Won-Joon Choi

It was so great to be back in Taiwan for #COMPUTEX2024 and share how Arm is enabling AI from the smallest edge devices all the way through to the cloud. Through Arm’s unparalleled pervasive compute platform, and the world’s largest software developer ecosystem, we expect more than 100 billion Arm devices ready for AI by the end of 2025.    There can be no question that AI will be a transformation not unlike the steam engine or the internet, but to unlock this untapped opportunity, developers must first get the necessary support to bring their innovations to life in an an easy and frictionless manner. That’s where Arm KleidiAI comes in. KleidiAI delivers best-in-class performance and efficiency to generative AI and large language models, helping developers seamlessly unleash the AI performance and capabilities of the Arm CPU platform, and in the process - accelerating the time to market.   AI will transform the technology landscape and only Arm's unrivalled leadership in power efficiency will allow us to meet the exponential growth in compute demands.   Watch the full keynote on YouTube here: https://lnkd.in/gYDfHSCt  

1,194

22 Comments

SK hynix: #GDDR7 Mass Production To Start in Q1'2025 Being a major JEDEC #memory standard, GDDR7 is slated to be produced by all three of the Big Three memory #manufacturers. But it seems that not all three vendors will be kicking off mass production at the same time. #SKhynix was at this year's #Computex trade show, showing off their full lineup of memory #technologies – including, of course, GDDR7. SK hynix is the last of the major memory vendor's we've seen promoting their memory, and fittingly, they seem to be the last in terms of their mass production schedule. According to company representatives at the show, the firm will not kick off mass production of their GDDR7 chips until the first quarter of 2025. Thanks again to Anton Shilov and AnandTech for the full article with more background and insights via the link below 💡🙏👇 https://lnkd.in/eEMBgD_N #semiconductorindustry #semiconductors #semiconductor #technology #tech #innovation #ai #semiconductormanufacturing #dram #mobile #it

62

2 Comments

semiStocks: The Week's Leaders, Laggards, Regional & Sector Performance                     ... for the week ending June 9, 2024 W/W Plc W/W Chg 7-Jun 1 23.4% Valens 2 15.7% Hua Hong 3 13.8% Navitas 4 13.1% Aehr Test 5 11.4% indie Semi 6 11.1% SMIC 7 11.1% Axcelis 8 10.3% NVIDIA 13.7% Average 83 -4.4% Hygon 84 -4.5% Microchip 85 -4.5% Montage 86 -5.6% Silicon Labs 87 -5.9% Loongson 88 -8.4% Impinj 89 -18.1% Lattice 90 -19.8% Semtech -8.9% Average W/W Chg 0.3% America 2.2% China 7.2% EMEA 0.4% Japan 3.6% Korea 2.6% Taiwan 0.2% Other Asia -2.0% Logic 3.1% AI 0.9% Auto 1.3% Mobile 5.0% Memory 1.2% Analog & Power 3.9% Foundry 2.5% OSAT Can you help me out? Please click the like button so I know these write-ups are of value for you. This for the greater good as I have no investments in these stocks. Also, no guarantees here. I am not a stock analyst. These are just observations and opinions. My opinions here are subject to many unknowns that will cause actual results and outcomes to be different from what is written. This is not investment advice and is not tailored to your specific investment needs. Do your homework and consult a professional investment advisor. #Semiconductor #StockstoWatch #stockmarket #techstocks

15

Go GaN - From power to RF: GaN’s journey to a US$4.35B market by 2029 !

15

#Samsung and #SKHynix engage in #LPCAMM battle over on-device #AI Samsung Electronics (Samsung) and SK hynix are both preparing to mass-produce the new concept of low-power #memory, #LPCAMM. Experts anticipate a rapid increase in #AI-enabled devices, including #smartphones, #notebooks, and #tablets, by 2025. According to #SouthKorea based media The Chosun Daily, Samsung and SK Hynix have scheduled test production for LPCAMM in the first half of 2024, with plans to begin mass production in the second half of 2024. LPCAMM is a high-capacity memory module incorporating multiple mobile DRAM #LPDDR5 #chips, positioned as a next-gen DRAM capable of replacing SO-DIMM. The key advantages of LPCAMM lie in its size and power consumption performance. The installation area can be reduced by up to 60%, and power consumption can be improved by up to 70%, enabling the core #computing system of on-device AI to reduce memory #chip power consumption. Industry experts believe that LPCAMM is currently the best substitute solution. If commercialized, LPCAMM could quickly replace SO-DIMM in the #notebook and #PC markets. Thanks again to Yu-Chun Lin , Charlene Chen and DIGITIMES Asia for the full article with more background and insights via the link below 🙏💡👇 https://lnkd.in/ehHmgp4K #semiconductorindustry #semiconductors #icdesign #ic #tech #ddr5 #it #computer #innovation #micron #mobile #ai #aiot #edge #technology

107

latency matters arm64 AmpereOne vs x86

38

Numemory reinvents Optane storage-class #memory 💡 Chinese fabless #semiconductor startup Numemory is developing storage-class memory (SCM) #chips using Optane-like phase change memory and crossbar #technology. Numemory, a business unit of Xincun Technology (#Wuhan), has launched the NM101, a storage-class memory #chip with 10x faster read and write speeds and 5x longer endurance than #NAND #chips. Xincun Technology was established in July 2022 and has developed a 3D selector-only memory (SOM) #technology using phase-change memory and a crossbar structure, which is remarkably similar to Intel Corporation and Micron Technology’s failed Optane #memory technology. Numemory’s website states: “The new memory is a type of SCM (storage-class memory) that can increase #storage capacity, reduce costs, and thus greatly improve the overall performance of the storage system. The new memory is considered to be a next-generation non-volatile storage technology that is very promising for large-scale mass production, and has become a hot topic of international concern in recent years.” Thanks again to Chris Mellor and Blocks and Files for the full article with more background and insights via the link below 💡🙏👇 https://lnkd.in/emUECvnU #semiconductorindustry #semiconductormanufacturing #technology #ai #tech #innovation #dram #nand #nandflash #flashmemory #server #computer #datacenter #foundry #it #computing #computer #ic

47

2 Comments

We are absolutely delighted to inform that the DVCon Taiwan 2024 Program Committee has select our paper abstract "Exploring Token-Based Strategies to Enhance Data Security and Memory Management in PCIe Devices". This is a defining moment, not only for Silicon Interfaces, but also for Portable Stimulus Standards (PSS) and Domain Specific Language (DSL), an Accellera standard, and also is the key that unlocks issues associated with security and data integrity for multi-Core environments, like RISC-V. That is something useful to the whole Semiconductor community! Please do join us at DVCon Taiwan Sep 10-11, 202,4 and hear how the SOCs can be more secure and have more data integrity. Narendra Modi Ashwini Vaishnaw Rajeev Chandrasekhar Amitesh Kumar Sinha India Electronics and Semiconductor Association indiasemiconductor forum VLSI Society of India (VSI) India Semiconductor Mission Vivek Tyagi Sumit Goswami Anirudh Devgan Jaswinder Ahuja Jitendra Chaddah (JC) Rajeev Khushu Vinay Shenoy Nishit Gupta Vivek Sharma Jensen Huang Lisa Su David W. Reed (芮德威) N. Chandrasekaran Banmali Agrawala Cristiano R. Amon Pat Gelsinger Aart de Geus Sassine Ghazi Raja Subramaniam Sanjay Mehrotra Nivruti Rai Hitesh Garg Lu Dai Dennis Brophy Penny Yang Ajay Singh Dr. Sanjay Arora Shamli Menon Neha Nigadikar IndoAmerican Chamber of Commerce, India Indo-German Chamber of Commerce Accellera Systems Initiative #dataintegrity #datasecurity #portablestimulus #pss #dsl #accellera #risc-v #multicore #data #semiconductors #microelectronics #vlsi #vlsidesign #verification #dvontaiwan

7

1 Comment

PCIe Protocol Series Part 1 - https://lnkd.in/gmvq5yA6 Part 2 - https://lnkd.in/g2ySkt95 Part 3 - https://lnkd.in/guG3vwtc Part 4 - https://lnkd.in/gsXn2wCy Part 5 - https://lnkd.in/gwxVVuQU Part 6 - https://lnkd.in/ghs2RvpH Quick insight into PCIe: History: PCIe was introduced in 2004 as a replacement for the aging PCI and AGP standards. Since then, it has become the de facto standard for connecting high-speed devices to a computer. Architecture: PCIe uses a point-to-point architecture, which means that each device is connected directly to the CPU (#cpu) via its own dedicated lanes, allowing for faster and more efficient data transfer. Speed: PCIe offers significantly higher data transfer rates than the earlier PCI and AGP standards, with a theoretical maximum bandwidth of up to 64GB/s for the latest PCIe 6.0 standard. Lanes: PCIe supports multiple lanes, with different numbers of lanes providing different levels of bandwidth. For example, a PCIe x1 slot provides a single lane and a maximum bandwidth of around 1GB/s, while a PCIe x16 slot provides 16 lanes and a maximum bandwidth of around 16GB/s. Compatibility: PCIe is backward compatible, which means that newer devices can be used with older motherboards, but they will be limited by the maximum bandwidth of the motherboard. Hot-swapping: PCIe also supports hot-swapping, which means that devices can be added or removed while the computer is running without the need for a reboot. Types: There are several types of PCIe slots, including PCIe x1, PCIe x2, PCIe x4, PCIe x8, and PCIe x16. These slots can be used for a wide range of devices, including graphics cards, sound cards, network cards, and storage devices. In summary, PCIe is a high-speed serial computer expansion bus standard that provides fast and efficient connectivity for a wide range of devices. Its point-to-point architecture, support for multiple lanes, and hot-swapping capabilities make it an essential component in modern computer systems. Join VLSI Jobs : Whatsapp Channel : https://lnkd.in/dHsz2TZh Download VLSI FOR ALL Community App : https://lnkd.in/dB_QDcHT FREE Mentorship : https://wa.me/918218599381 Call us on (+91)-9643070368 Apple Users App Link : https://lnkd.in/dA_7YmYJ Write us on info@vlsiforall.com #vlsiforall #vlsijobs #freshers #internship #vlsitraining #Semiconductor #Innovation #Technology #Beginners #PCIe #VLSI #Electronics #Engineering

62

#SouthKorea's #Chip Stockpiles Shrink at Fastest Pace in 10 Years - Demand for #memory #chips increases in line with #AI development - #Samsung, #SKHynix vie to become main supplier to NVIDIA South Korea’s #semiconductor inventories dwindled by the most since 2014, underscoring the pace at which demand is outstripping supply as customers rev up their purchases of devices needed to develop #artificialintelligence #technologies. Chip stockpiles dropped 33.7% from a year earlier in April in the largest decrease since late 2014, according to data released Friday by the national statistical office. It also marked the fourth month of declines in inventories, which coincided with a pickup in South Korea’s semiconductor exports. Thanks again to Sam Kim and Bloomberg News for the full article with more background and insights via the link below 💡🙏👇 https://lnkd.in/eKF5c37V #semiconductormanufacturing #semiconductorindustry #technology #it #datacenter #artificialintelligence #computing #computer #cloud #tech

90

1 Comment

https://lnkd.in/ghJBR_vf SK Hynix develops computing HBM to widen the AI memory gap for the late comers. SK Hynix is ​​entering the high bandwidth memory (HBM) market. It was discovered that a new concept, next-generation HBM with added calculation and communication functions is being developed. The strategy is to widen the gap with latecomers in the HBM market, where competition is heating up with differentiated technology and performance. Implementation of computing and network memory with systesemiconductors in HBM4, Significantly improved performance, power, size, etc. IP cooperation with fabless and foundry. #skhynix #nvidia #artificialintelligence #llms #generativeai #gpu #servers #datacenter #cloud #ai #Semiconductorindustry #electronics #semiconductor #technology #innovation #business credit: etnews

8

Live from CES - BrainChip debuts edge AI box with partner ecosystem for gesture, security and vision . BrainChip  has launched the Akida Edge AI Box, a compact AI/ML processing appliance for various industries. The Akida Edge AI Box is developed in partnership with VVDN Technologies and features an embedded Linux solution with multiple connectivity options including ethernet, bluetooth, and USB interfaces to provide a complete edge AI computing platform based on Brainchip’s Akida AKD1000 IC. Key partners include Edge Impulse for model training . https://lnkd.in/dDAJP4gp

81

2 Comments

Samsung Electronics is set to unveil its first 400-layer (V10) vertical NAND flash chip by 2026, aiming to secure leadership in the expanding storage market fueled by artificial intelligence. This roadmap, shared by Korea Economic Daily, outlines a strategy that mirrors the hybrid bonding technique previously adopted by competitors like YMTC and KIOXIA/WDC. Samsung’s approach, called bonding vertical NAND (BVNAND), could mark a significant leap in NAND flash performance. The integration of hybrid bonding enables separation of peripheral circuitry and memory arrays, optimizing thermal conditions to improve cell quality attributes like data retention, read disturbance, and current stability. Earlier adopters saw improved development timelines and memory performance, but cost concerns slowed adoption. However, with AI driving demand for high-performance storage, hybrid bonding in 3D NAND is expected to gain momentum rapidly. For clients interested in technical analysis, TechInsights’ "3D NAND – Samsung’s Path to 1XXX-Layer" report and Q3 2024 update feature detailed insights on YMTC and KIOXIA/WDC’s techniques. Thanks again to TechInsights of the Treasury for the full article with more background and insights click the source link in the comments below 💡 🙏👇 #Samsung #NANDFlash #3DNAND #MemoryTechnology #AI #Semiconductor #TechInsights #DataStorage #HybridBonding #Innovation

103

2 Comments

Phison Electronics Corps. Delivering "Made In India" aiDAPTIV+ Innovation from MiPhi Semiconductors Private Limited with Customisations & User Friendly Interfaces, Enabling Easier & Efficient AI Deployments in Data Centers, R&D Organisations, Educational Institutions and AI Deployers Tested and Approved by Leading Computing Governance Establishments of India and Deployed Across Top Technology Institutions & Research Establishments of the Country, Delivering Unimaginable TCO Savings on AI Deployments. #ai #artificialintelligence #aimodel #aitraining #aidatacenter #datacenter #machinelearning #techinnovation #GenerativeAI

9

1 Comment

Samsung Electronics 'New Product Requirements'... "Restore AI Semiconductor Competitiveness" Samsung Electronics has implemented a major executive reshuffle focused on its semiconductor operations as part of its 2025 strategy. Vice Chairman Jeon Young-hyun now leads both the Memory Division and Device Solutions Division, while Han Jin-man has been appointed head of the Foundry Division. This move aims to strengthen technological competitiveness in the high-bandwidth memory (HBM) sector, where Samsung currently lags behind rivals. Han Jin-man, drawing on his extensive experience within Samsung and in previous roles at Micron and startups, is expected to revitalize the foundry business, which faces significant challenges but also new opportunities. His leadership will focus on improving ties with major U.S. tech clients to enhance competitiveness against firms like TSMC. The creation of a Chief Technology Officer position in the Foundry Division further emphasizes Samsung's dedication to advancing semiconductor technology. Jeon Young-hyun will take on operational roles in the Memory Business Division in response to poor performance metrics from recent quarterly reports. Anticipated leadership changes may follow the reshuffle, yet the Device Experience Division seeks continuity, keeping leaders like Noh Tae-moon in place. Recent appointments in marketing and brand strategy reflect a renewed commitment to global brand presence and strategic oversight, showcasing Samsung’s efforts to adapt to a challenging market. #AI #SouthKoreaWant more analysis of major developments in AI, semiconductors, and cloud computing in Asia? Subscribe now to the Cognitive Asia newsletter--free of charge--at Substack https://lnkd.in/g4s6bqfi https://lnkd.in/ghYX54Mw

1

Samsung & SK Hynix stack LPDDR #DRAMs to cut #AI power consumption #SouthKorean #memory giants Samsung Electronics and SK hynix, in addition to High Bandwidth Memory (HBM), will also start the competition in #smartphone #LPDDR DRAM stacking. South Korean media Seoul Economic Daily reported that #Samsung aims to complete the research and development of "VCS DRAM" in the first quarter of 2025, and mass production in 2026. VCS memory is a product in which LPDDR DRAM is stacked upward into four layers like #HBM. #SKHynix, on the other hand, has invested in R&D under the name of "VFO," which is also a stacked LPDDR DRAM product and will be presented at the Semiconductor Packaging Technology Conference in the US, emphasizing that VFO will be the optimal solution for the #mobile market, where #wafers need to be thinner. LPDDR is a DRAM for mobile devices such as cell #phones and NBs, and its biggest feature is its low power consumption. Compared with HBM, which is a stacked DDR DRAM, stacking LPDDR will help increase #batterylife. Thanks again to Daniel Chiang , Judy Lin 林昭儀 and DIGITIMES Asia for the full article with more background and insights via the link below 💡🙏👇 https://lnkd.in/etTJWT5g #semiconductorindustry #semiconductors #semiconductors #technology #semiconductormanufacturing #innovation #computer #computer #chips

87