Tobias Becker

The need for microsecond speed machine learning (ML) inference for particle physics experiments has emerged in recent years, in particular for the forthcoming upgrades to the experiments at the Large Hadron Collider at CERN. A community has grown around the need to develop the custom hardware platforms and tools required. The material presented in this report is drawn from the latest workshop held by the fast ML for science community and comprises of a collection of perspectives on the status of fast ML in different scientific domains, and the supporting technology.

BACKGROUND:Left ventricular assist device (LVAD) implantation may improve kidney function, but in patients awaiting heart transplantation, the long-term effects of LVAD implantation on renal function and subsequent clinical outcome are unclear. MATERIAL AND METHODS:We analyzed data in patients with LVAD implants (n=139) and without LVAD implants (n=1038) who were listed for a heart transplant at our institution between 2000 and 2019. The primary endpoint was an impairment in renal function (decrease of creatinine-based estimated glomerular filtration rate [eGFR] by ≥30%) up to a maximum of 2 years after listing. Secondary endpoints were chronic kidney disease stage 4 or 5, heart transplantation, survival during listing, and 1-year survival after transplantation. RESULTS:Values for eGFR increased after LVAD implantation (P=0.001) and were higher at the time of waitlisting in the LVAD group than in the non-LVAD group (P=0.002), but were similar between groups at the end of waitlisting (P=0.75). Two-year freedom from renal impairment was 50.6% and 66.7% in the LVAD and non-LVAD groups, respectively, with a multivariable-adjusted hazard ratio for the LVAD versus the non-LVAD group of 1.78 (95% confidence interval 1.19-2.68; P=0.005). Two-year freedom from chronic kidney disease stages 4-5 was similar between study groups (LVAD group: 83.5%; non-LVAD group: 80.1%; =0.50). The 2-year probability of transplantation was slightly lower in the LVAD group than in the non-LVAD group (50.0% and 55.8%, respectively, P=0.017). However, 2-year survival on the waiting list and 1-year survival after transplantation did not differ significantly between study groups (P-values >0.20). CONCLUSIONS:Our data indicate a transient improvement in creatinine-based eGFR values by LVAD implantation without influencing survival.

The next generation high-performance computing platforms will need to support exascale computing. A promising path in achieving exascale is to embrace heterogeneity and specialised computing in the form of reconfigurable accelerators. However, assessing the feasibility of heterogeneous exascale systems requires fast and accurate performance prediction. This paper proposes PERKS, a novel performance estimation frame-work for reconfigurable dataflow platforms (RDPs). PERKS uses machine and application parameters to build an analytical model for predicting the performance of multi-accelerator systems. Moreover, model calibration is automatic, making the model flexible and usable for different machine configurations and applications. Our experimental results demonstrate that PERKS can predict the performance of current workloads and RDPs with an accuracy above 95%. We also demonstrate how the modelling scales to exascale workloads and exascale platforms.

Building and successfully deploying applications on high-end heterogeneous resources such as GPUs, MICs or FPGAs, and typically with several library dependencies, is a complex task. Modern containerization provides a lightweight virtualization environment which can help solve some of these complex deployment and execution issues. By using containers, the software can be packaged with all the dependencies and tested in a single environment and deployable on heterogeneous architectures, easily. In this paper, we present our experiences with the container-based virtualized solutions that we developed for the use case applications in the EU H2020 project CloudLightning. We present specifics on their management and orchestration with specific software on the heterogeneous resources of our test-bed. The use cases include a Genomics application targeting FPGA-based DFEs, an Upscaling application for reservoir modeling, a ray tracing application targeting the MIC (Intel Xeon Phi co-processor), and a BLAS application with libraries, optimized for both CPU and GPU. An overview of the CloudLightning project and how the use case applications have been developed to be used as cloud services in the self-organizing, self-managing cloud technology, is also included.

This paper presents the cloud infrastructure of the AEGLE project, that targets to integrate cloud technologies together with heterogeneous reconfigurable computing in large scale healthcare systems for Big Bio-Data analytics. AEGLEs engineering concept brings together the hot big-data engines with emerging acceleration technologies, putting the basis for personalized and integrated health-care services, while also promoting related research activities. We introduce the design of AEGLE's accelerated infrastructure along with the corresponding software and hardware acceleration stacks to support various big data analytics workloads showing that through effective resource containerization AEGLE's cloud infrastructure is able to support high heterogeneity regarding to storage types, execution engines, utilized tools and execution platforms. Special care is given to the integration of high performance accelerators within the overall software stack of AEGLE's infrastructure, which enable efficient execution of analytics, up to 140× according to our preliminary evaluations, over pure software executions.