📢 A reminder that I am looking for a capable and ambitious researcher for my #EPSRC#NIA#project on Carbon-Aware Scalable Processing from early next year. Applications for the #PDRA/#RA position are due on 21-Nov-2024 at 23:45h UK time: https://t.co/RU8eWwv8w8 📢 #hiring
Great talk yesterday by @ps0mas on a full room! Come and find him today at the poster session in the morning to learn more about Elastic Translations! #MICRO2024@cslab_ntua
Universities should be spaces of free expression, not silenced by force. This was the European Researcher's Night, a public university event meant to engage the community. This is madness.
What was meant to be an open outreach event at @ntua was shut down by force, with armed police guarding the doors, just because of a peaceful protest for Palestine by the researchers’ union at their own night. This is a disgrace.
@ERC_Research @MSCActions
Greetings from #ICPP2024 after a successful presentation of our work (cc @miquel_pericas) on co-designing convolutions and long vector RISC-V CPUs for CNN model serving (https://t.co/siEfZmJHsf)
Reproducibility matters! Consider joining the effort to evaluate artifacts, to ensure robust and reliable results. Your expertise can make a difference.
Me, @portokalidis and Christian Pinto are chairing the artifact evaluation committee for @EurosysConf#EuroSys2025. We welcome self-nominations for committee members! 🙋♀️🙋♂️
The fact that papers, that are clearly plagiarizing AI, made it through the peer review process says a lot about the quality and value of the current peer review process.
#PhDVoice@PostdocVoice
The boom in demand for #AI has led to a surge in focus on #datacenter capacity, with extreme stress on electricity grids, generation capacity, and the environment
https://t.co/4mUhDp1ILt
#HPC via @dylan522p@SemiAnalysis_
GPU Flops-per-byte went crazy over the years. It's even more insane if you plot this for interconnect (PCIE, NVLink, XeLink etc) bandwidth
We were close to 1:1 when the first floating point shaders were introduced in 2002 on Radeon 9700 Pro.
Re-balancing flops/byte to more saner levels should yield compelling performance/$/watt improvement for AI workloads (probably most GPU workloads)
Celebrating International Day of Women and Girls in Science, Feb 11! This year, delve into the impact of women in science on SDGs. IDWIGS strengthens global bonds, shares innovative solutions, and pioneers a science workshop for visually impaired girls. https://t.co/23WgJ21VSE
(1/n) Excited to announce the "Cerebras PyTorch Sparsity Library," which democratizes access to #sparse training for all #ML researchers and developers.
Read more here: https://t.co/qffivdvcbg
Business as usual will not be sufficient: @KathyYelick will give the opening keynote on “Beyond #Exascale Computing” at #ISC24
https://t.co/gBoEIp4or2
#HPC#AI
Remember our work, PARALiA, for multi-GPU BLAS? Here is @PetrosAnastasi6 from @cslab_ntua presenting results for communication-optimized, multi-GPU, high-performance, energy-efficient GEMM with PARALiA, at @hipeac Session 2 on Runtime Optimization!
Do you run GEMM on multiple GPUs? Want to speed it up by ~70% and save some energy? Have your cake and eat it too with PARALiA, a new heterogeneous BLAS runtime by @PetrosAnastasi6.
Read the paper: https://t.co/ZyarG5Jcv0
Grab the code: https://t.co/DAy9RzdW8Y @cslab_ntua
If you've been working diligently to meet the AsHES @IPDPS deadline, here is some good news for you: the deadline has been extended to January 26 AoE! We are looking forward to your contributions! #hpc#heterogeneous#EmergingTech#IPDPS#callforpapers