AI Eats The World, And Most Of Its Flash Storage
… Just for fun, Matson walked us through the math that he recently used to explain the current situation to upper management at Solidigm. …
Search
Aiming At Hyperscalers And Edge, Nvidia Cuts Down To The A2 Accelerator
… The GA107 GPU was never great at the 64-bit FP64 processing that many AI training and HPC simulation and modeling workloads need, and the cut down GA107 GPU used in the A2 accelerator is really bad at FP64 math – and so by design, given its use case. …
Oracle’s Financing Primes The OpenAI Pump
… But having said all of that, here is the basic math: At $60 billion per gigawatt, a good, high-end ballpark figure for a datacenter tricked out with the best AI gear, that 4.5 gigawatts costs $270 billion. …
Microsoft Committed To Doubling AI Infrastructure In Two Years
… If you do the math, that means the OpenAI backlog is now $265.6 billion, down 5.6 percent sequentially. That math implies that OpenAI spent $15.65 billion on Azure capacity in Q3 F2026 – but I would be careful jumping to that conclusion. …
Building A Hassle-Free Way To Port CUDA Code To AMD GPUs
… With CPUs now getting both vector and matrix math units and with a shortage of GPU compute engines compared to the huge demand thanks to the generative AI boom, there is without question not only a need for alternatives to the CUDA parallel programming environment that Nvidia created as the core of… …
Cisco Wins Over AI Customers With Merchant Silicon And Optics
… Well, IBM actually just stopped talking about even AI orders, and probably because it is mostly a services play for Big Blue unless it starts counting Power Systems and Z mainframe revenues as AI just because its processors have native matrix math accelerators. …
Sometimes, Air Is The Only Way For AI Systems To Keep Their Cool
… What is also interesting in the presentation for the MI350P is that AMD is being perfectly honest about how it is positioned against not only the air-cooled system boards based on the MI350X and MI355X GPUs that are the flagship GPUs from the company, but how you position the MI350P against running… …
Building The Imperfect Beast
… The jump in performance of Mythos over Opus 4.6 on Humanity’s Last Exam perhaps not ironic, and not multiple choice but including problems models cannot solve and domain experts – meaning people – struggle with but can solve and on the Charxiv reasoning benchmark checks how models reason from chart… …
With TPU 8, Google Makes GenAI Systems Much Better, Not Just Bigger
… The TPU 8t TensorCore has a big wonking vector unit and two matrix math units, but there are some microarchitecture changes that are not obvious in the block diagram above. “TPU 8t is a powerhouse optimized for training,” explained Vahdat in his keynote. “We have redefined performance capability by… …
Contemplating Meta’s Homegrown MTIA Compute Engine Roadmap
… I happen to think that this insight, to make DLRMs more like LLMs, and whatever algorithms and math underlie the HSTU technique and its generative recommendation engine, are driving the MTIA AI compute engine effort at Meta Platforms. …