Tag: GPU

Source URL: https://www.theregister.com/2024/10/30/ualink_consortium_incorporated/ Source: The Register Title: No-Nvidias networking club convenes in search of open GPU interconnect Feedly Summary: Ultra Accelerator Link consortium promises 200 gigabits per second per lane spec will debut in Q1 2025 The Ultra Accelerator Link Consortium – an alliance of enterprise tech vendors that pointedly excludes Nvidia because it wants…

The Register: AMD teases its GPU biz ‘approaching the scale’ of CPU operations

Oct 30, 2024

—

by

Source URL: https://www.theregister.com/2024/10/30/amd_q3_2024/ Source: The Register Title: AMD teases its GPU biz ‘approaching the scale’ of CPU operations Feedly Summary: Q3 profits jump 191 percent from last quarter on revenues of $6.2 billion, helped by accelerated interest in Instinct AMD continued to ride a wave of demand for its Instinct MI300X AI accelerators – its…

Slashdot: OpenAI Builds First Chip With Broadcom and TSMC, Scales Back Foundry Ambition

Oct 29, 2024

—

by

Source URL: https://hardware.slashdot.org/story/24/10/29/2022236/openai-builds-first-chip-with-broadcom-and-tsmc-scales-back-foundry-ambition?utm_source=rss1.0mainlinkanon&utm_medium=feed Source: Slashdot Title: OpenAI Builds First Chip With Broadcom and TSMC, Scales Back Foundry Ambition Feedly Summary: AI Summary and Description: Yes Summary: OpenAI is collaborating with Broadcom and TSMC to develop its first in-house AI chip aimed at enhancing AI inference capabilities, while reducing dependence on Nvidia GPUs. This strategic move…

The Register: xAI picked Ethernet over InfiniBand for its H100 Colossus training cluster

Oct 29, 2024

—

by

Source URL: https://www.theregister.com/2024/10/29/xai_colossus_networking/ Source: The Register Title: xAI picked Ethernet over InfiniBand for its H100 Colossus training cluster Feedly Summary: Work already underway to expand system to 200,000 Nvidia Hopper chips Unlike most AI training clusters, xAI’s Colossus with its 100,000 Nvidia Hopper GPUs doesn’t use InfiniBand. Instead, the massive system, which Nvidia bills as…

The Register: The troublesome economics of CPU-only AI

Oct 29, 2024

—

by

Source URL: https://www.theregister.com/2024/10/29/cpu_gen_ai_gpu/ Source: The Register Title: The troublesome economics of CPU-only AI Feedly Summary: At the end of the day, it all boils down to tokens per dollar Analysis Today, most GenAI models are trained and run on GPUs or some other specialized accelerator, but that doesn’t mean they have to be. In fact,…

The Register: ParTec expands supercomputer patent fight from Microsoft to Nvidia

—

by

Source URL: https://www.theregister.com/2024/10/28/partec_expands_supercomputer_patent_fight/ Source: The Register Title: ParTec expands supercomputer patent fight from Microsoft to Nvidia Feedly Summary: Wants injunction on GPUs that use what it alleges is its own IP German HPC vendor ParTec is taking legal action against Nvidia for alleged patent infringement, seeking an injunction to stop its GPUs being sold in…

Hacker News: Using reinforcement learning and $4.80 of GPU time to find the best HN post

—

by

Source URL: https://openpipe.ai/blog/hacker-news-rlhf-part-1 Source: Hacker News Title: Using reinforcement learning and $4.80 of GPU time to find the best HN post Feedly Summary: Comments AI Summary and Description: Yes Summary: The text discusses the development of a managed fine-tuning service for large language models (LLMs), highlighting the use of reinforcement learning from human feedback (RLHF)…

The Register: GenAI’s dirty secret: It’s set to create a mountainous increase in e-waste

—

by

Source URL: https://www.theregister.com/2024/10/28/genai_dirty_secret/ Source: The Register Title: GenAI’s dirty secret: It’s set to create a mountainous increase in e-waste Feedly Summary: New modelling shows tech could massively increase current volumes of electronic landfill by 2030 Computational boffins’ research claims GenAI is set to create nearly 1,000 times more e-waste than exists currently by 2030, unless…

Hacker News: GDDR7 Memory Supercharges AI Inference

—

by