ARM en Nvidia Tegra K1 Claims Fame as First 64-Bit ARM Chip for Android <!--paging_filter--><h3><img src="/files/u69/tegra_k1.jpg" alt="Nvidia Tegra K1" title="Nvidia Tegra K1" width="228" height="163" style="float: right;" />Android enters the 64-bit ARM era</h3> <p>Say hello to <strong>"Denver," the codename for Nvidia's 64-bit Tegra K1 System-on-Chip (SoC), which also happens to be the first 64-bit ARM processor for Android</strong>. The new version of Nvidia's Tegra K1 SoC pairs the company's Kepler architecture-based GPU with its own custom-designed, 64-bit, dual-core "Project Denver" CPU, which Nvidia says is fully ARMv8 architecture compatible.</p> <p>So, what's special about this chip besides a 64-bit instruction set? Nvidia designed Denver to offer the highest single-core CPU throughput and industry-leading dual-core performance. Each Denver core (and there are two) sports a 7-way superscaler microarchitecture and includes a 128KB 4-way L1 instruction cache, a 64KB 4-way L1 data cache, and a 2MB 16-way L2 cache that services both cores.</p> <p>Using a process called Dynamic Code Optimization, Denver optimizes frequently used software routines at runtime into dense, highly tuned microcode-equivalent routines stored in a dedicated 128MB main-memory based optimization cache. This allows for faster access and execution, which translates into faster performance, in part because it lessens the need to re-optimize the software routine.</p> <p>Denver will also benefit Android platforms with new low latency power-state transitions. This is in addition to extensive power-gating and dynamic voltage and clock scaling routines based on workloads. The end result is more efficient power usage, which allows Denver's performance to rival even some mainstream PC-class CPUs at significantly reduced power consumption, <a href="" target="_blank">Nvidia says</a>.</p> <p>If you want to dig even further into the architecture, you can get more details <a href="" target="_blank">here</a>.</p> <p><em>Follow Paul on <a href="" target="_blank">Google+</a>, <a href="!/paul_b_lilly" target="_blank">Twitter</a>, and <a href="" target="_blank">Facebook</a></em></p> 64-bit android ARM cpu denver Hardware nvidia processor tegra k1 News Tue, 12 Aug 2014 17:32:21 +0000 Paul Lilly 28334 at ARM Strengthens Bottom Line Through Licensing Expansion <!--paging_filter--><h3><img src="/files/u69/arm_1.jpg" alt="ARM" title="ARM" width="228" height="144" style="float: right;" />More than 1,100 companies license ARM technology</h3> <p>While AMD and Intel were watching from the sidelines as the market transitioned to mobile, ARM was busy "earning and burning, snapping necks and cashing checks," to borrow a line from Step Brothers. It's a bit more competitive today, though ARM was able to gain a foothold in the mobile market and continues to ride the momentum. As such, <strong>ARM said it added 41 licensed customers to its portfolio</strong>, bringing the total number of licenses signed to more than 1,100.</p> <p>ARM reported total revenue of $309.6 million in the second quarter of 2014, up 17 percent from the $264.3 million it collected in the same quarter a year ago, and up 9 percent year-on-year. That resulted in a profit of $86.7 million for the quarter, and $126.7 million for the first half of the year.</p> <p>"Our continued strong licensing performance reflects the intent of existing and new customers to base more of their future products on ARM technology. The 41 processor licenses signed in Q2 were driven by demand for ARM technology in smart mobile devices, consumer electronics, and embedded computing chips for the Internet of Things, and include further licenses for ARMv8-A and Mali processor technology," ARM CEO Simon Segars <a href="" target="_blank">said in a statement</a>. "This bodes well for growth in ARM’s medium and long term royalty revenues."</p> <p>As more manufacturers jumped into mobile, ARM saw its licensing revenue jump 42 percent compared to last year. The company also shipped 2.7 billion ARM-based shipped in Q2, up 11 percent versus the same quarter a year ago. Shipment growth has been especially strong in enterprise networking and microcontroller markets, ARM said.</p> <p><em>Follow Paul on <a href="" target="_blank">Google+</a>, <a href="!/paul_b_lilly" target="_blank">Twitter</a>, and <a href="" target="_blank">Facebook</a></em></p> ARM cpu financial report Hardware license mobile processor News Tue, 22 Jul 2014 16:44:34 +0000 Paul Lilly 28212 at Cheaper Chromebooks in the Offing <!--paging_filter--><h3><img src="" alt="Chromebooks" title="Chromebooks" width="228" height="211" style="float: right;" /></h3> <h3>Use of inexpensive ARM SoCs could pave the way for sub-$200 Chromebooks</h3> <p>When Acer recently introduced the C720 Chromebook, a Haswell Core i3-toting device, we couldn’t help but wonder if users would be comfortable shelling out $350 or more for a Chromebook. This is an especially pertinent question because if there’s one thing that has helped these nifty little devices carve a niche for themselves, it is their greater affordability compared to entry-level Windows machines. The good news is that <strong>Chromebooks are likely to get even more affordable in the near future</strong>.</p> <p>This is because <a href=";utm_medium=referral&amp;utm_source=pulsenews" target="_blank">inexpensive ARM-based SoCs from the likes of MediaTek and Rockchip could end up powering future Chromebook models</a>. Taiwan-based MediaTek, in fact, has already begun its efforts to get in on the Chrome OS action.</p> <p>“A new experimental ARM Cortex A7 board named Moose has recently been added to the open-source <a href="" target="_blank">Chromium OS repository</a> by MediaTek,” Google’s François Beaufort revealed in a <a href="" target="_blank">Google Plus post Saturday</a>.</p> <p>With A7 chips from the likes of MediaTek being cheaper than even the ARM-based Exynos chips Samsung currently uses in a couple of its Chromebooks (the only company to do so), any device based on such SoCs will no doubt be cheaper than existing ARM- and X86-based Chromebooks, albeit while being considerably slower as well. However, all such dirt cheap devices are still quite far off — that is, if they show up at all.</p> <p>Follow Pulkit on <a href="" target="_blank">Google+</a></p> A7 ARM chrome os chromebooks chromium os Hardware laptops mediatek netbooks News Mon, 14 Jul 2014 05:23:14 +0000 Pulkit Chandna 28155 at Nvidia Tegra K1 Could Find Its Way Into Servers <!--paging_filter--><h3><img src="" alt="Tegra K1 Realistic Face Demo" title="Tegra K1 Realistic Face Demo" width="228" height="128" style="float: right;" /></h3> <h3>Expects 64-bit Tegra K1 variant to enter production ‘long before’ year’s end</h3> <p>Nvidia earlier this year <a href="" target="_blank">unveiled the Tegra K1 SoC (system-on-chip) with much fanfare, touting it as a “192-core super chip.”</a> There are two variants of the Tegra K1 : one that combines its 192-core Kepler GPU with a 32-bit quad-core CPU, and another that packs an 64-bit ARMv8-A architecture CPU (codenamed “Project Denver”) instead. While the former has already <a href="" target="_blank">begun shipping</a>, Nvidia CEO Jen-Hsun Huang told investors during the company’s recent Q115 earnings call that the <strong>64-bit variant is expected “to be in production long before the end of the year.”</strong></p> <p>During the call, Huang identified <a href=";p=qanda&amp;l=last" target="_blank">devices, automotive, and gaming as the main areas of focus of its mobile strategy</a>, prompting Stifel, Nicolaus &amp; Co. analyst Kevin E. Cassidy to question him about the likelihood of the 64-bit TK1 variant finding its way into micro servers. In his reply, Huang said there was no reason to rule it out.</p> <p>“I think we’re seeing a lot of interest in putting something like Tegra in micro servers, but one step at a time, one step at a time,” Huang said. “I think the most important thing about micro servers and putting TK1 in the server is really the software stack. And the software stack we’re building for GRID can very well eventually be used on top of Tegra as well.”</p> <p>It shouldn’t surprise anyone if Nvidia eventually turns its attention to servers with the Tegra K1, for its arch rival AMD is already doing it with its <a href="">Opteron A1100 64-bit ARM processors (codenamed “Seattle”)</a>.</p> <p>Follow Pulkit on <a href="" target="_blank">Google+</a></p> ARM grid nvidia tegra k1 project denver server News Mon, 12 May 2014 05:59:01 +0000 Pulkit Chandna 27790 at ARM Extends Reach in First Quarter, Posts 38 Percent Jump in Licensing Revenue <!--paging_filter--><h3><img src="/files/u69/arm_0.jpg" alt="ARM Scrabble" title="ARM Scrabble" width="228" height="174" style="float: right;" />ARM-based chip shipments totaled 2.9 billion in Q1</h3> <p>AMD and Intel have their work cut out for them as both companies try to wrestle mobile market share away from ARM. <strong>To kick off the first quarter of 2014, ARM Holdings reported total revenues of $305.2 million</strong>, up 16 percent compared to the same quarter a year ago. That includes a 38 percent year-on-year increase in processor licensing revenue derived from 2.9 billion ARM-based chip shipments in Q1, which itself is up 11 percent compared to last year.</p> <p>It's not just smartphones and tablets that are driving ARM-based processor shipments, but also servers, supercomputers, and other segments.</p> <p>"Q1 was a good start to the year for ARM, with more customers choosing to license ARM technology for their future products, which helped drive ARM’s revenues. Licenses are a precursor to future royalty revenues. Our customers are signing licenses with a view to designing ARM technology into an increasingly wide range of markets from servers and supercomputers to embedded sensors and enterprise networking applications and thereby underpinning ARM's future royalty opportunity," <a href="" target="_blank">noted ARM CEO Simon Segars</a>.</p> <p>Licensing is a big part of ARM's business strategy. Out of the $305.2 million in total revenues it collected in Q1, $129.9 million come from licensing its technologies to companies like Nvidia and Qualcomm. However, ARM isn't immune to market conditions and was affected by lower than expected sales of high-end smartphones running chips based on its IP. As a result, ARM's royalty revenue was $5 million lower than it was in the same quarter a year ago.</p> <p>"ARM’s royalty revenues in Q1 2014 were impacted by an inventory correction which particularly affected mobile and consumer electronics. As a result, both the year-on-year growth in ARM’s royalty revenues and the outperformance compared to overall industry growth are lower this quarter than seen in most recent periods," Segars added. "Inventory corrections of this type occur in the industry from time to time, the last one occurring in H1 2012."</p> <p>Another explanation is that the high-end smartphone market is starting to become saturated, though ARM is firm in its belief that the drop in royalty revenue is a temporary blip.</p> <p>Image Credit: <a href="" target="_blank">Flickr (Simon Cunningham)</a></p> <p><em>Follow Paul on <a href="" target="_blank">Google+</a>, <a href="!/paul_b_lilly" target="_blank">Twitter</a>, and <a href="" target="_blank">Facebook</a></em></p> ARM financial report mobile processor revenue News Wed, 23 Apr 2014 15:16:47 +0000 Paul Lilly 27678 at AMD's A-Series 64-bit ARM Server SoCs Begin Sampling; Launch Set for Fourth Quarter <!--paging_filter--><h3><img src="" alt="Opteron A1100" title="Opteron A1100" width="228" height="173" style="float: right;" /></h3> <h3>Opteron A1100 chips support up to eight 28nm Cortex A57 cores</h3> <p><strong>AMD began sampling its Opteron A1100 64-bit ARM processors (codenamed “Seattle”) last month</strong>, the chipmaker announced during its first quarter financial results conference call Thursday. Hailing it as a key milestone “in our ambidextrous strategy,” AMD CEO Rory Read said that the company planned to <a href="" target="_blank">begin shipping the chips in the final quarter of 2014.</a></p> <p>"We have introduced Seattle, our first 64-bit ARM server processor and the industry's first at 28nm technology, positioning AMD as the only SoC provider to bridge the x86 and ARM ecosystems for server applications," Read said. "We're catching it just as the wave is forming. This is going to be an important market over the next three, five, 10 years."</p> <p>Meanwhile, Lisa Su, AMD's SVP and GM of global business units, talked of the immense interest Seattle has managed to inspire in customers. But even she isn’t sure how much of that interest will actually translate into revenue for the beleaguered company. "Relative to revenue, it's probably too early to tell what's going to happen in 2015,” she said, “but I'd say the interest in the platform is quite high.”</p> <p>“I think the important thing for us and what we're working with the customers on is platform development and software development and ensuring that we get some of the ecosystem there.”</p> <p>When asked about Seattle finding its way into SeaMicro systems, Su had this to say: “One of the advantages of having a systems business is that we can do co-development between our chip development and our systems development. So it will be quite important for us to have Seattle in SeaMicro systems, and that’s in development.” However, she refused to put a date on the launch of such SeaMicro Fabric servers, clarifying that the “Q4 statement was a chip statement.”</p> <p>The upcoming 64-bit ARM-based server SoCs (system-on-chip) from AMD will be available with four or eight ARMv8-based Cortex A57 cores, up to 4MB of shared Level 2 cache, 8MB of shared Level 3 cache, eight PCI-Express Gen 3 lanes, two 10 GB/s Ethernet, and eight SATA 3 ports. Further, the <a href="" target="_blank">A-series chips support up to 128GB of DDR3 or DDR4 ECC memory as unbuffered DIMMs, registered DIMMs or SODIMMs. </a></p> <p>Follow Pulkit on <a href="" target="_blank">Google+</a></p> amd ARM opteron a1100 seattle server soc News Mon, 21 Apr 2014 05:39:28 +0000 Pulkit Chandna 27666 at ARM Unveils Cortex-A17 Processor for Mid-Range Mobile and Consumer Devices <!--paging_filter--><h3><img src="/files/u69/arm.jpg" alt="ARM" title="ARM" width="228" height="126" style="float: right;" />New chip from ARM takes aim at smartphones, tablets, smart TVs, and more</h3> <p><strong>ARM on Tuesday added to its mobile arsenal by announcing its Cortex-A17 processor</strong>, a mid-range part the company hopes will find its way into a number of different devices. According to ARM, the Cortex-A17 boasts a 60 percent performance uplift over Cortex-A9 processors, making it the fastest mid-range solution available. ARM also claims that it offers superior efficiency.</p> <p>"We expect to see a rich set of innovation in the mid-range mobile phone segment which is forecast to become a half a billion unit market annually from 2015 and the Cortex-A17 processor will be a key component in that growth," <a href="" target="_blank">said Ian Ferguson</a>, vice president of segment marketing, ARM. "To date, the ARM Partnership has shipped more than 50 billion ARM-based chips and the continued broadening of our processor family will enable our partners to further optimize their offerings in existing and new product categories."</p> <p>The 28nm Cortex-A17 is based on the ARMv7-A architecture with support for over 1 million apps. It's scalable up to four cores, each of which offers a full out-of-order 11+ stage pipeline. It also has a fully integrated, low-latency L2 cache controller. ARM says it expects frequencies to run north of 2GHz.</p> <p style="text-align: center;"><img src="/files/u69/cortex-a17.jpg" alt="Cortex-A17" title="Cortex-A17" width="608" height="509" /></p> <p>ARM sees the Cortex-A17 being paired with its high-efficiency Cortex-A7 as part of its big.LITTLE platform, which consists of higher-end processors for CPU intensive tasks and power efficient processors for longer battery life.</p> <p><em>Follow Paul on <a href="" target="_blank">Google+</a>, <a href="!/paul_b_lilly" target="_blank">Twitter</a>, and <a href="" target="_blank">Facebook</a></em></p> ARM cortex-a17 mobile processor News Tue, 11 Feb 2014 17:30:48 +0000 Paul Lilly 27239 at AMD Flexes Its First ARM Based Server SoC <!--paging_filter--><h3><img src="/files/u69/amd_development_board_0.jpg" alt="AMD Development Board" title="AMD Development Board" width="228" height="173" style="float: right;" />AMD's foray into ARM-based server SoCs begins with the Opteron A Series</h3> <p>A milestone has been reached in Sunnyvale less than a month into 2014. Chip designer <strong>AMD formally introduced its first 64-bit ARM-based server system-on-chip (SoC)</strong> previously codenamed "Seattle" and now called Opteron A1100. The chip is fabricated using a 28-nanometer process technology and is the first of its kind from an established server vendor. Along with the new SoC, AMD also unveiled a new development platform intended to make software design on the Opteron A1100 Series quick and easy.</p> <p>AMD's new SoC supports 4-core or 8-core ARM Cortex A57 processors and has up to 4MB of shared L2 cache and 8MB of shared L3 cache. Other features include configurable dual DDR3 or DDR4 memory channels with ECC at up to 1866 MT/second; up to four SODIMM, UDIMM, or RDIMM; eight lanes of PCI-Express Gen 3; eight SATA III ports; two 10-Gigabit Ethernet ports; ARM TrustZone technology; and Crytpo and data compression co-processors.</p> <p>"The needs of the data center are changing. A one-size-fits-all approach typically limits efficiency and results in higher-cost solutions," <a href="" target="_blank">said Suresh Gopalakrishnan</a>, corporate vice president and general manager of the AMD server business unit. "The new ARM-based AMD Opteron A-Series processor brings the experience and technology portfolio of an established server processor vendor to the ARM ecosystem and provides the ideal complement to our established AMD Opteron x86 server processors."</p> <p>AMD's new development kit is packaged in a micro-ATX form factor and includes an Opteron A1100 Series processor, four registered DIMM slots for up to 128GB of DDR3 RAM, PCI Express connectors configurable as single x8 or dual x4 ports, and eight SATA connectors. They're compatible with standard power supplies and can be used stand-alone or mounted in a standard rack-mount chassis.</p> <p><em>Follow Paul on <a href="" target="_blank">Google+</a>, <a href="!/paul_b_lilly" target="_blank">Twitter</a>, and <a href="" target="_blank">Facebook</a></em></p> amd ARM cpu enterprise Hardware opteron a1100 processor seattle Servers News Wed, 29 Jan 2014 15:58:54 +0000 Paul Lilly 27157 at Google Looking to Design Proprietary ARM Chips <!--paging_filter--><h3>A mammoth undertaking for Google</h3> <p><strong><img src="/files/u160391/google..jpg" width="250" height="141" style="float: right;" />Google </strong>and <strong>Intel </strong>have been sharing similar strategies for a while now, though that might be getting ready to be taken to the next level. Reports from <a href="">Bloomberg</a> indicate that Google is considering constructing their own ARM-based servers. </p> <p>Google opening up the floor to their own ARM chips does require some serious consideration on their part, even if it would mean putting them into direct competition with Intel. Like <a href="" target="_blank">ExtremeTech</a> projects, this wouldn't be a project Google would simply tackle without some sort of meticulously-detailed plan in place. </p> <p>Do you think Google will go ahead with the plan, or do you think they're barking up the wrong tree?</p> ARM chips Google intel news News Sun, 15 Dec 2013 05:15:16 +0000 Brittany Vincent 26890 at ARM Denies Claims Of Working On 128-bit Chips <!--paging_filter--><h3><img src="/files/u160391/arm.jpg" width="250" height="125" style="float: right;" />ARM has "no plans" for chips because they "aren't needed"</h3> <p>Rumors have buzzed surrounding <a href="" target="_blank"><strong>ARM Holdings'</strong></a> possible release of 128-bit chip designs to power various new smartphones. Most recently, via <a href="" target="_blank">PCPro UK</a>, the company was cited by the <a href="" target="_blank">Korea Herald</a> to promise 128-bit architecture "within the next two years."&nbsp;</p> <p>It wouldn't be surprising to see these chips being manufactured, since according to claims made by the Korea Herald, these types of chips would be all but required if smartphones are to feature more striking features like facial recognition and fingerprint scanning. Despite this story, however, ARM has denied all claims that they're hard at work on said chips.&nbsp;</p> <p>Chief marketing officer Ian Drew fired back with a blog post via ARM company blog, in which he states that there are "absolutely no plans underway for 128 bit ARM-based chips because they simply aren't needed."&nbsp;</p> <p>A statement like that leaves little room for speculation, but it's not as if there isn't space on the market for these types of chips. What's your take on the matter?</p> ARM blogs chip manufacturing news smartphones News Sat, 23 Nov 2013 23:35:48 +0000 Brittany Vincent 26757 at Your Tablet Benchmarks Suck <!--paging_filter--><h3><img src="/files/u154082/now_online_comments_december.jpg" alt="intel" title="intel" width="250" height="190" style="float: right;" />The mobile benchmarking scene is facing the same pitfalls as the PC experienced</h3> <p>At least that’s what <a title="Intel" href="" target="_blank">Intel</a> engineers are saying as the chip giant finally prepares to go toe-to-toe with <a title="arm" href="" target="_blank">ARM</a>-based tablets with its new “<a title="bay trail" href="" target="_blank">Bay Trail</a>” <a title="atom" href="" target="_blank">Atom</a> chips.</p> <p>In talks with the hardware press just before showing off Intel’s new Bay Trail part, principal Engineers Ronen Zohar and Francois Piednoel pointed out several “cringe” worthy issues they found in the source code for many benchmarks being used to test tablets today.</p> <p>The identities of the benchmarks weren’t disclosed but Zohar pointed out several issues with the source of popular benchmarks that don’t actually test what they claim to test. For example, one memory bandwidth benchmark didn’t even stress the tablet’s RAM. Another test used an unrealistic math function that the vast majority of research doesn’t match popular use.</p> <p>In another test, the developer hoped to create a CPU performance test but going through the source code with the media, Zohar said it was apparent the test didn’t do that. Instead the test only really tested how fast it could update the status bar.&nbsp;</p> <p>“Your CPU benchmark ‘tests’ how fast you can update the status bar and how fast you can update the clock,” Zohar said he learned from examining the app's source code.</p> <p>Perhaps worse than inadequate benchmarks is gaming of the tests by vendors. Zohar said in one example he witnessed, running the stock browser on a device and pointing it at a web-based browser resulted in the CPU ratcheting up to higher clocks.</p> <p>Those who have followed the PC benchmarking scene for ages will feel a sense of déjà vu all over again as the PC went through this too in the early days. &nbsp;Piednoel agrees there are echoes of the early days of the PC when benchmarking was a bit of a wild west.&nbsp;</p> <p>It’s not just Intel that believes this either.&nbsp;</p> <p>“I agree with Intel that mobile benchmarking has gotten completely out of hand, and, it does remind me of the 90's,” said analyst Pat Moorhead of Moor Insights &amp; Strategy. A former AMD exec turned analyst, Moorhead said mobile benchmarks have some maturing to do. He also agreed that cheating is rampant but also not quite as black and white as people make it out to be.</p> <p style="text-align: center;"><img src="" alt="antutu" title="antutu" width="580" height="326" /></p> <p style="text-align: center;"><strong>AnTuTu is a very popular mobile benchmark</strong></p> <p>Intel's hands aren’t entirely clean in this either. This summer the company was accused of cheating in the popular <a title="antutu" href="" target="_blank">AnTuTu</a> benchmark in showdowns with its <a title="clover trail" href="" target="_blank">Clover Trail</a>+ SOC. When run on the popular AnTuTu benchmark, ARM-based tablets would run the full calculation but when the benchmark was run on Clover Trail+ platforms, the benchmark would run the full calculation and then take a shortcut for the rest of the run.</p> <p>When asked to about the summer dustup over AnTuTu—after having just accused competing ARM vendors of benchmarketing—Zohar chalked it up to legitimate compiler optimization. But, Zohar said, the optimizations had been made many moons ago and made sense. He said Intel had no interaction with the AnTuTu developer except to provide a compiler which was optimized for Intel hardware—not exactly illogical. The optimizations didn’t fabricate numbers he said, the compiler just knew that if the code is asking for the exact same thing it asked for, there’s no reason to waste time and energy since you already know the answer. Why not just give the same answer? &nbsp;When AnTuTu was written to execute the same workload over and over again rather than take the compiler short cut, the Intel CPU actually trailed the ARM chip slightly.</p> <p>Just days ago AnTuTu Labs, the developer of the AnTuTu benchmark said it has <a title="antutu anti cheating" href="" target="_blank">implemented anti-cheating techniques</a> in AnTuTu X.&nbsp;</p> <p>Moorehead said he doesn’t think Intel’s framing of how it all unrolled doesn’t match what he knows but allegations of “cheating” isn’t exactly unusual.</p> <p>“I don't know of any relevant hardware company who hasn't been accused of cheating, particularly in CPUs and graphics,” Moorhead said. “The gray area is that one man's cheat is another's optimization. One man's piling on of resources in a benchmark consortium is interpreted as manipulation.”</p> <p>&nbsp;</p> <hr /> <p>&nbsp;</p> <p>For the most part, cheating on the PC has been mostly tamed by the move from the early days of purely synthetic benchmarks to an emphasis on “real-world” tests. The move was motivated when the benchmarking community began seeing driver optimizations that increased performance in benchmarks that actually hurt gaming performance. The theory behind the emphasis on real-world testing is that if a vendor is “optimizing” for a game, the end user still benefits. So call it cheating or optimizing, the result is still a better experience for the consumer. At least, that’s the theory. Reality doesn’t always match though.&nbsp;</p> <p style="text-align: center;"><img src="" alt="quake 3" title="quake 3" width="640" height="480" /></p> <p style="text-align: center;"><strong>More than a decade ago, ATI got in hot water for fudging performance numbers in Quake III Arena</strong></p> <p>One of the most famous cases of “optimizations” involved Quake III Arena. Tech site <a title="hardocp" href="" target="_blank"></a> found that changing the name of the executable from Quake3.exe to Quack3.exe would cause performance of the ATI Radeon 8500 to drop. When changed back, the performance would increase. Further testing by others found that the “optimization” appeared to be at the <a title="hardocp image quality" href="" target="_blank">cost</a> of image quality.&nbsp;</p> <p>ATI defended itself by saying that it was indeed an optimization made to give gamers the best combination of performance and visual quality but the fact that people still remember this more than 12 years later tells you how history remembers it.</p> <p>Years ago, Intel was also caught up in another benchmark brouhaha when it was found that applications compiled with Intel’s compiler didn’t use Streaming SIMD Extensions 2 (SSE2) properly on AMD CPUs that had the feature. The only way to enable the support on AMD CPUs was to make the application appear to be an Intel CPU that supported SSE2. The end result was even if an AMD CPU had SSE2 support, an application compiled with Intel’s compiler would run far slower using a different code path without SSE2 support. This, in fact, was an allegation of AMD’s anti-trust suit against Intel which both eventually agreed to settle with AMD receiving a $1.25 billion payment.</p> <p>But showing just how gray “optimizations” can be, defenders of Intel argue that the Intel’s C++ compiler is specifically designed for Intel CPUs and it’s not Intel’s job to validate AMD CPUs with a tool made to extract the most performance out of an Intel CPU. Others, of course, argue that Intel’s foot print on the industry is so large and if its compiler was violating Intel’s own guidance to explicitly check for CPU feature set support rather than just the CPUID string the only answer can be blatant cheating.</p> <p>On the PC though, these incidents are more the exception than the rule thanks to the bad PR that’s usually generated and a generally skeptical press. Reliance on using real-applications, such as how long it takes to encode a video using Handbrake, has also kept the benchmark controversies to a minimum lately.</p> <p>That’s not the same with tablets and smartphones right now. Samsung’s name was recently dragged through over allegations that the Galaxy S4 and Galaxy Note 3 were maxing out on cores and clock speeds—but only during popular benchmarks. This practice though apparently wasn’t confined to Samsung, <a title=" cheating" href="" target="_blank"></a> found multiple vendors were targeting benchmarks including Asus, HTC, LG as well as Samsung.&nbsp;</p> <p>As the first fingered for optimizing solely for benchmarks, Samsung has denied it’s intentionally trying to cheat, but only wants to give the highest performance when running stressful workloads. Afterall, when you’re running a test that’s supposed to measure an SOC’s theoretical performance, don’t you want the SOC to be running at maximum clock speed with all of the cores active?</p> <p>In the end, Intel argues that synthetic mobile benchmarks are still misleading.&nbsp;</p> <p>“It’s not because you have 25 potatoes that you have a good cell phone,” Piednoel said. &nbsp;“At the end of the day we are asking you to look at the new breed of benchmarks coming from benchmark vendors. Try to measure user experience, stop trying to measure potatoes.”</p> amd android antutu ARM ati benchmarks intel issue mobile problem samsung galaxy News Thu, 10 Oct 2013 23:41:56 +0000 Gordon Mah Ung 26334 at Samsung Doubles the GPU Performance in Updated Exynos 5 Octa Processor <!--paging_filter--><h3><img src="/files/u69/samsung_exynos_5_octa.jpg" alt="Samsung Exynos 5 Octa" title="Samsung Exynos 5 Octa" width="228" height="144" style="float: right;" />It's a GPU thing</h3> <p>Now that we have reasonably fast dual-core and quad-core processors powering our mobile devices, attention is starting to shift to the GPU to drive gaming and advanced graphics technology. Enter <a href=""><strong>Samsung's</strong></a> Exynos 5420, the newest addition to the company's Exynos 5 Octa family of eight-core processors, though that bit needs a little bit of explaining before we dive into the GPU side.</p> <p>Though Samsung markets the Exynos 5 Octa family as being eight-core processors, they're broken up into two quad-core segments using ARM's <a href="" target="_blank">big.LITTLE architecture</a>. In this particular case, what you have are four ARM Cortex A15 cores clocked at 1.8GHz and four Cortex A7 cores clocked at 1.3GHz. You never have access to all four cores at once, and instead tap into the faster A15 cores during heavy workloads, and the slower A7 cores for less intensive tasks. It's a battery saving measure that gives you a boost in performance when you need it.</p> <p>That said, the real story here is the six-core ARM Mali T628 GPU processor. According to Samsung, this is the GPU's first appearance, the upside being that it boasts 3D graphic processing capabilities over two times that of the previous generation Exynos 5 Octa processor.</p> <p>"Demand for richer graphic experiences is growing rapidly nowadays," <a href="" target="_blank">said Taehoon Kim</a>, vice president of System LSI marketing, Samsung Electronics. "In order to meet that demand from both OEMs and end users, we developed this processor which enables superb graphical performance without compromising power consumption."</p> <p>Outside of raw graphics performance, the Exynos 5420 boasts memory bandwidth of 14.9GB/s paired with a dual-channel LPDDR3 at 933MHz, delivering support for Full HD Wi-Fi displays.</p> <p>The new chip is currently sampling to customers and will go into mass production in August. With any luck, we could see new mobile devices built around the Exynos 5420 by the end of the year.</p> <p><em>Follow Paul on <a href="" target="_blank">Google+</a>, <a href="!/paul_b_lilly" target="_blank">Twitter</a>, and <a href="" target="_blank">Facebook</a></em></p> ARM exynos 5 octa gpu graphics Hardware mali t628 mobile processor samsung News Tue, 23 Jul 2013 16:45:28 +0000 Paul Lilly 25992 at Nvidia Following in ARM's Footsteps, Plans to License Graphics Technology <!--paging_filter--><h3><img src="/files/u69/nvidia_gpu_stage.jpg" alt="Nvidia GPU Stage" title="Nvidia GPU Stage" width="228" height="184" style="float: right;" />A potentially big move for Nvidia.</h3> <p>It's hard to argue against the success ARM has had in the mobile space. Arguably no other company has a larger footprint in the smartphone and tablet market, and it's all done through licensing its technology to third parties. Nvidia, whose own Tegra line is based on ARM's architecture, has been paying attention to ARM's business savvy and will soon begin licensing its Kepler GPU technology to other companies.</p> <p>So says <em>Reuters</em> based on a conversation it had with Nvidia chief Jen-Hsun Huang. As the landscape changes and trends toward mobile, Huang wants to be sure his company remains flexible and is willing to adapt.</p> <p>"The bottom line is the world has changed and we're expanding our business model to serve markets that we historically could not serve by selling chips alone," <a href="" target="_blank">Huang told <em>Reuters</em></a>.</p> <p>It's an interesting proposition and one that would inevitably pit Nvidia against ARM. As to whether Nvidia has any plans of licensing its GPU technology to Apple and Samsung, Huang played it coy saying it's early and his company has only begun interacting with the marketplace.</p> <p>"We will target customers where their capacity and desire to build their own application processors is great," Huang said.</p> <p><em>Follow Paul on <a href="" target="_blank">Google+</a>, <a href="!/paul_b_lilly" target="_blank">Twitter</a>, and <a href="" target="_blank">Facebook</a></em></p> ARM graphics jen-hsun huang license nvidia technology News Wed, 19 Jun 2013 14:59:58 +0000 Paul Lilly 25764 at 10 Things You Need To Know About Intel’s New Atom <!--paging_filter--><h3>How Intel's new Atom CPUs may be a game changer</h3> <p><span style="font-weight: bold;">10. It won’t suck.</span></p> <p><strong><a title="Intel Atom maximum pc" href="" target="_blank">Atom</a></strong> parts have long been the butt of our jokes for being the anti-performance parts that inspired the Netbook but anyone who ever tried to drive a Netbook for anything beyond browsing knows how much Atom’s sucked in performance. A dual-core, Hyper-Threaded 1.6Ghz Atom N2600 gives up a Cinebench 11.5 score of 0.47. That’s just barely faster than &nbsp;a single-core Athlon 64 3200’s score of 0.42. For reference, a Core i7-2600K gives up about 8.1 and a 3.2GHz Core 2 Duo E8200 gives you about 1.91. The actual performance isn’t known, but the new “<strong>Silvermont</strong>” version of Atom should offer far more performance than we've ever seen before.</p> <p style="text-align: center;"><img src="/files/u154082/5-6-2013_11-47-20_am.jpg" alt="intel atom" title="intel atom" width="620" height="361" /></p> <p style="text-align: center;"><strong><em>Meet the <a title="new Intel Atom CPUs" href="" target="_blank">new Intel Atom family</a></em></strong></p> <p><strong>9. No more process sloppy seconds</strong></p> <p>Atom has long been the bottom bin of <a title="intel Maximum PC" href="" target="_blank">Intel</a> CPUs. It didn’t get access to the latest process technologies and while Core 2 and Core i7 have been on a “tick tock” strategy where two new designs are produced on each process, Atom has plowed along with a one “new” design for each process. With Silvermont, the chip gets moved to the latest 22nm process 3D transistors. Even better for Atom will be the next-generation. With the introduction of the 14nm process, Intel will introduce a “tick” Airmont and then a “tock” chip that doesn’t even have a codename yet. This is just another sign of just how important Intel sees Atom to its future.</p> <p style="text-align: center;"><img src="/files/u154082/5-6-2013_11-47-58_am.jpg" alt="intel atom" title="intel atom" width="620" height="361" style="text-align: center;" /></p> <p style="text-align: center;"><strong><em>Intel Atom core block diagram</em></strong></p> <p><strong>8. Silvermont probably won’t bring back Netbooks.</strong></p> <p>The Netbook was the hottest piece of tech that was introduced at the right time. Who didn’t want a $300 mini-notebook when the entire financial world was collapsing? Unfortunately most users who bought them quickly became disenfranchised with the performance of the Netbook. In fact, Netbook sales numbers look like a failed rocket launch. Netbook sales peaked in 2010 with 32 million Netbooks sold, according to IHS. This year, 3.97 million Netbooks will sell with IHS predicting just 264,000 Netbooks will sell next year before Netbooks go extinct by 2015. Many blame the death of the Netbook on the iPad and other ARM-based tablets but we like to blame Atom. If Netbooks had had decent performance from day one, they might not have cratered so badly. Even with Silvermont and say, Core i7-lite performance, will OEMs try Netbooks again or have they had enough? We think OEMs have moved beyond the Netbook which is a bit of a shame because if they had decent performance years ago, maybe they wouldn’t be the Dodo bird of PCs.</p> <p style="text-align: center;"><img src="/files/u154082/5-6-2013_11-49-13_am.jpg" alt="intel silvermont" title="intel silvermont" width="620" height="336" style="text-align: center;" /></p> <p style="text-align: center;"><strong><em>Silvermont's wide range of operation</em></strong></p> <p><strong>7. Think of it as Atom i7 or Atom 2 Quad</strong></p> <p>Silvermont will be built around a modular design. Each module will feature two cores and Intel can stitch together up to eight cores on a die. Unlike AMD’s modular design that shares chip resources, Silvermont’s cores are separate cores that only share a common L2 cache. All previous Atom chips have continued to use the ancient front side bus to connect the chips but Silvermont will feature a point-to-point interface connecting to a system agent which will hook into the memory controller. Also important in Silvermont is the move from the in-order design of all previous Atoms to an out-of-order design. Out-of-order designs allow instructions to be executed out of order to greatly increase performance over in-order designs. For perspective, out-of-order CPU designs have been used by Intel since the Pentium Pro chip. Most ARM-based CPUs have also been in order up until the Cortex A9 chips.</p> <p>The main penalty to out-of-order designs has been an increase in power consumption and die space which is why Intel turned to an in-order design for the original Bonnell-based Atoms in 2008. &nbsp;Intel says with its advanced 22nm process, it can now do an out-of-order design while keeping power consumption and die space to a minimum. Intel has also completely redesign Silvermont with larger branch predictors, improved decoders, redesigned execution units, larger L2 and reduced L2 cache latency. In a nutshell, performance of Silvermont will be a factor of 3x over the current fastest Atom’s with 5x lower power consumption.</p> <p>Oh yeah, Silvertmont also gets SSE3.1, SSE4.2, hardware AES-NI encyprtion, hardware random number generation and several other instructions from the Westmere generation of CPUs.</p> <p><strong>6. Atom now gets Turbo too.</strong></p> <p>Intel’s popular and effective Turbo Boost makes an appearance in Silvermont which can now “burst” different cores up depending on the load. Intel says Atom also has the ability to run cores at different speeds as well. The company has previously pooh-poohed such an approach and still says it's more efficient to run all cores at the same speed when needed, but certain server, notebook, tablet and phone makers may want to intentionally run cores at asymmetric speeds to reduce power consumption sometimes.&nbsp;</p> <p style="text-align: center;"><img src="/files/u154082/5-6-2013_11-48-41_am.jpg" alt="atom cores" title="atom cores" width="620" height="314" style="text-align: center;" /></p> <p style="text-align: center;"><strong><em>Intel Atom Power Sharing</em></strong></p> <p><em>Click the next page to read about how it's faster and more efficient than ARM</em></p> <hr /> <p><strong>5. Yup. From servers to phones</strong></p> <p>Silvertmont will go into micro-servers and scale down to phone iterations after introduced. And no, you aren’t likely to get a tablet or phone with an 8-core Silverton variant. Those are likely aimed at micro-servers which aren’t as power sensitive as a phone or tablet.</p> <p style="text-align: center;"><img src="/files/u154082/5-6-2013_11-50-18_am.jpg" alt="atom tablets" title="atom tablets" width="620" height="361" style="text-align: center;" /></p> <p style="text-align: center;"><strong><em>Intel Atom in tablets</em></strong></p> <p><strong>4. No damnit, it’s not ARM</strong></p> <p>Analysts and self-appointed Internet experts have long speculated that Intel needs an ARM chip to compete with ARM because x86 can’t "get er done." For what it’s worth, Intel had an entire ARM division when DEC sold it the StrongARM family which turned into XScale. Intel sold XScale to Marvell in 2006 and apparently still has no regrets about it. Silvermont is pure x86.</p> <p style="text-align: center;"><img src="/files/u154082/5-6-2013_11-48-15_am.jpg" alt="intel atom module" title="intel atom module" width="620" height="242" style="text-align: center;" /></p> <p style="text-align: center;"><strong><em>Intel Atom workflow</em></strong></p> <p><strong>3. Faster and more efficient—than ARM</strong></p> <p>The battle of the last three years has clearly not been Intel vs. AMD, but x86 vs ARM. ARMchair commandos have long said x86 can’t compete because it’s just too power hungry. But remember, ARM is no brainiac chip. Even the super weak sauce old iterations of Atom have been performance and on power parity with ARM chips (non-believers see <a title="anandtech" href="" target="_blank">here</a>). Intel says Silvermont will easily stomp all ARM chips into the dirt. While, Intel didn’t actually directly say the ARM word during press briefings but you don’t have to be Steven Hawking to guess what CPU architecture Intel is comparing Silverton to. Even with the power consumption of ARM chips far exceeding Silvermont, those CPUs still can’t match Silvermont’s performance. Up against four competing ARM chips, Intel says at the same power use, Silvermont will be from 1.6 times to 2.3 times faster and consume from 3 to 5.8 times less power.</p> <p><strong>2. But the competition has eight cores.</strong></p> <p>Intel’s arrogance is pretty well known. But the truth is you can’t be arrogant if your products suck. That gloat has been fading of late but during our media briefings with Intel engineers, we could see the spring in their PowerPoint decks. We know, the proof is in the pudding, but Intel says Silvermont’s better cores out-perform competing ARM CPUs that use inefficient quad-cores. And yeah, that thing where they have eight cores? Remember four of the cores are low power cores that take over when the high performance cores aren’t needed. Intel says it's long explored such little core, big core approach and it’s never been optimal.&nbsp;</p> <p style="text-align: center;"><img src="/files/u154082/5-6-2013_11-50-00_am.jpg" alt="atom cores" title="atom cores" width="620" height="361" style="text-align: center;" /></p> <p style="text-align: center;"><strong><em>Intel Atom's cores</em></strong></p> <p><strong>1. Intel’s war begins with Silvermont</strong></p> <p>When did the x86 vs. ARM war begin? That’s hard to say. Some say 2010, others say 2011 or even 2012. We’d say that when the war began didn’t really matter. What matters is how each side reacts. Intel has long been a dangerous dragon tends to slumber when not challenged. Need proof? Go and Bing Pentium 4 or Intel’s lack of interest at the high-end desktop game today. &nbsp;So does anyone want to piss that dragon off? ARM does. Over the last few years it and a merry band of dwarves have walked into the slumbering dragon’s cave, stuck a short sword in its side and threatened the dragon’s hoard of gold. Silvermont—if Intel’s claims are true—could very well indicate that someone’s going to get roasted.</p> airmont ARM bonnell cpu Intel Atom New processor saltwell silvermont silverton tablet News Features Mon, 06 May 2013 19:57:42 +0000 Gordon Mah Ung 25488 at Intel Plans to Push Atom Prices Down Further to Compete With Arm <!--paging_filter--><p><img src="/files/u46173/intelatom.png" alt="Intel Atom" title="Intel Atom" width="228" height="155" style="float: right;" /></p> <h3>Intel refuses to surrender the lower-end of the market.</h3> <p>Years ago AMD was putting pressure on Intel to continue innovating on the high end, but fast forwarded to 2013 and <strong><a href="">Intel</a></strong> is the last man standing. The new war is in ultra-low powered chips, and the company is years behind. Intel’s response to ARM was the ATOM series of processors, but they were stuck trying to power a heavy and bloated Microsoft OS, while ARM had custom designed operating systems that extended battery life, and created an entirely new market. This year the two companies are destined to meet in the middle, and it will be a pivotal moment in the history of computing. Intel has<a href=""> announced its plans to compete with the current crop of dirt cheap ARM based devices</a>, and to the winner goes the spoils.</p> <p>Intel’s decision to take on the low end market isn’t without precedent, but it’s also a moving target. The price point for computing has been falling faster than industry costs, and now the company finds itself scrapping with the competition for a low margin market they aren’t even sure they want. Outgoing CEO Paul Otellini understands the value of winning against ARM, but it remains to be seen if his successor shares his passion.</p> <p>Microsoft announced last Thursday that small inexpensive form factor PC’s are part of its strategy, so it's likely Intel is just embracing the ever shifting winds of technological change. "We are...working closely with [PC makers] on a new suite of small touch devices powered by Windows. These devices will have competitive price points, partly enabled by our latest OEM offerings designed specifically for these smaller devices, and will be available in the coming months," Microsoft's chief financial officer, Peter Klein, said during the company's earnings call.</p> <p>Either way it sounds like $200 Android powered Intel tablets are a strong possibility this year, with $250/$300 Windows 8 based versions not far behind. Will they be a fad like Netbooks? Or are these the devices everyone has been waiting for?&nbsp;</p> <p><span style="font-style: italic;">Follow Justin on </span><a style="font-style: italic;" href="">Twitter</a>&nbsp;or on <a href="" target="_blank">Google+</a></p> android ARM atom Hardware intel mobile netbooks paul otellini tablets Windows windows 8 News Mon, 22 Apr 2013 02:46:08 +0000 Justin Kerr 25396 at