Google Builds Video-Transcoder Chip
To better support its massive YouTube business, Google has created its own video-transcoding chip. The VCU design offers a 40x performance gain compared with CPUs.
To better support its massive YouTube business, Google has created its own video-transcoding chip. The VCU design offers a 40x performance gain compared with CPUs.
Loihi 2 Advances Spiking R&D
Intel Labs has developed a second-generation neuromorphic processor using the Intel 4 process. It’s also offering the open-source Lava framework to advance neuromorphic software.
Intel Labs has developed a second-generation neuromorphic processor using the Intel 4 process. It’s also offering the open-source Lava framework to advance neuromorphic software.
Anton 3 Cracks Molecular Dynamics
Billionaire David E. Shaw has spent two decades quietly building highly specialized supercomputers. Anton 3 is his newest ASIC for accelerating molecular-dynamics simulations.
Billionaire David E. Shaw has spent two decades quietly building highly specialized supercomputers. Anton 3 is his newest ASIC for accelerating molecular-dynamics simulations.
Cadence Takes Tensilica AI to the Max
The new Tensilica AI lineup comprises three product tiers that Cadence calls AI Base, AI Boost, and AI Max, along with a new accelerator and a multicore model that can deliver up to 32 TOPS.
The new Tensilica AI lineup comprises three product tiers that Cadence calls AI Base, AI Boost, and AI Max, along with a new accelerator and a multicore model that can deliver up to 32 TOPS.
Samsung Demos In-Memory Processing
The company disclosed architecture and performance details for the processor-in-memory (PIM) version of its popular Aquabolt HBM2. The HBM-PIM, dubbed Aquabolt-XL, adds compute to a drop-in HBM2 stack.
The company disclosed architecture and performance details for the processor-in-memory (PIM) version of its popular Aquabolt HBM2. The HBM-PIM, dubbed Aquabolt-XL, adds compute to a drop-in HBM2 stack.
Dimensity 810 Trims Phone Power
The new Dimensity 810 boosts the single-thread CPU throughput by 17% compared with the Dimensity 800 while reducing power consumption, while the Dimensity 920 upgrades gaming performance.
The new Dimensity 810 boosts the single-thread CPU throughput by 17% compared with the Dimensity 800 while reducing power consumption, while the Dimensity 920 upgrades gaming performance.
Enflame Lifts Veil on Training Chip
The well-funded Chinese AI startup revealed architectural details of its first chip shortly after announcing its T20 second-generation design with 4x better inference performance plus TF32 for training.
The well-funded Chinese AI startup revealed architectural details of its first chip shortly after announcing its T20 second-generation design with 4x better inference performance plus TF32 for training.
Kompanio Powers 5G Tablets
MediaTek has added a new brand to its portfolio by introducing two tablet processors, refreshing its Arm compute strategy by reusing its latest Dimensity smartphone chips.
MediaTek has added a new brand to its portfolio by introducing two tablet processors, refreshing its Arm compute strategy by reusing its latest Dimensity smartphone chips.
Qualcomm Spills Cloud AI 100 Guts
Shortly after demonstrating outstanding performance per watt in recent MLPerf Inference results, Qualcomm revealed internal-architecture details of its Cloud AI 100 accelerator at last month’s Hot Chips.
Shortly after demonstrating outstanding performance per watt in recent MLPerf Inference results, Qualcomm revealed internal-architecture details of its Cloud AI 100 accelerator at last month’s Hot Chips.
SOT MRAM Improves Access Speed
Startup Antaios is leading the commercialization of spin-orbit-torque (SOT) magnetoresistive RAM (MRAM), which promises greater longevity and faster access times than previous MRAMs.
Startup Antaios is leading the commercialization of spin-orbit-torque (SOT) magnetoresistive RAM (MRAM), which promises greater longevity and faster access times than previous MRAMs.
Cerebras Performs Extreme Training
The company hopes to meet the demands of huge neural networks by offering its new MemoryX and SwarmX systems, which aim to reduce GPT-3 training time to less than one day.
The company hopes to meet the demands of huge neural networks by offering its new MemoryX and SwarmX systems, which aim to reduce GPT-3 training time to less than one day.
Tesla D1 Tackles AI Training
In a surprising move, Tesla has developed its own AI training accelerator. The D1 design crams 25 chips into an enormous wafer-scale package that generates an industry-leading 9 petaflop/s.
In a surprising move, Tesla has developed its own AI training accelerator. The D1 design crams 25 chips into an enormous wafer-scale package that generates an industry-leading 9 petaflop/s.
Alder Lake Extends Battery Life
Alder Lake is Intel’s first mainstream PC processor with a hybrid architecture, combining big and small cores to increase both performance and battery life.
Alder Lake is Intel’s first mainstream PC processor with a hybrid architecture, combining big and small cores to increase both performance and battery life.
Mount Evans Puts Intel in DPU Range
The company recently disclosed architectural details of its first data-processing unit, combining a control plane and a programmable data plane to handle Ethernet speeds up to 200Gbps.
The company recently disclosed architectural details of its first data-processing unit, combining a control plane and a programmable data plane to handle Ethernet speeds up to 200Gbps.
Sapphire Rapids Spans Tiles
The next-generation Xeon processor is Intel’s biggest leap forward in years, delivering greater core counts, more per-core performance, integrated accelerators, and the latest memory and I/O interfaces.
The next-generation Xeon processor is Intel’s biggest leap forward in years, delivering greater core counts, more per-core performance, integrated accelerators, and the latest memory and I/O interfaces.
