Online GPUAnalysis using Adaptive DMA Controlled by Softcore for 2D Detectors
| Autor: | Wassim Mansour, R. Ponsard, V. Fristot, Nicolas Janvier, Dominique Houzet |
|---|---|
| Přispěvatelé: | European Synchrotron Radiation Facility (ESRF), Centre Interuniversitaire de Micro-Electronique (CIME), Institut National Polytechnique de Grenoble (INPG) |
| Rok vydání: | 2020 |
| Předmět: |
Memory buffer register
MicroBlaze Computer science Controller (computing) CMA 02 engineering and technology DMA 01 natural sciences PCIe [INFO.INFO-TS]Computer Science [cs]/Signal and Image Processing 0103 physical sciences Synchronization (computer science) 010306 general physics Field-programmable gate array Throughput (business) FPGA PCI Express business.industry Data stream mining Microblaze GPUDirect 021001 nanoscience & nanotechnology [INFO.INFO-PF]Computer Science [cs]/Performance [cs.PF] Embedded system 0210 nano-technology business |
| Zdroj: | DSD Euromicro DSD 2020 Euromicro DSD 2020, Aug 2020, Portorož (Virtual), Slovenia. ⟨10.1109/DSD51259.2020.00075⟩ |
| DOI: | 10.1109/dsd51259.2020.00075 |
| Popis: | International audience; New generation X-ray detectors enables cutting-edge experiments that can produce very high throughput data streams that are challenging to manage and store. This paper presents an evaluation of a configurable data placement mechanism from an FPGA device collecting detector raw data to a burst-cache memory and concurrently to a GPU accelerator, bypassing hardware and software extraneous copies and bottlenecks via PCI-Express. It includes a DMA controller dynamically configured in real-time by a Microblaze soft-processor. A low-latency synchronization mechanism using GPUDirect technology is presented as well. Multi-GB, DMA-able memory buffer allocation, leveraging Linux contiguous memory allocator is investigated. As illustrative workloads, real-time raw-data correction as foreseen in Serial Synchrotron X-ray experiments were processed. Obtained results showed that if one could reach a data throughput of 12.7GB/s to CPU memory when using PCIe gen3 x16, a 12-cores OpenMP CPU application processes the raw data only up to 2.7GB/s and is outperformed by a GPU accelerator (NVIDIA RTX 6000). |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|