#AI chips

610 stories · 85 in 7 days · 237 in 30 days
Oct 30
Oct 29
Oct 28
IndustrySelected17:07
Faraday Technology lands 2nm AI chip design win on Samsung process

The move from 4nm to 2nm on performance and power grounds signals that leading AI accelerator designs are now pushing past the point where older nodes suffice, and hybrid-bonded 3D DRAM stacking is entering mainstream custom silicon. Chip designers and data-centre hardware planners should track how Samsung's 2nm competes with TSMC's advanced nodes for AI workloads. Faraday's record-setting NRE also reflects rising design-service demand from US customers sourcing advanced packaging in Asia.

IT之家Source
Oct 27
Oct 26
Oct 24
Oct 23
Oct 22
IndustryUS vs ChinaMulti-sourceSelected71°19:33
SemiAnalysis benchmarks Huawei Ascend 950PR against Nvidia H100

The interesting split here is software versus hardware: Huawei's kernels achieve the best measured MFU on compute-bound shapes, but the memory subsystem and power efficiency put the chip at A100-class overall. Practitioners evaluating non-Nvidia inference hardware can see exactly where Ascend wins and where it lags. The gap to B300 also frames how far Chinese accelerators trail at the leading edge.

Event
Teortaxes@teortaxesTexSource
Oct 21