Realistically the biggest wild card would be some kind of extensive PyTorch support from Apple, which could alter model development workflows—but would not break Nvidia’s grip on foundation models.
HARDWARE
-
Foundation Models Converging on Nvidia Hardware Dependency
By
–
Foundation model developers—including at least one beyond Google/FB/OpenAI that I know of—will all be relying on Nvidia hardware going forward. The OSS/smaller model devopers might swerve (maybe w/ Apache TVM) but will also most likely rely on cuda-enabled hardware.
-
Nvidia’s ML Training Dominance Through A100/H100 and CUDA
By
–
With Nvidia going nuclear today would just re-up my last piece about how Nvidia has built a stranglehold on ML training with its A100/H100 series and, more importantly, Cuda’s deep connection with PyTorch.
-
Groq ISC 2023 Workshop: DataFlow Architectures for GNN Workloads
By
–
Join @GroqInc for an ISC 2023 Workshop Paper presentation exploring the use of dataflow architectures for GNN workloads on May 25 at 2 pm. In collaboration with @Argonne
, we will demonstrate how DataFlow architectures such as GroqChip™ overcome inefficiencies of traditional GPUs -
Analog Devices Stock Falls Amid China Sales Slowdown
By
–
Analog Devices drops eight percent, but CEO promises a ‘soft landing’ The main culprit is China, where sales deteriorated after a healthy start in February. https://
thetechnologyletter.com/the-posts/anal
og-devices-drops-eight-percent-but-ceo-promises-a-soft-landing
… $ADI #semiconductors #earningsseason #China #automotive #LifeSciences -
Apple Headset Strategy: Merrill Lynch Investment Analysis
By
–
Apple headset: don’t dismiss it, says Merrill Lynch How bad could it be? https://
thetechnologyletter.com/the-posts/appl
e-headset-dont-dismiss-it-says-merrill-lynch
… $AAPL #virtualreality #WWDC -
Cerebras Wafer-Scale Cluster Simplifies LLM Training
By
–
Large language models are extremely challenging to train. Our latest blog discusses how the Cerebras Wafer-Scale Cluster eliminates the need for huge compute budgets and complex distributed compute techniques. Read how we train models with a few clicks: https://
hubs.li/Q01R433-0 -
Tech Stock Winners Predicted by Major Analysts in AI Semiconductor Sector
By
–
Mammoth AI reports from Needham, Deutsche, others predict tech stock winners Picks such as Cerence and Vicor are interesting food for thought. https://
thetechnologyletter.com/the-posts/mamm
oth-ai-reports-from-needham-deutsche-others-predict-tech-stock-winners
… $NVDA $MU $CRNC $VICR $MPWR #artificialintelligence #AI #semiconductors #ChatGPT #investing -
Metis AI Platform Debuts at Embedded Vision Summit
By
–
Great first day at @EmbVisionSummit
! Swing by booth #604 to see our Metis #AI platform! https://
embeddedvisionsummit.com -
SambaNova Enables Complex Multi-Physics AI Simulations
By
–
"Multi-physics simulation is complex… The @SambaNova system helps us (
@Livermore_Lab
) create these cognitive simulations.” Coverage via @TheRegister https://
theregister.com/2023/05/23/lln
l_sambanova_ai_systems/
…
#hpc #ISC23 #ISChpc