AMD has long signaled that embedded chips will play a major role in its future, and it’s backing up that claim by providing a glance at its 2014 roadmap. The highlight is Bald Eagle, a 35-watt x86 processor designed for demanding tasks like gaming; it should include up to four Steamroller cores, and it will optionally sport on-chip Graphics Core Next video. Devices that need even more visual power will use Adelaar, a dedicated graphics chipset that includes both GCN and 2GB of built-in memory. It’s reportedly fast enough to be useful for PC video cards, not just the usual set-top boxes and smart TVs.
Two system-on-chip designs are also joining AMD’s lineup. Hierofalcon is built with data centers in mind, and carries up to eight ARM Cortex-A57 cores; Steppe Eagle, meanwhile, is a combination of upgraded Jaguar x86 cores and GCN that should speed up AMD’s low-power G-series processors. With the exception of Hierofalcon, all of the new embedded chips should be available in the first half of next year. AMD hasn’t named any early customers, but its embedded silicon tends to reach products that you’d recognize. Check out the roadmap after the break.
Samsung’s Exynos 5 Octa processors have thus far been of a variety that allows for four of the eight cores to be used at one time. This will no longer be the reality beginning in the fourth quarter of this year, however, with Samsung announcing that with the use of Heterogeneous Multi-Processing the Exynos 5 […]
When MediaTek announced that it would be producing true eight-core mobile processors later this year, we knew it was only a matter of time before its main rival Qualcomm chimed in. As illustrated by a set of guitar amplifiers, the San Diego gang explains that while they rebuild their CPU cores for each generation (the latest architectures being Krait 300 and Krait 400), they claim that “Our Competitor” — which is labeled with the same font and colors as MediaTek’s logo — simply “chooses to duplicate the same old cores” based on ARM’s slower Cortex-A7 architecture. That said, it’s worth a reminder that Qualcomm’s cheaper Snapdragon 400 range also uses Cortex-A7.
Later on in the video, Qualcomm uses a Guitar Hero-like visualization to compare the performance difference, as well as show how octa-core is overrated for most apps. Apparently only 17 out of the top 20 Android apps in China use two cores at most, hence the bare fretboard for the octa-core side. The Snapdragon side, meanwhile, combs through a denser bunch of apps at a higher speed. Of course, there’s bound to be some bias here, so only time will tell how close to reality this argument is. Until then, enjoy the cheeky clip after the break.
It might not just be your ears that are stuffed with wax for long—because researchers from the University of Michigan want to pack your phone full of the stuff, too.
It’s a rare occurrence to see Google, NVIDIA and IBM working lockstep towards a common goal, but the aforesaid trio has joined Mellanox and Tyan in order to launch a development group for data centers. The aptly-titled OpenPOWER Consortium is hailed as a “development alliance based on IBM’s POWER microprocessor architecture.” We’re told that it intends to concoct “advanced server, networking, storage and GPU-acceleration technology aimed at delivering more choice, control and flexibility to developers of next-generation, hyperscale and cloud data centers.” Sounds pretty bold, and it’ll require IBM to offer up open-source POWER firmware to those participating. Moreover, NVIDIA and IBM will be jointly working to integrate the CUDA GPU and POWER ecosystems, but beyond that, it’s not entirely clear what the immediate impact on mankind will be. You’re more than welcome to take stabs in comments, though.
It’s a material world, and Samsung’s just living in it. And, evidently, it’s tired of doing things in the same manner it has been. According to a report from ETNews, the aforesaid company is looking to produce an Exynos processor using a “redesigned ARM core platform” that’ll be whipped up in Samsung’s Austin, Texas-based R&D facility. If the plans materialize, it’ll mark the first time that Samsung has engineered an Exynos chip “based on its own architecture platform by redesigning the ARM cores.” For those unaware, Qualcomm and Apple are amongst the precious few that have concocted their own platforms after inking an architecture licensing deal with ARM, and it sounds as if Sammy’s tired of being on the outside looking in. The report also states that development is expected “to be completed early next year,” and you can bet that future Galaxy and Note products will be the first to benefit.
This week NVIDIA is once again blurring the lines between desktop and mobile graphics with a note on the introduction of Kepler technology into their next-generation mobile processor. NVIDIA suggests that, “from a graphics perspective, this is as big a milestone for mobile as the first GPU, GeForce 256, was for the PC when it was introduced 14 years ago.” This is the first set of details we’re getting on Project Logan, the next processor architecture in the Tegra chipset family.
You’ll remember the comic book character collection of code-names for the processors that’ve become the Tegra 3 and Tegra 4 – and what we must assume will be the Tegra 5 as well. Here with what’s still called Project Logan, NVIDIA makes clear their intent to bring graphics processing abilities until now reserved for desktop machines to the mobile realm; for tablets, smartphones, and everything in between.
In addition to deploying Kepler’s efficient processing powers to the Logan mobile SoC, NVIDIA intents on bringing the excellence in a form that the company will be able to license to others. This licensing was outlined earlier this year amid the latest Kepler integrations into GPUs such as the NVIDIA GeForce GTX 760.
NVIDIA suggests that the technology deployed with mobile Kepler is able to use one-third the power of “GPUs in leading tablets, such as the retinal iPad”, while it performs identical renderings. They also note that this efficiency is achieved with mobile Kepler without compromising graphics capabilities, working with OpenGL ES 3.0, OpenGL 4.4, and everything else in the OpenGL universe.
Though we’re expecting this architecture to hit Google’s Android – as NVIDIA has been hitting for the past several years – they do mention that the technology also supports DirectX, the latest graphics API from Microsoft. Think Windows RT and Windows 8 – NVIDIA’s been there before.
Working with these current and next-generation APIs allow NVIDIA to bring on graphics unlike any seen in the mobile universe, developers taking hold of these environments with a variety of high-end rendering and simulation techniques. NVIDIA runs down three of the most powerful:
Tessellation – which creates geometry dynamically and efficiently on the GPU from high-level descriptions, sizing triangles optimally based on the user’s viewpoint. By comparison, fine detail in a traditional pre-generated approach is inefficient, requiring excess geometry to deal with all possible viewpoints.
Compute-based deferred rendering – which calculates the effect of all lights in the scene in a single deferred rendering pass. This OpenGL 4 capability greatly improves deferred rendering efficiency and scalability compared to current OpenGL ES based implementations, which require an extra pass for each light source in the scene. The scalability of the compute-based approach paves the way to even more advanced lighting models, such as using virtual points of lights to approximate global illumination effects.
Advanced anti-aliasing and post-processing – which deliver better image quality, particularly in areas of very sharp color contrast, by making multi-sampling more programmable and allowing applications to implement their own anti-aliasing filters. These also enable more efficient film-quality post-processing effects, such as motion blur and depth of field.
NVIDIA makes clear that a lovely collection of processing-heavy tasks will be able to be carried out with this next-generation solution including computer vision, augmented reality, computational imaging, and speech recognition. Showed off this week at Siggraph was a return of the digital head now known as “Ira”, aka Faceworks.
Stick around as we continue to jump deeper into the next big superhero-themed processor, one that’ll break barriers beyond what we’re only just seeing now with the NVIDIA Tegra 4 – living inside NVIDIA SHIELD and getting pumped up for benchmarks sooner than later!
Intel teased us with the prospect of a fanless Haswell chip back at Computex, but it didn’t say just how it would achieve such a feat. Now we know: it’s introducing more energy-efficient versions of the Y-series Core processors that were announced earlier this year. The new chips consume as little as 4.5W in a typical scenario, letting them run in tablets and detachable convertibles without the fans needed by their 6W peers. Don’t expect blistering performance at this reduced power level, however. While Intel isn’t divulging clock speeds just yet, the 4.5W Y-series chips have the same 11.5W thermal design power rating as their 6W siblings — they’ll still need active cooling to perform at their best. The company also isn’t providing ship dates or naming customers, although it does promise that the extra-miserly Core CPUs should be available in the “coming months.” We have an idea as to who might be interested.
Last week, Samsung teased the update to its Exynos 5 Octa processor, promising that more details would be arriving this week. It’s only Monday, and those details have already been published, showing more power than the current iteration and the use of a six-core ARM Mali-T628 GPU. We’ve got the rest of the details after the jump.
The Exynos 5 Octa processor was the first to utilize ARM’s big.LITTLE architecture, featuring eight processing cores with four of them being utilized at a time. Four of the cores are Cortex-A15 offerings, while the others are Cortex-A7 cores. Depending on the needs of the device at any given time, either the more powerful four cores are used, or the lesser powerful.
The new Exynos 5 Octa offering still follows this, but both sets of cores have seen a boost in clock speed. The quad-core Cortex-A15 in the original has a clock speed of 1.6GHz, while the new version offers 1.8GHz. The A7 cores in the original were 1.2GHz, while the new chipset’s cores are set at 1.3GHz. This represents a 20-percent increase in power over the original version.
Some other changes includes 14.9GB/s memory bandwidth alongside a dual-channel LPDDR3 at 933MHz, which Samsung says allows for rapid data processing, as well as support for full HD wifi display. For playing and recording 1080p video, the processor offers full HD 60fps video hardware codec. There’s support for OpenGL ES 3.0 and Full Profile Open CL 1.1.
ARM’s Media Processing Division Executive VP and General Manager Pete Hutton said: “ARM welcomes the latest addition to the successful Exynos Octa 5 series, which uses ARM’s Mali GPU solution to dramatically improve graphics performance. ARM big.LITTLE™ and ARM Artisan® Physical IP technologies continue to be at the heart of the Octa series and now complement the new functionality brought by ARM GPU Compute. This combination enables unprecedented capabilities in areas such as facial detection and gesture control, and brings desktop-quality editing of images and video to mobile devices.”
Samsung couldn’t help itself last week when it teased a new Exynos 5 Octa system on a chip, and now it’s dishing out the full details. The fresh 5420 variant of the SoC is based on Mali-T628 MP6 silicon, packs a quartet of ARM Cortex-A15 cores running at 1.8GHz and four 1.3GHz Cortex-A7s in an ARM big.LITTLE configuration. Seoul claims that the package packs 20 percent more CPU processing punch, and has two times greater 3D graphics power than its predecessor. Dual-channel LPDDR3 at 933MHz gives the processor a screaming memory bandwidth of 14.9 GBps, which lends it full HD WiFi display support. Baked inside is an image compression solution that makes for energy efficient multimedia loading, and squeezes out more hours of use with high-res displays. There’s no word on which devices might use the new SoCs, but the chips are already being sampled by Samsung’s customers, and mass-production is slated for August.
This is site is run by Sascha Endlicher, M.A., during ungodly late night hours. Wanna know more about him? Connect via Social Media by jumping to about.me/sascha.endlicher.
