Updates
Part Two - Porting AI codes from CUDA to SYCL and oneAPI, one llama at a time
Prelude In our first part we looked at the conversion from CUDA to SYCL via the whole project migration tool, SYCLomatic. Now we are going to take this portable code, and run it across an NVIDIA and Intel GPU. Building on the NVIDIA system Now we are going to build...
Cuda
Sycl
Oneapi
Porting
Part One - Porting AI codes from CUDA to SYCL and oneAPI, one llama at a time
Introduction The rapid advancement of LLMs can be attributed to their ability to effectively tackle complex problems, such as those encountered in chatbots, virtual assistants, content generation, and language translation. Their performance, which matches human capabilities, places LLMs at the forefront of AI models. The classical general purpose graph frameworks...
Cuda
Sycl
Oneapi
Porting
Accelom and SYCLOPS
Due to significant improvements in genomic sequencing technology, it is now possible to sequence a full human genome for less than $1000 and receive genetic data on several molecular levels of the cell (genomic, transcriptome, protein expression, copy-number variations, and so on). However, in order to turn this genomic Big...
Accelom
Syclops
DAC 2024 – Showcasing the future of RISC-V through EDA
This week, the 2024 edition of RISC-V Summit Europe took place in lovely Munich, Germany. Those of us who attended last year’s edition in Barcelona might not have expected the same weather but Munich was up for the challenge and served us a sunny, hot week, only interrupted by a...
Codasip
Risc-v
Conference
Summit
Sunny skies and electric energy: RISC-V Summit Europe 2024 shines in Munich
As I sat on the plane in Boston it’s fair to say that I was curious about what DAC 2024 would bring. The previous year was much better than I expected but a cold June in San Francisco wasn’t exactly what I was dreaming about. Afterall, while I was heading...
Codasip
Risc-v
Energy
AdaptivePerf and SYCLOPS
AdaptivePerf is an open-source architecture-agnostic code profiler for Linux, based on the custom-patched “perf”. Using just one single command adaptiveperf "<command to be profiled>", the tool: samples both on-CPU and off-CPU activity, traces every spawned thread and process, minimises the risk of broken profiled stacks by detecting inappropriate kernel and...
Press-release
Linux
Performance
Github
Seven EU Projects Collaborate Around Shared Mission to Extract Meaningful Insights from Extreme Data
As part of the Horizon Europe funding programme, seven pioneering projects have received funding to address the challenges posed by extreme data generated by IoT, industrial, business, administration, environmental, scientific and societal sources. These projects are pursuing innovative approaches that integrate cutting-edge technologies, including Artificial Intelligence (AI), Internet of Things...
Press-release
Construction-kit
Oneapi
Codeplay
The oneAPI Construction Kit and SYCLOPS
With software evolving quickly, it has become increasingly common for hardware vendors to create specialist AI processors that run their software more efficiently than would be possible with out-of-the-box hardware. While these custom processors can offer the advantage of performance, they come with challenges for developers. Primary among these is...
Oneapi
Construction-kit
Custom Compute for Edge AI: Accelerating innovation with Lund University and Codasip University Program
In recent years, the rapid advancement and adoption of Artificial Intelligence (AI) on the edge has brought about a surge in development. As AI models like ChatGPT become more prevalent and accurate, the computational requirements for inference also escalate. This necessitates architectural innovations aimed at reducing both power consumption and...
Codasip
Risc-v
Edge
Hpc
Ai
A custom RISC-V vector instruction to accelerate structured-sparse matrix multiplications
A novel AI-acceleration paper presents a method to optimize sparse matrix multiplication for machine learning models, particularly focusing on structured sparsity. Structured sparsity involves a predefined pattern of zero values in the matrix, unlike unstructured sparsity where zeros can occur anywhere. The research was conducted by Democritus University of Thrace...
Codasip
Risc-v
Ai
Matrix
Effectively hiding sensitive data with RISC-V Zk and custom instructions
Cryptographic hash functions play a critical role in computer security providing a one-way transformation of sensitive data. Many information-security applications benefit from using hash functions, specifically digital signatures, message authentication codes, and other forms of authentication. The calculation of hash functions such as SHA512, SHA256, MD5 etc is a potential...
Codasip
Risc-v
Cryptographic
Introducing the SYCLOPS Consortium
Announced in May this year, the SYCLOPS Project brings together 8 leading organizations from across Europe, collaborating to democratize AI acceleration and break the existing monopoly on the acceleration market. Our combined vision is using open standards to enable a healthy, competition-driven ecosystem with widespread adoption of both RISC-V®, an...
Consortium
Announcement
Codasip 700 RISC-V processor family: Bringing the world of Custom Compute to everyone
Today, technology innovators must have new ways to create differentiated products. How are they supposed to meet the demand for more computational performance when semiconductor scaling laws are showing their limits? There is only one way: having a compute that is custom for specific needs. And what do we need...
Codasip
Risc-v
How the SYCLOPS project democratizes AI acceleration
Codasip Labs is all about innovation, and specifically the commercialization of that innovation. Naturally, with the rise of Artificial Intelligence (AI) and Machine Learning (ML), these areas have become a key focus for us. At the beginning of 2023, we joined the New Horizon Europe Project SYCLOPS (Scaling extreme analYtics...
Codasip
Risc-v
Ai
Ml
Re-targetable LLVM C/C++ compiler for RISC-V
RISC-V is a modular instruction set architecture (ISA) with great customization capabilities that enable innovation and differentiation without fragmentation. On top of the baseline modules from ratified/standard ISA extensions, such as integer instructions or floating-point instructions, designers can add custom instructions: pure design freedom! And the reasons for adding instructions...
Codasip
Risc-v
Llvm
No one-size-fits-all approach to RISC-V processor optimization
As the demand for high-performance processors continues to grow and semiconductor scaling laws continue to show their limits, the need for processor optimization is inevitable. As I explained in a previous blog, RISC-V is designed to enable this. However, there is no one-size-fits-all approach to processor optimization. As each workload...
Codasip
Risc-v
Optimization
Launch of the New Horizon Europe Project SYCLOPS
8 leading European organisations join forces to bring together RISC-V and SYCL standards to demonstrate ground-breaking advances in scalability of extreme data analytics via fully-open AI acceleration The wide-spread adoption of AI has resulted in a market for novel hardware accelerators that can efficiently process AI workloads. Unfortunately, all popular...
Launch
Update
RISC-V customization, HW/SW co-optimization, and custom compute
Do we still need to introduce and define RISC-V? You know, the open-source instruction set architecture (ISA) that is gaining popularity thanks to its flexibility, scalability, and modularity. Okay, we just did, just to be sure we are all on the same page. One of the key benefits and the...
Codasip
Risc-v
Customization