Addds the GenVectorX update.

_collections/_updates/2025-03-05-syclops-genvector-x.md

@@ -0,0 +1,41 @@
+---
+title: "SYCLOPS GenVectorX"
+date: 2025-03-05
+layout: update
+tags:
+  - syclops
+  - GenVectorX
+---
+
+
+CERN hosts the world's largest particle accelerator, the LHC, where beams of particles are accelerated and made to
+collide to study the fundamental structures of our universe. The LHC produces petabytes of physics data regarding
+particle collisions that need to be analysed efficiently. These collisions are expressed as operations on particles,
+represented as 4-dimensional time-space vectors, also known as Lorentz Vectors. Analysing these massive amounts of data
+naturally requires highly efficient and complex software solutions. ROOT is an open-source platform-independent C++ data
+analysis framework designed and developed for this purpose. The [ROOT project](https://root.cern) provides the core
+ingredients for high energy physics (HEP) analysis tasks, including optimised data storage, a user-friendly and
+interactive interface, math libraries, and visualisation capabilities. ROOT’s high-level data analysis
+interface, [RDataFrame](https://root.cern/doc/master/classROOT_1_1RDataFrame.html), currently contains
+support for implicit parallelism in multi-threaded and multi-node distributed environments, but GPU parallelism has not
+been attempted before this work. Given the increasing heterogeneity in computing facilities, it becomes crucial to
+efficiently support GPGPUs to take advantage of the available resources. SYCL allows for a single-source implementation,
+which enables support for different architectures.
+
+![Image 1](/static/images/updates/2025-03-05-syclops-genvector-x/image.webp)
+
+Within ROOT, the GenVector is a large package intended to provide classes and functionalities to represent physical
+vectors and transformations (such as rotations and Lorentz transformations) in 2, 3, and 4 dimensions, according to the
+needs of High Energy Physics researchers. [GenVectorX](https://arxiv.org/abs/2312.02756) extends GenVector, providing a
+SYCL implementation of the Lorentz Vector classes that facilitate computations with physical vectors. We leverage SYCL 
+as a performance portable framework to migrate and modernize the fundamental GenVector package. Typically, in a HEP 
+analysis the data are first read from disk and decompressed; then irrelevant data are filtered out, new quantities 
+are defined by processing the remaining particles, possibly more data are filtered out, and last the results are 
+aggregated, e.g. by computing an histogram. The GenVectorX package addresses the define action, even if the histogram 
+computation has been ported. An extensive test campaign shows that SYCL can achieve competitive performance with 
+respect to native CUDA on NVIDIA GPUs with near-one code similarity, meaning that the codebase is more sustainable 
+in terms of development efforts and less error-prone as a single source needs to be maintained. Moreover, we acquire 
+support for accelerators beyond NVIDIA GPUs - before SYCL, RDataFrame only supported multithreaded execution across 
+bulks. These features haven’t been put in production yet, but if you’re curious you can take a look at the 
+related [development branch!](https://github.com/mdessole/root/tree/genvectorx_gpu_histogram_bulk-2.0)
+