Recent #HBM news in the semiconductor industry

11 months ago
➀ Google is rumored to switch to TSMC's N3E process for Tensor G5; ➁ The report also clarifies that Google has chosen not to use 2nm technology for Tensor G6; ➂ The move could impact the competition in the AI and smartphone chip markets.
2nm3D IC3nmAIAI PCAI chipAMDASUSArmCPUChipletDRAMDellEDAGDDRGPUGaNHBMHPCInfineonLinuxNPUNVIDIAPCIePrivacyRaspberry PiSEMICONDUCTORSK hynixSSDSwitchTITSMCautomotivecoolingcybersecuritygamingiOSlaptopmemorymicrochipmonitorsoftware
11 months ago
➀ Arm is exploring the feasibility of running LLMs on mobile devices; ➁ Arm's optimization techniques for LLMs on mobile; ➂ The importance of practical use cases for LLMs in mobile devices
2nm3D IC3nmAIAI PCAI chipAMDASUSArmCPUChipletDRAMDellEDAEMIBEUVGDDRGPUGaNHBMHPCInfineonLinuxMobileNPUNVIDIAPCIePrivacyRaspberry PiSEMICONDUCTORSK hynixSSDSwitchTIautomotivecoolingcybersecuritygamingiOSlaptopmemorymicrochipmonitorsoftware
about 1 year ago
1. SK hynix is accelerating the development of HBM4 and HBM4E memory, originally planned for mass production in 2026 and 2027 respectively. 2. Now, HBM4 is set for mass production in 2025, and HBM4E will enter mass production in 2026, aligning with NVIDIA's accelerated AI accelerator release cycle. 3. The next-generation HBM4 offers a 40% increase in bandwidth and a 70% reduction in power consumption compared to HBM3E, which is the fastest memory in the world.
AIDRAMHBM
over 1 year ago
SK Hynix is planning to start mass production of HBM4E in 2026, a company research said.The South Korean memory maker originally planned to start mass production of the Gen 7 high-bandwidth memory (HBM) in 2027.But SK Hynix team leader and researcher Kwi Wook Kim said during International Memory Wee
HBMHynix