COMPUTEX 2024 臺北國際電腦展於昨日(6/4)正式開幕,AMD超微執行長蘇姿丰(Lisa Su)現身演講,說明AI和高效能運算(HPC)運用在AMD的轉型升級中扮演關鍵角色,並進一步說明:AI在各產業領域的變革創新潛力、AMD Zen 5的技術亮點及預計上市規劃、第三代Ryzen AI處理器領導AI PC的進程、數據中心與雲端解決方案以及未來生成式AI的發展。
COMPUTEX 2024 kicked off yesterday (June 4), and AMD CEO Dr. Lisa Su took the stage to highlight the pivotal role AI and high-performance computing (HPC) play in AMD's transformative journey.
She discussed AI’s potential to revolutionize various industries, the key features and launch plans for AMD’s Zen 5, the progress of the third-generation Ryzen AI processors in leading AI PCs, data center and cloud solutions, and the future of generative AI.
(翻攝自Youtube @AMD:AMD at Computex 2024: AMD AI and High-Performance Computing with Dr. Lisa Su)
演講筆記 Keynote
AI整合與推動 AI Integration and Advancement
蘇姿丰強調AI在各產業的變革潛力,致力於創新與合作,推動尖端解決方案的開發。
AMD透過先進的GPU、CPU和NPU及開放軟體,使AI潛力概念轉變為實際產品,涵蓋AI PC、雲端等多元領域。
Dr. Lisa Su emphasized AI’s transformative potential across industries, stressing AMD’s commitment to innovation and collaboration in developing cutting-edge solutions.
AMD turns AI concepts into real products through advanced GPUs, CPUs, NPUs, and open software, covering AI PCs, cloud, and more.
Zen 5技術亮點 Zen 5 Technological Highlights
Zen 5架構廣泛應用於超級計算機、數據中心和PC中。
新技術亮點,包括雙平行管道前端,改善分支預測準確性和降低延遲,並設計了更寬的CPU引擎和指令窗口來提升效能。
Zen 5產品預計將於7月上市,將領先性能帶到主流價格點。
The Zen 5 architecture is set to be used extensively in supercomputers, data centers, and PCs.
Key advancements include dual parallel pipeline frontends, improved branch prediction accuracy, and reduced latency. The wider CPU engine and instruction window enhance performance.
Zen 5 products are expected to hit the market in July, bringing top-tier performance to mainstream price points.
第三代Ryzen AI處理器 Third-Generation Ryzen AI Processors
AMD從去年1月推出初代Ryzen AI以來,持續領導著AI PC的轉型。
第三代Ryzen AI處理器在運行生成式AI時,可將效能提高至兩倍,並提供50 TOPS的INT8 AI性能。
AMD與微軟合作推出Copilot+ PC,搭載第三代Ryzen AI處理器,可帶來最佳NPU驅動體驗。
Since launching the first-generation Ryzen AI in January last year, AMD has continued to lead the AI PC revolution.
The third-generation Ryzen AI processors double performance in running generative AI and offer 50 TOPS of INT8 AI performance.
AMD’s collaboration with Microsoft on the Copilot+ PC, featuring the third-generation Ryzen AI processors, delivers the best NPU-driven experience.
與合作夥伴的協作 Collaboration with Partners
微軟副總裁Pavan Davuluri談及與AMD合作推出的「Windows Copilot Runtime」,一款讓開發者能自由訪問Windows新AI功能的API庫。
AMD與Lenovo聯想合作,推出搭載第三代Ryzen AI處理器的Copilot+ PC,並引入專為創作者設計的Creator Zone軟體。
Microsoft VP Pavan Davuluri spoke about the partnership with AMD to launch the “Windows Copilot Runtime,” an API library that gives developers access to new Windows AI features.
AMD and Lenovo have teamed up to release the Copilot+ PC with third-generation Ryzen AI processors, introducing the Creator Zone software designed for creators.
AI在各領域的應用 AI Applications Across Industries
AMD技術廣泛應用於數據中心、5G網路、醫療保健、製造業、汽車、PC和遊戲領域,每天影響數十億人的生活。
Subaru使用Versal技術在其EyeSight ADAS系統中,分析前置鏡頭捕捉的每幀畫面,並警示潛在的安全隱患。
Hitachi Energy利用AMD自適應計算產品,檢測高壓直流系統中的潛在電氣問題。
AMD’s technology is widely applied in data centers, 5G networks, healthcare, manufacturing, automotive, PCs, and gaming, impacting billions of lives daily.
Subaru utilizes Versal technology in its EyeSight ADAS system to analyze each frame from front-facing cameras, alerting potential safety hazards.
Hitachi Energy uses AMD adaptive computing products to detect potential electrical issues in high-voltage DC systems.
數據中心與雲端解決方案 Data Center and Cloud Solutions
EPYC處理器是雲計算的首選,支持所有主要雲供應商的內部工作負載。
下一代數據中心CPU和GPU產品,Turin和MI300X,可提供領先全球的推理性能、記憶體大小和計算能力。
EPYC processors are the go-to for cloud computing, supporting internal workloads of all major cloud providers.
The next-generation data center CPUs and GPUs, Turin and MI300X, offer world-leading inference performance, memory size, and computing power.
Generative AI發展 Generative AI Development
Stability AI與AMD合作,優化SD3模型,實現了革命性的視覺化體驗提升,並利用MI300X的192GB HBM記憶體進行更快、更有效率的模型訓練。
AMD計劃每年推出新的系列產品,2025年將推出MI350系列,進一步提升性能和記憶體容量。
Stability AI collaborates with AMD to optimize the SD3 model, achieving a revolutionary enhancement in visual experiences and using the MI300X’s 192GB HBM memory for faster, more efficient model training.
AMD plans to release new product lines annually, with the MI350 series slated for 2025, promising further improvements in performance and memory capacity.
AMD的未來展望 AMD's Future Outlook
AMD將致力於開放性AI網路基礎系統開發,提高螢幕更新率及降低延遲。
AMD成為唯一能提供完整的CPU、GPU和網路解決方案的企業,滿足現代數據中心的所有需求。
AMD is committed to developing open AI network infrastructure, improving screen refresh rates, and reducing latency.
As the only company providing a complete suite of CPU, GPU, and network solutions, AMD is poised to meet all modern data center needs.
Comentários