Michalis Papadimitriou
@mikepapadim
Followers
303
Following
7K
Media
36
Statuses
431
Research Fellow @OfficialUoM working on compilers, ML and open-source software. Part of the #TornadoVM team.
UK/Greece
Joined January 2015
Part 5: Project Leyden, Babylon, Panama + @TornadoVM At Devoxx Belgium 2025, Frank Delporte took the camera and microphone to talk with several speakers : Moritz Halbritter, @mikepapadim and Christos Kotselidis https://t.co/ByfnsaeEx6
1
5
8
introducing gpuup: you no longer have to put any effort into setting up CUDA toolkit + drivers on a node (single or multi gpu). just copy paste a short command (in replies)
24
57
738
Congratulations to @mikepapadim for presenting the #TornadoViz tool at @MPLR_Conf! It is an #opensource visualisation tool for the #TornadoVM Bytecodes that orchestrate the data transfers, execution, etc. of complex workloads. Paper: https://t.co/mHcI753oko
0
6
5
Final MPLR talk for 2025: Dynamic and Static Code Analysis for Java Programs on Heterogeneous Hardware
0
3
6
Next week at @MPLR_Conf & VMIL in Singapore! We are excited to share that our partners from @csmcr @CarvForth @RedHatJava are presenting 4 #research outcomes on #Java, #TornadoVM, #GC and #tools. #opensource #software
@EU_HaDEA
@UKRI_News All preprints on Zenodo 👇
1
6
7
Yesterday I had the chance to present in @Devoxx Belgium our work on building #GPULlama.java with @tornadovm . The first GPU-ready inference library in pure #Java that has no extra dependencies and it's fully integrated with @langchain4j . https://t.co/9nCb2T5H1i
2
7
20
Here's some MLX code that can help understand what a matmul with CUDA tensor cores is doing on the GPU. If you understand this you understand the core algorithm. Most of the rest of the complexity comes from efficiently loading the data and feeding the GPU in a pipeline and
6
27
309
Χθεσινή δημοσίευση στο έγκυρο Lancet: Σχεδόν 55,000 παιδιά κάτω των 6 ετών σε κατάσταση οξέος υποσιτισμού στη Λωρίδα της Γάζας, αντιμετωπίζουν αυξημένο κίνδυνο θανάτου. Για να κατανοήσουμε το μέγεθος της τραγωδίας: Η Γάζα είναι περίπου το 1/5 της Ελλάδας σε πληθυσμό (2 εκ).
77
121
392
The #TornadoVM deep dive session is live on YouTube! 👉 https://t.co/z9cFtq2qOP Next up: https://t.co/1IPlxDs1D4 at #Devoxx with live coding demo showing Java-native GPU acceleration! - 🗓 Tomorrow (Oct 9) | 🕔 17:40–18:30 | 📍 Room 6 Don’t miss it! #opensource #Java #AI #GPU
0
8
16
0
9
19
Yesterday at the #TornadoVM deep dive, we presented #TornadoInsight - an #opensource IntelliJ plugin that helps devs learn the TornadoVM API via static analysis & dynamic inspection. Extended in @P2CODE_eu, and coming to #MPLR25 next week! #Devoxx
https://t.co/d6DhXBmgKN
1
9
16
Behind the cameras! @CKotselidis and @mikepapadim discussing with @FrankDelporte about our technology at @Devoxx and how it can be used for fast #AI in #Java! The video is coming soon…
0
4
10
In today's @Devoxx we presented how we use @graalvm to generate efficient GPU code with @tornadovm and https://t.co/pEHpA4oprX
#java #ai #llms
0
6
12
My colleague @mikepapadim gives live #code demo with #TornadoVM at @Devoxx Belgium! #opensource #software #Java #AI
0
5
11
Good morning from #Devoxx Belgium and the #TornadoVM team! Our #TornadoVM deep dive talk is today at 13.30 in Room 4. #opensource #Java #AI #GPU
1
7
14
#Java + #GPUs + #AI = #TornadoVM We’re speaking at #Devoxx 2025! I will be there along with @mikepapadim @MXekalaki @CKotselidis for a deep dive on @tornadovm! 📅 Oct 6 | 13:30–16:30 | Rm 4 @hipeac
@EU_CloudEdgeIoT
@AERO_Project_EU
@P2CODE_eu
@tangoproject_eu
#opensource
0
7
11
Επειδή θα αποκρύψουν τι είπε η #Κοβεσι κ δεν θα αναφέρουν τίποτα για κλέφτες, σκάνδαλα και σταύλους του Αυγεία, πάρτε το βίντεο. #ΟΠΕΚΕΠΕ
52
1K
4K
Also, in another session we will showcase how @graalvm allows @tornadovm to produce extremely effiecient GPU kernels for fast LLM inference for multiple hardware vendors https://t.co/5qcz66VSGs
m.devoxx.com
By Michalis Papadimitriou. This session introduces GPULlama3.java, an open-source framework enabling efficient, GPU-accelerated inference for Llama and similar models directly in Java using Tornado...
TornadoVM Deep Dive 🚀 This session will explore how @TornadoVM extends GraalVM to unlock GPU acceleration, covering parallelism, integration with existing tooling, and a live demo optimizing an LLM pipeline. https://t.co/Wdo8X5GPgn
0
4
8
First integration of https://t.co/pEHpA4oXhv with @langchain4j. Now, you can compile and run pure-Java inference on GPUs with @tornadovm for Llama3, Mistral, Qwen2.5/3 and Phi-3 https://t.co/sIwhJTjnjY
github.com
This pull request integrates GPU-accelerated large language model (LLM) inference into LangChain4j by incorporating GPULlama3.java](https://github.com/beehive-lab/GPULlama3.java), a Java-native imp...
0
6
13