which is a very nice feature to have going into 2013. We should see more than 50 products include the Snapdragon 800 in the near future, with the quad-core 2.3GHz chip expected to arrive in even more devices before the year is up.
Time to spill the beans on Intel's upcoming CPU and chipset plans, if our tomato sauces are correct. One of our sources tipped us that Haswell will be coming at Computex (June 2013) on the Z87 platform as the mid-range solution (much alike the previous-gen Z77), though this is something that is widely available through Google, and mostly widely known among inner circles as being fact.
More importantly, the source let us know that the next-generation performance range of CPUs will be coming by the end of the year, and not 2014 as previous speculated. This news comes from a source that is very close to the subject.
While we suspect this to be Ivy Bridge-E based, our source wasn't exactly clear and it could end up being Haswell-E, with Ivy Bridge-E being skipped. We do know, however, that the next-gen performance chip will make use of a chipset called X99. A quick Google for "X99" turns up relatively little and basically nothing about an Intel chipset.
We'll be sure to keep you up-to-date with all of the latest Intel CPU and chipset news as soon as we hear anything more.
MWC 2013 - Intel had a fairly big announcement this morning at the 2013 Mobile World Congress. The company has teased us with a promise of its third-generation 22nm Atom System-on-Chip processor by mid 2014.
The announcement came as Intel debuted it's new Atom Z2580, Z2560 and Z2520 SoC's which are built on dual 32nm processing cores running at 2GHz, 1.6GHz and 1.2GHz respectively. The new Atom chips support Hyper-Threading which brings the total simultaneous processes count up to four, allowing it to compete with recent ARM processors.
Codenamed Merrifield, the next-generation 22nm Atom's will feature tri-gate transistors, and a big decrease in power consumption. All of this sounds like exciting news, but with issues like Android Apps refusing to install on Atom based devices still plaguing Intel, we do not see the mobile market jumping on the Atom bandwagon just yet. No official release date has been set, but we should see 22nm based devices on shelves by Q2 2014.
Alongside today's GTX Titan launch, NVIDIA announced a new integrated Tegra processor that features LTE. The Tegra 4i is the first chip from NVIDIA to feature LTE and should allow the GPU giant to enter the smartphone market. Previously, Tegra chips had been relegated to tablets due to the lack of an LTE radio.
NVIDIA claims that the new Tegra 4i is the fastest and most efficient mobile chip. Jointly developed by ARM and NVIDIA, it makes use of the R4 Cortex-A9 CPU for its four cores and features a fifth battery saver core. The GPU features 60 custom NVIDIA GPU cores, so graphics performance should be stellar for a mobile device.
"NVIDIA is delivering for the first time a single, integrated processor that powers all the major functions of a smartphone," said Phil Carmack, senior vice president of the Mobile business at NVIDIA. "Tegra 4i phones will provide amazing computing power, world-class phone capabilities, and exceptionally long battery life."
Along with the Tegra 4i announcement, NVIDIA debuted a reference smartphone platform known as "Phoenix." NVIDIA hopes that smartphone manufacturers will be able to quickly bring devices to market using the Tegra 4i. They can consult the reference device to enable quicker development and deployment.
NVIDIA CEO Jen-Hsun Huang has gone on record saying that we will not see its latest-generation and flagship mobile processor, the Tegra 4, ship until July. This will almost certainly push back launch dates of anticipated devices such as NVIDIA's own Project Shield.
During a conference call, Huang said:
"We will ship Tegra 4 starting in Q2. We are ramping production now and we will have full production release. The second quarter is when we ship to customers. Q2 is also when we ship the Tegra 4 based Shield device. Although it is in the latter part of Q2, it is going to be in Q2."
NVIDIA's Tegra 4 uses ARM's latest design, the Cortex-A15, which uses 45 percent less power than previous models. The Tegra 4 also features a GeForce GPU with 72 cores with up to six times the graphics performance of the Tegra 3.
We are still four or five months away from the launch of Haswell, Intel's next generation CPU architecture, and that puts us at a minimum of three to four months from seeing reviews begin popping up, right? Russian overclocking site OCLab.ru has posted what appear to be the first benchmarking results from a Haswell part. The results are interesting.
An Intel Haswell B0 stepping that was clocked at 2.8GHz was used, and they compared it to an Ivy Bridge clocked at 2.8GHz. The Haswell finished a Super PI 32M run in 11:27, while the Ivy Bridge completed the run in 11:49. And a WPrime 1024M run was completed by Haswell in 07:11.181 and the Ivy Bridge finished in 07:11.8.
As with every "leaked" bit of information, we must be careful not to read much into it. If these really are genuine Haswell results then Haswell is looking pretty good so far. Our sources have told us that we should read very little into this as if this is true, it is a very early engineering sample and the results might not be indicative of true Intel Haswell performance, as things such as drivers may not yet be fully optimized. If you would like to see the rest of the results, visit source #2 below.