Current MPR Articles
May 30, 2016
Nvidia Brings Pascal to the Masses (MPR)
Nvidia has launched the GTX 1080, its first consumer GPU based on the Pascal graphics architecture.
Cypress Adds WICED to IoT Lineup (MPR)
To expand its IoT efforts beyond microcontrollers, Cypress has acquired Broadcom’s line of integrated Wi-Fi, Bluetooth, and ZigBee processors.
May 23, 2016
Intel Exits Mobile as Atom Bombs (MPR/MCR)
Intel has canceled all future Sofia products as well as the Broxton processor, officially ending its efforts to become a leader in the smartphone and Android tablet markets.
X-Gene 3 Performance Piques Interest (MPR)
According to AppliedMicro’s latest simulations, the forthcoming X-Gene 3 will match the per-socket and per-thread performance of a high-end Xeon E5v4 processor at a similar power level.
May 16, 2016
Octeon Expands Market With ARM (MPR/NWR)
Cavium has extended its 48-core ThunderX architecture across a broad range of embedded processors, offering the new ARM-based Octeon TX alongside the MIPS-based Octeon III.
Broadcom Offers First 64-Bit StrataGX (MPR/NWR)
Broadcom has launched its first 64-bit StrataGX SoCs, which offer four Cortex-A57 CPUs and a host of accelerators for realizing virtual network functions.
NXP’s Cortex-A72 Quartet (MPR/NWR)
Four new QorIQ processors, including the octa-core LS2088A and quad-core LS2048A, combine ARM’s best CPU with NXP’s second-generation packet-processing hardware.
May 9, 2016
Nvidia Hits HPC First With Pascal (MPR)
Based on the Pascal architecture, the 16nm Tesla P100 GPU uses 56 cores to deliver 5.3Tflops double-precision or 21.1Tflops half-precision while burning 300W.
NXP Economizes With LS2080A (MPR/NWR)
NXP’s newest QorIQ processors, the octa-core LS2080A and quad-core LS2040A, bring lower-power and lower-cost options to the ARMv8-based LS2-series.
Editorial: AMD Goes to China (MPR)
AMD is the latest chip vendor to partner with a Chinese company, but similar partnerships have not worked out well for foreign companies that share their valuable IP.
Convertible PCs Sail on Apollo Lake (MPR)
Apollo Lake is Intel’s next-generation entry-level processor for PCs. The 14nm device, which integrates the new Goldmont CPU, succeeds Cherry Trail for mini PCs and 2-in-1 convertible tablets.
May 2, 2016
HBM2 Doubles Bandwidth, Capacity (MPR)
The second-generation High Bandwidth Memory (HBM2) increases memory capacity and bandwidth, adds ECC support, and improves memory-controller efficiency.
April 25, 2016
NXP Debuts QorIQ Cortex-A72 (MPR/NWR)
The industry’s first embedded processors to use Cortex-A72, the quad-core LS1046A and dual-core LS1026A improve performance, particularly for mixed control- and data-plane networking.
AMD FirePro Reinvigorates HPC (MPR)
The new FirePro S9300x2 is the industry’s highest-performance single-precision GPU. The Radeon Open Compute tool chain adds support for C/C++ and Cuda software.
MediaTek’s Gateways to IoT (MPR)
The MT7697 combines ARM’s Cortex-M4F with integrated Wi-Fi and Bluetooth. The MT2511 is an analog companion chip for MediaTek’s wearable processors.
April 18, 2016
Baikal Aims for Desktops, Microservers (MPR)
Russian semiconductor startup Baikal recently announced its second design, a 28nm ARMv8 processor for desktop computers, communications equipment, and microservers.
April 11, 2016
Editorial: Cost Will Hamper 5G Rollout (MPR/NWR/MCR)
Without Moore’s Law to help, ultrafast 5G handsets will remain expensive, slowing the rollout of the next-generation cellular technology and diluting its potential benefits.
Silicon Labs Goes Wireless (MPR)
The new Wireless Gecko microcontrollers integrate a radio that supports Bluetooth, Thread, and ZigBee protocols, providing highly integrated solutions for low-cost IoT devices.
April 4, 2016
Broadwell-EP Moves Xeon E5 to 14nm (MPR)
Intel’s new socket-compatible upgrade to the Xeon E5 server platform provides modest performance gains on general-purpose software but greater improvements for specific workloads.
March 28, 2016
RISC-V Offers Simple, Modular ISA (MPR)
RISC-V is a new general-purpose instruction-set architecture that can be configured with a 32-, 64-, or 128-bit integer base and optional extensions.
New Intel SoCs Target Autos and IoT (MPR)
The Atom x3-M7272 and the Quark-based XMM 7315 are two new Intel processors with integrated modems for connected cars and cellular IoT applications.
Cortex-A32 Is First 32-Bit ARMv8 CPU (MPR/MCR)
The new Cortex-A32 is ARM’s first 32-bit CPU core to use the ARMv8 ISA, providing an upgrade path from the older Cortex-A5 and Cortex-A7 designs.
AMD Expands Embedded G-Series (MPR)
AMD introduced the Excavator CPU to its G-Series embedded lineup with new I and J processor families. The new low-cost LX family instead uses the Jaguar CPU.
March 21, 2016
Editorial: Clouds Part for ARM (MPR)
Microservices and hyperconvergence will lead data centers to adopt SoCs that are power efficient and that offer hardware acceleration.
Xeon D Doubles Core Count (MPR)
Intel recently released six new Xeon D server and embedded processors, packing 12–16 cores. New 65W variants offer enough compute performance to compete with the Xeon E5 line.
March 14, 2016
NXP’s Lowest-Power 64-Bit ARM (MPR)
NXP will soon sample the QorIQ LS1012A, which appears to be the world’s smallest, lowest-power embedded processor using the 64-bit ARMv8 architecture.
PowerVR 8XE Moves Down the Chain (MPR/MCR)
Imagination Technologies’ new PowerVR Series8XE family offers OpenGL ES 3.2 and Vulkan compatibility, improved performance, and reduced die size compared with the previous Series7XE.
ARM Revs Up Cortex-R8 for 5G (MPR)
To address the requirements of Release 13 of the LTE-Advanced specification as well as future 5G systems, ARM has added the new Cortex-R8 to its family of real-time embedded-processor cores.
March 7, 2016
Ceva’s New Gen-X DSPs Target 5G (MPR)
Ceva’s new X4 DSP core for baseband processors includes features that support the new 3GPP LTE-Advanced Release 13 (LTE-Advanced Pro) as well as future 5G modems.
Accelerating Machine Learning (MPR)
ISSCC papers from Intel and MIT indicate that hardware accelerators for machine-learning algorithms can deliver up to 30x better performance per watt than CPUs or GPUs.