- Популярные видео
- Авто
- Видео-блоги
- ДТП, аварии
- Для маленьких
- Еда, напитки
- Животные
- Закон и право
- Знаменитости
- Игры
- Искусство
- Комедии
- Красота, мода
- Кулинария, рецепты
- Люди
- Мото
- Музыка
- Мультфильмы
- Наука, технологии
- Новости
- Образование
- Политика
- Праздники
- Приколы
- Природа
- Происшествия
- Путешествия
- Развлечения
- Ржач
- Семья
- Сериалы
- Спорт
- Стиль жизни
- ТВ передачи
- Танцы
- Технологии
- Товары
- Ужасы
- Фильмы
- Шоу-бизнес
- Юмор
"AI models can take weeks or months to run. What happens when something goes wrong during training?"
In this episode of Utilizing Tech, Davide Villa of Xinnor joins hosts Stephen Foskett from Tech Field Day and Ace Stryker from Solidigm to explore how software RAID can keep GPUs fully utilized while ensuring data reliability.
Davide explains how xiRAID leverages server resources to handle AI’s demanding workloads without additional hardware, minimizing risks like write amplification and performance drops during failures.
Видео "AI models can take weeks or months to run. What happens when something goes wrong during training?" канала Xinnor
Davide explains how xiRAID leverages server resources to handle AI’s demanding workloads without additional hardware, minimizing risks like write amplification and performance drops during failures.
Видео "AI models can take weeks or months to run. What happens when something goes wrong during training?" канала Xinnor
Комментарии отсутствуют
Информация о видео
6 февраля 2025 г. 14:41:13
00:00:35
Другие видео канала
Deploying xiRAID Classic with IBM Storage Scale: Full Demo & Performance Test
How to build a top-IO500 storage cluster using commodity hardware?
How to build an AI / HPC storage stack that balances speed, resiliency & cost
xiSTORE | Software-Defined Storage Solution for HPC & AI Workloads
All-NVMe Storage Cluster with IBM Storage Scale Accelerated by xiRAID on Supermicro Servers
Introducing xiSTORE: The Future of SDS for HPC & AI | Webinar | November 29, 2023
xiRAID + WD Ultrastar Transporter
xiNAS Demo: NAS Troubleshooting and Performance Tuning with AI Agents via MCP
“61.44 TB rebuild in 5 hours”
High Availability Support & Unprecedented Performance: Keep Data Safe and Sound with xiRAID v4.1
xiRAID + Lustre outperforming CephFS
Scaling That Just Works
How to get maximum capacity, performance, reliability from existing NVMes?
Performance per Hardware Efficiency
High-Performance Storage Solution for PostgreSQL Database in Virtual Environment
Measuring raw NVMe headroom (tens of millions IOPS)
Revolutionizing PCIe 5 Storage with Unmatched RAID Performance | xiRAID by Xinnor
Leverage Your HPC Expertise for AI
Be bold — but be realistic
xiRAID：极速高性能 NVMe RAID解决方案 —— 高 IOPS、低延迟、容错与软件定义存储之选，适用于 HPC、AI、大数据、云计算及数据中心
