
www.csd.cs.cmu.edu
Preview meta tags from the www.csd.cs.cmu.edu website.
Linked Hostnames
17- 92 links towww.csd.cs.cmu.edu
- 8 links towww.cmu.edu
- 8 links towww.cs.cmu.edu
- 3 links tocalendar.google.com
- 2 links tocoursecatalog.web.cmu.edu
- 1 link tobsky.app
- 1 link tocmu.wd5.myworkdayjobs.com
- 1 link toenr-apps.as.cmu.edu
General Meta Tags
6- titleHome | Carnegie Mellon University Computer Science Department
- charsetutf-8
- GeneratorDrupal 10 (https://www.drupal.org)
- MobileOptimizedwidth
- HandheldFriendlytrue
Link Tags
29- canonicalhttps://www.csd.cs.cmu.edu/
- icon/sites/default/files/favicon_0.ico_0_0.gif
- preconnecthttps://fonts.googleapis.com
- preconnecthttps://fonts.gstatic.com
- shortlinkhttps://www.csd.cs.cmu.edu/
Links
125- http://coursecatalog.web.cmu.edu
- http://coursecatalog.web.cmu.edu/schools-colleges/schoolofcomputerscience/undergraduatecomputerscience/#bscurriculumtextcontainer
- https://bsky.app/profile/csdatcmu.bsky.social
- https://calendar.google.com/calendar/render?action=TEMPLATE&dates=20250613T150000/20250613T153000&ctz=America/New_York&text=CRA+Research+Policy+Virtual+Town+Hall+-+Faculty+-+registration+required&details=%0A++++++++++++The+Trump+Administration%27s+recently+unveiled+Fiscal+Year+2026+budget+request+poses+a+significant+threat+if+passed+into+law%2C+especially+for+those+who+rely+on+National+Science+Foundation+funding.%0A++++++
- https://calendar.google.com/calendar/render?action=TEMPLATE&dates=20250617T140000/20250617T153000&ctz=America/New_York&text=Doctoral+Thesis+Proposal+-+Daiyaan+Arfeen&details=%0A++++++++++++%3Cp%3ELLM+training+requires+massive+amounts+of+compute+due+to+large+model+and+dataset+sizes%2C+so+it+is+not+unusual+to+train+LLMs+on+tens+or+hundreds+of+thousands+of+GPUs+to+complete+training+in+a+reasonable+amount+of+time+%28days+or+weeks%29.+However%2C+GPU+failures+%28which+are+common+at+these+scales%29+and+data-dependencies+%28introduced+by+the+training+algorithms%29+can+lead+to+severe+GPU+underutilization.%26nbsp%3B%26nbsp%3B%3C%2Fp%3E%0A++++++&location=+Designing+Scalable+DNN+Training+Systems+to+Overcome+Algorithmic+Constraints