CES 2014: NVIDIA Announces Tegra K1 SoC with 192 Kepler CUDA Cores, Denver ARMv8 Option

Subject: Processors, Mobile | January 5, 2014 - 11:43 PM |
Tagged: tegra k1, tegra, SoC, nvidia, kepler, CES 2014, CES

Update: Check out our more in-depth analysis of the Tegra K1 processor from NVIDIA.

Today during its CES 2014 press conference, NVIDIA announced the Tegra K1 SoC as the successor to the Tegra 4 processor.  This new ARM-based part includes 192 Kepler-based CUDA cores, sharing the same GPU architecture as the current GeForce GTX 700-series discrete graphics cards. 

k1-chip.jpg

NVIDIA also announced the Epic has Unreal Engine 4 up and running on the Tegra K1, bringing an entirely new class of games to mobile Android devices.  We got to see some demonstrations from NVIDIA running on the K1 and I must admit the visuals were stunning.  Frame rates did get a bit choppy during the subway demo of UE4 but it's still early.

As an added surprise, NVIDIA is announcing a version of Tegra K1 that ships with the same quad-core A15 (4+1) design as the Tegra 4 BUT ALSO have a version that uses two NVIDIA Denver CPU cores!!  Denver is NVIDIA's custom CPU design based on the ARMv8 architecture, adding 64-bit support to another ARM partner's portfolio.

denver3.jpg

Tegra K1 is offered in two pin-to-pin compatible versions - a 32-bit quad-core (4-Plus-1 ARM Cortex-A15 CPU) and a custom, NVIDIA-designed 64-bit dual Super Core CPU. This CPU (codenamed “Project Denver”) delivers very high single-thread and multi-thread performance. Both versions deliver stunning graphics and visual computing capabilities powered by the 192-core NVIDIA Kepler GPU. 

NVIDIA has only had Denver back for a few days from the fab but there able to showcase it running Android.  It's been a long time since the initial announcement of this project and its great to finally see a result.

dieshot.jpg

Tegra K1 with quad-core A15 processor

We'll have an in-depth story on the Tegra K1 on Monday morning, 6am PST right here on PC Perspective so check back then!!

Coverage of CES 2014 is brought to you by AMD!

PC Perspective's CES 2014 coverage is sponsored by AMD.

Follow all of our coverage of the show at http://pcper.com/ces!

Intel is stacking memory on top of the new Xeons

Subject: General Tech | November 22, 2013 - 12:29 PM |
Tagged: Supercomputing Conference, Intel, SoC, Near Memory, knights landing

Intel spilled more beans about the new Near Memory architecture that will be accompanying their new Xeon release.  The memory will be stacked directly onto the CPU giving much quicker access than you would normally see from DDR3 which has to travel over the motherboard.  They have not disclosed expected speeds, which could be up to what we see in current CPU caches only in much larger sizes.  This is not quite a Xeon SoC but in the presentation The Register heard of Intel's plans to incorporate optical fabrics and switches onto the CPUs as well with size being the only limit.  Perhaps they do have a leg to stand on when they claim the return to power of homogeneous computing.

xeon_plus_memory_stack.jpg

"According to an EE Times report, Intel's Rajeeb Hazra, a VP and general manager of its data centre group, said Intel would customise high-end Xeon processors and Xeon Phi co-processors by closely integrating memory, both by adding memory dies to a processor package and, at a later date, integrating layers of memory dies into the processor along with optical fabrics and switches."

Here is some more Tech News from around the web:

Tech Talk

Source: The Register

Imagination Technologies Unleashes Warrior MIPS P5600 CPU Core Aimed at Embedded and Mobile Devices

Subject: Editorial, General Tech, Networking, Processors, Mobile | October 19, 2013 - 01:45 AM |
Tagged: SoC, p5600, MIPS, imagination

Imagination Technologies, a company known for its PowerVR graphics IP, has unleashed its first Warrior P-series MIPS CPU core. The new MIPS core is called the P5600 and is a 32-bit core based on the MIPS Release 5 ISA (Instruction Set Architecture).

The P5600 CPU core can perform 128-bit SIMD computations, provide hardware accelerated virtualization, and access up to a 1TB of memory via virtual addressing. While the MIPS 5 ISA provides for 64-bit calculations, the P5600 core is 32-bit only and does not include the extra 64-bit portions of the ISA.

Imagination Technologies Warrior MIPS P5600 CPU Core.png

The MIPS P5600 core can scale up to 2GHz in clockspeed when used in chips built on TSMC's 28nm HPM manufacturing process (according to Imagination Technologies). Further, the Warrior P5600  core can be used in processors and SoCs. As many as six CPU cores can be combined and managed by a coherence manager and given access to up to 8MB of shared L2 cache. Imagination Technologies is aiming processors containing the P5600 cores at mobile devices, networking appliances (routers, hardware firewalls, switches, et al), and micro-servers.

MIPS-P5600-Coherent-multicore-system.png

A configuration of multiple P5600 cores with L2 cache.

I first saw a story on the P5600 over at the Tech Report, and found it interesting that Imagination Technologies was developing a MIPS processor aimed at mobile devices. It does make sense to see a MIPS CPU from the company as it owns the MIPS intellectual property. Also, a CPU core is a logical step for a company with a large graphics IP and GPU portfolio. Developing its own MIPS CPU core would allow it to put together an SoC with its own CPU and GPU components. With that said, I found it interesting that the P5600 CPU core was being aimed at the mobile space, where ARM processors currently dominate. ARM is working to increase performance while Intel is working to bring its powerhouse x86 architecture to the ultra low power mobile space. Needless to say, it is a highly competitive market and Imagination Technologies new CPU core is sure to have a difficult time establishing itself in that space of consumer smartphone and tablet SoCs. Fortunately, mobile chips are not the only processors Imagination Technologies is aiming the P5600 at. It is also offering up the MIPS Series 5 compatible core for use in processors powering networking equipment and very low power servers and business appliances where the MIPS architecture is more commonplace.

In any event, I'm interested to see what else IT has in store for its MIPS IP and where the Warrior series goes from here!

More information on the MIPS 5600 core can be found here.

Move over Intel, AMD has an announcement too

Subject: General Tech | September 10, 2013 - 04:10 PM |
Tagged: Steppe Eagle, SoC, Hierofalcon, GCN, Bald Eagle, amd, Adelaar

AMD have announced their new mobile roadmap and have changed their naming scheme drastically for these new processors.  The first of their ARM based processors will be called Hierofalcon and feature up to eight Cortex A57 processors capable of hitting 2GHz with a pair of 64-bit ECC DDR3 or DDR4 memory channels.  It will be a true SoC and feature both network and PCIe controllers as well as support for ARM's TrustZone.  Bald Eagle will have Steamroller cores and will be low TDP processors with a maximum of 35W and allow you to configure the maximum TDP to even lower levels if you so wish.  The final announcement dealt with the new GCN-based embedded series of GPUs called Adelaar which arrive in three different packages, a multi-chip module, a mobile PCIe module and a discrete GPU.  You can glean a bit more about these new families at DigiTimes.  

1_b.jpg

"AMD has disclosed its roadmap for the embedded computing market, as it becomes the first company to offer both ARM and x86 processor solutions for low-power and high-performance embedded compute designs. The new lineup includes two x86 accelerated processing units (APUs) and CPUs, a high-performance ARM system-on-chip (SoC) and a new family of discrete AMD Embedded Radeon graphics processing units (GPUs) expected to launch in 2014."

Here is some more Tech News from around the web:

Tech Talk

Source: DigiTimes

Microsoft Shows Off Xbox One SoC At Hot Chips 25

Subject: General Tech | August 29, 2013 - 11:22 PM |
Tagged: xbox one, SoC, microsoft, gaming, console, amd

At the Hot Chips conference earlier this week, Microsoft showed off several slides detailing the SoC used in its upcoming Xbox One gaming console.

The Xbox One uses a System on a Chip (SoC) designed by AMD’s Semi-Custom Business Unit. The processor features eight “Jaguar” AMD CPU cores, an AMD GCN (Graphics Core Next) based GPU with 768 shader cores, an audio co-processor, and 32MB of on-chip eSRAM.

The SoC, measuring 363mm^2 is manufactured on TSMC’s 28nm HPM process. The chip can interface with 8GB of DDR3 main memory with bandwidth of 68.3 GB/s or utilize the on-chip SRAM which has bandwidth of 102GB/s. The embedded SRAM is in addition to the smaller L1 and L2 caches. The slides indicate that the GPU and CPU can at least access the SRAM, though it still remains frustratingly unknown if the SoC supports anything like AMD’s hUMA technology which would allow the CPU and GPU to both read and write to the same memory address spaces without having to copy data between CPU and GPU-accessible memory space. It may be that the CPU and GPU can use the SRAM, but the same memory spaces can not be shared, though that may be the pessimist in me talking. On the other hand, there could be something more, but it’s impossible to say from the block diagram spotted by Semi-Accurate at the Microsoft presentation.

Microsoft Xbox One Gaming Console.png

With that said, the slides do reveal a few interesting figures about the SoC that were not known previously. The Xbox One SoC has 47MB of on-chip memory including 32MB eSRAM used by the CPU and GPU and 64KB of SRAM used by the audio co-processor. The chip’s GPU is rated for Microsoft’s DirectX 11.1 and above graphics API. Further, Microsoft rates the GPU at 1.31 TFLOPS, 41 Gigatexels-per-second, and 13.6 Gigapixels-per-second. Additionally, the GCN-based GPU is able to hardware-encode multi-stream H.264 AVC MVC video and hardware decode multiple formats, including H.264 MVC. The hardware encoder is likely being used for the console’s game capture functionality.

The audio processors in the Xbox One SoC use two 128-bit SIMD floating point vector cores rated at 15.4 GFLOPS and “specialized hardware engines” and “signal processing optimized vector and scalar cores.”

The final interesting specification I got from the slides was that the SoC is able to go into a low power state that is as low as 2.5% of the chip’s full power using power islands and clock gating techniques.

You can find all of the geeky details in these slides over at SemiAccurate.

Source: SemiAccurate

Leaked Bay Trail Product Lineup Reveals Atom, Celeron, and Pentium Branded Processors

Subject: General Tech | July 8, 2013 - 02:41 AM |
Tagged: valleyview, SoC, silvermont, pentium, Intel, celeron, Bay Trail, atom

A leaked Intel lineup reveals that the company's upcoming Bay Trail processors will also fall under not only the traditional Atom branding, but the Pentium and Celeron brands as well. The new lineup includes Bay Trail-D, Bay Trail-I, and Bay Trail M processors (note that Valleyview is the CPU codename, Bay Trail is the platform codename, with the CPU based on Intel's 22nm Silvermont architecture). The Bay Trail SoCs, which are based on the company's new 22nm Silvermont micro-architecture, include five processors in the Atom family, two in the Pentium family, and five processors that are part of the Celeron family.

All five of the Atom branded processors are Bay Trail-I chips. The leaked Atom lineup includes the following SKUs.

  • Atom E3810 (Bay Trail-I): Single core at 1.46 GHz with 400 MHz GPU and 5W TDP
  • Atom E3821 (Bay Trail-I): Dual core at 1.33 GHz with 533 MHz GPU and 6W TDP
  • Atom E3822 (Bay Trail-I): Dual core at 1.46 GHz with 667 MHz GPU and 7W TDP
  • Atom E3823 (Bay Trail-I): Dual core at 1.75 GHz with 792 MHz GPU and 8W TDP
  • Atom E3840 (Bay Trail-I): Quad core at 1.91 GHz with 792 MHz GPU and 10W TDP

Further, there will be one Bay Trail-M and one Bay Trail-D  Silvermont-based CPU under the Pentium brand. Specifications on those two chips are below.

  • Pentium N3510 (Bay Trail-M):  Quad core at 2 GHz with 750 MHz GPU and 7.5W TDP (4.5W SDP)
  • Pentium J2850 (Bay Trail-D): Quad core at 2.41 GHz with 792 MHz GPU and 10W TDP

Finally, the new Bay Trail-M and Bay Trail-D SoCs under the Celeron brand includes two quad cores and three dual core CPUs.

According to this PDF, the N2805, N2810, and N2910 Celeron CPUs will have an MSRP of $132, though it seems as though the N2805 should be cheaper than that since it has much lower specifications than the other two. The new Celeron-branded chips have the following specifications.

  • Celeron J1750 (Bay Trail-D): Dual core at 2.41 GHz with 792 MHz GPU and 10W TDP
  • Celeron J1850 (Bay Trail-D): Quad core at 2 GHz with 792 MHz GPU and 10W TDP
  • Celeron N2805 (Bay Trail-M): Dual core at 1.46 GHz with 667 MHz GPU and 4.5W TDP (sub-2.5W SDP)
  • Celeron N2810 (Bay Trail-M): Dual core at 2 GHz with 756 MHz GPU and 7.5W TDP (4.5W SDP)
  • Celeron N2910 (Bay Trail-M): Quad core at 1.6 GHz with 756 MHz GPU and 7.5W TDP (4.5W SDP)

Stay tuned to PC Perspective for more information on Bay Trail and Intel's first OoOE Atom micro-architecture as it develops.

Also read:

Source: Fanless Tech

ECS Reveals its First Motherboard with the AMD Kabini - KBN-I

Subject: Motherboards | June 19, 2013 - 02:10 PM |
Tagged: SoC, KBN-I, Kabini, ECS, A6-5200

ECS' new KBN-I board simply needs you to add some RAM and storage to become a fully functional computer.  The integrated 2.0Ghz A6-5200 with its accompanying HD 8400 graphics are enough power for general computing duties and perhaps even newer titles if you turn the settings down somewhat. If you do plan on gaming there is a PCIe 3.0 16x slot for an expansion card, though you might be better served with a TV capture card instead of a graphics card.   

kbn_color.png

19th Jun 2013, Taipei, Taiwan - Elitegroup Computer Systems (ECS) reveals its latest Mini-ITX motherboard family-KBN-I-featuring the world 1st Quad Core SoC (system-on-a-chip) processor with the newest “Jaguar” 28 nanometer architecture. The ECS KBN-I is integrated AMD Quad Core A6-5200 APU with AMD Radeon HD 8400 Graphics, support new generation DirectX 11.1 graphics, and offer 50% longer battery life performance. The ECS KBN-I with official Windows 8 WHCK certification (Windows Hardware Certification Kit) allows users to take full advantage of the touch browser and function-oriented design of Windows 8. The NM70-I is optimized for home, general computing productivity and multimedia applications.

Up to 25% More Power Efficiency
Designed to save space and energy, the ECS KBN-I is the ideal solution for small-form factor systems, with the AMD Kabini Quad Core SoC processor and maximum of 25W low power consumption. It improves a better power efficiency with up to 25% more power efficiency thru clock gating and unit redesign compared with current generation low power platform. Thanks to new generation design of increasing the amount of core, instructions per clock(IPC), and boost clock speed, the ECS KBN-I reducing energy usage by operating at lower frequencies when performance is not necessary. The ECS KBN-I also has 9W fanless Heat Sink design, full electrostatic discharge protection and 100% solid capacitors to maximize the reliability and longevity of your system.

Excellent Extension Support
Even with the mini-ITX size of the KBN-I, ECS has kept in mind the expansion needs of future-proof minded customers. The ECS KBN-I supports video and storage upgrades with 1 x PCI-E x16, 2 x SATA 6GB/s connections, 2 x USB 3.0 ports and 4 x USB 2.0 ports. For industrial applications, the KBN-I is equipped with Serial port output (COM). The Mini-PCIe slot is designed for laptops and other small-footprint computer systems and supports Mini-PCI card (one full-card and one half-card) for mini-Wireless, mini-TV tuner and so on. Moreover, users can enjoy high definition media as the KBN-I support HD 1080p output including HDMI and standard analog D-Sub VGA output. The NM70-I is bundled with additional software including Norton anti-virus, Muzee, Cyberlink Media Suite, and the ECS iEZ utility, which combines eBLU BIOS Live Update Utility, eDLU Drivers Live Update Utility and the eSF Smart Fan Utility.

Source: ECS

Samsung Galaxy Tab 3 10.1: Intel inside an Android?

Subject: General Tech, Graphics Cards, Processors, Mobile | June 3, 2013 - 03:00 AM |
Tagged: Intel, atom, Clover Trail+, SoC, Samsung, Galaxy Tab 3 10.1

While Reuters is being a bit cagey with their source, if true: Intel may have nabbed just about the highest profile Android tablet design win possible. The, still currently unannounced, Samsung Galaxy Tab 3 10.1 is expected to embed Intel's Clover Trail+ System on a Chip (SoC). Samsung would not be the largest contract available in the tablet market, their previous tablets ship millions of units each; they are a good OEM vendor to have.

Source: BGR India

Samsung is also known for releasing multiple versions of the same device for various regions and partners. The Galaxy Tab 10.1 and Galaxy Tab 2 10.1 did not have a variety of models with differing CPUs like, for instance, the Galaxy S4 phone did; the original "10.1" contained an NVIDIA Tegra 2 and the later "2 10.1" embed a TI OMAP 4430 SoC. It is entirely possible that Intel won every Galaxy Tab 3 10.1 tablet ever, but it is also entirely possible that they did not.

Boy Genius Report India (BGR India, video above) also claims more specific hardware based on a pair of listings at GLBenchmark. The product is registered under the name Santos10: GT-P5200 being the 3G version, and GT-P5210 being the Wi-Fi version.

These specifications are:

  • Intel Atom Z2560 800-933 MHz dual-core SoC (4 threads, 1600 MHz Turbo)
  • PowerVR SGX 544MP GPU (OpenGL ES 2.0)
  • 1280x800 display
  • Android 4.2.2

I am not entirely sure what Intel has to offer with Clover Trail+ besides, I would guess, reliable fabrication. Raw graphics performance is still about half of Apple's A6X GPU although, if the leaked resolution is true, it has substantially less pixels to push without being attached to an external display.

Maybe Intel made it too cheap to refuse?

Source: Reuters

NVIDIA's i500 SDR LTE Modem Achieves 150Mbps Throughput During CITA 2013 Demo

Subject: Mobile | May 22, 2013 - 07:46 PM |
Tagged: Tegra 4i, software defined radio, SoC, nvidia, i500, 4g lte

NVIDIA's Tegra 4i System on a Chip includes a software defined radio that works as a LTE modem. This i500 LTE modem uses general purpose deep execution processors (DXP) and is as much as 40% smaller than a hardware LTE modem according to the company.

At Mobile World Congress earlier this year, the modem was able to reach 100Mbps throughput. After a recent software update, the Tegra 4i SoC in NVIDIA's Pheonix reference platform achieved 150Mbps throughput in a demo at CITA 2013 in Los Angeles this week.

The reference phone was connected to a test network during the demo rather than a live cellular network. The cellular network test equiptment showed the Pheonix platform was connected at the full 150Mbps link speed. In addition to this, NVIDIA showed the Tegra 4i-powered Pheonix phone connected to a live AT&T LTE network streaming video and making voice calls.

The interesting bit about the i500 modem in the Tegra 4i is its software defined nature. NVIDIA was able to upgrade the modem's capabilites through software rather than needing to redesign the hardware. This would be a big plus to consumers as they would be able to take advantage of the faster network speeds as they become available without needing to replace their phones. NVIDIA did note that in addition to the LTE Cat 4 support, the i500 is also backwards compatible with LTE Cat 3, 3G, and 2G networks. I'm interested to see what the power consumption of thei500 is like compared to LTE modems implemented in specialized hardware. The i500 is smaller and more flexible, but SDR can use more power due to its general purpose hardware units.

Read more about NVIDIA's Tegra 4i SoC at PC Perspective!

Source: NVIDIA

Semi-Custom Business Unit Creates New Revenue Stream For AMD

Subject: General Tech | May 13, 2013 - 10:28 AM |
Tagged: x86, SoC, semi-custom chip, Patent, ip, APU, amd

Advanced Micro Devices (AMD) has an extensive intellectual property (IP) portfolio. The company has a range of products from CPUs and graphics cards to video acceleration hardware. It is also the only other major player to have a license to build chips with the x86 ISA. With the launch of its Semi-Custom Business Unit, AMD plans to take advantage of the engineering experience and patent portfolio to create a new revenue stream. AMD will work with other companies to create customized processors that integrate custom IP cores and technology but use AMD's existing products as a base to cut down on engineering time and R&D costs.

The first such customized chip is the System on a Chip used in Sony's PlayStation 4 gaming console. AMD intends to market its modular SoC technology and custom IP integration services to makers of set top boxes, smart TVs, tablets, PCs, networking hardware, and High Performance Computing applications. AMD argues that using its Semi-Custom Business Unit to create a customized SoC is cheaper and faster to design and produce than a fully-custom design, which makes sense since most of the engineering work is already done. AMD could stand to make quite a bit of extra money here, especially if it can land design wins for governmental and industrial design contracts. Intel's x86 license scarcity may actually benefit AMD here, in fact.

AMD's Semi-Custom Business Unit consists of an engineering team led by AMD Corporate Vice President and General Manager Saeid Moshkelani. I think doing this is a smart move for the x86 underdog, and it will be interesting to see how well the division does for the company's bottom line.

Source: AMD