Search NASA⌕ Search

SEARCH · Search NASA

Results for “MPI Endpoints”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

Enabling communication concurrency through flexible MPI endpoints

MPI defines a one-to-one relationship between MPI processes and ranks. This model captures many use cases effectively; however, it also limits communication concurrency and interoperability between MPI and programming models that utilize threads. This paper describes the MPI endpoints extension, which relaxes the longstanding one-to-one relationship between MPI processes and ranks. Using endpoints, an MPI implementation can map separate communication contexts to threads, allowing them to drive communication independently. Endpoints also enable threads to be addressable in MPI operations, enhancing interoperability between MPI and other programming models. These characteristics are illustrated through several examples and an empirical study that contrasts current multithreaded communication performance with the need for high degrees of communication concurrency to achieve peak communication performance.

Communication↗

Optimizing point‐to‐point communication between adaptive MPI endpoints in shared memory

Summary Adaptive MPI is an implementation of the MPI standard that supports the virtualization of ranks as user‐level threads, rather than OS processes. In this work, we optimize the communication performance of AMPI based on the locality of the endpoints communicating within a cluster of SMP nodes. We differentiate between point‐to‐point messages with both endpoints co‐located on the same execution unit and point‐to‐point messages with both endpoints residing in the same process but not on the same execution unit. We demonstrate how the messaging semantics of Charm++ enable and hinder AMPI's implementation in different ways, and we motivate extensions to Charm++ to address the limitations. Using the OSU micro‐benchmark suite, we show that our locality‐aware design offers lower latency, higher bandwidth, and reduced memory footprint for applications.

White, Sam↗

Optimizing the hypre solver for manycore and GPU architectures

The solution of large-scale combustion problems with codes such as Uintah on modern computer architectures requires the use of multithreading and GPUs to achieve performance. Uintah uses a low-Mach number approximation that requires iteratively solving a large system of linear equations. The Hypre iterative solver has solved such systems in a scalable way for Uintah, but the use of OpenMP with Hypre leads to at least slowdown due to OpenMP overheads. The proposed solution uses the MPI Endpoints within Hypre, where each team of threads acts as a different MPI rank. This approach minimizes OpenMP synchronization overhead and performs as fast or (up to 1.44) faster than Hypre's MPI-only version, and allows the rest of Uintah to be optimized using OpenMP. The profiling of the GPU version of Hypre shows the bottleneck to be the launch overhead of thousands of micro-kernels. The GPU performance was improved by fusing these micro-kernels and was further optimized by using Cuda-aware MPI, resulting in an overall speedup of 1.16—1.44 compared to the baseline GPU implementation. The above optimization strategies were published in the International Conference on Computational Science 2020 [1]. This work extends the previously published research by carrying out the second phase of communication-centered optimizations in Hypre to improve its scalability on large-scale supercomputers. Additionally, this includes an efficient non-blocking inter-thread communication scheme, communication-reducing patch assignment, and expression of logical communication parallelism to a new version of the MPICH library that utilizes the underlying network parallelism [2]. The above optimizations avoid communication bottlenecks previously observed during strong scaling and improve performance by up to 2 on 256 nodes of Intel Knight's Landing processor.

97 MATHEMATICS AND COMPUTING↗

OpenMP in VASP: Threading and SIMD

Abstract The Vienna Ab initio Simulation Package (VASP) is a widely used electronic structure code that originally exploits process‐level parallelism through the Message Passing Interface (MPI) for work distribution within and across nodes. Architectural changes of modern parallel processors urge programmers to address thread‐ and data‐level parallelism as well to benefit most from the available compute resources within a node. We describe for VASP how to approach for an MPI + OpenMP parallelization including data‐level parallelism through OpenMP SIMD constructs together with a generic high‐level vector coding scheme. We can demonstrate an improved scalability of VASP and more than 20% gain over the MPI‐only version as well as a 2× increased performance of collective operations using the multiple‐endpoint MPI feature. The high‐level vector coding scheme applied to VASP's general gradient approximation routine gives up 9× performance gain on AVX512 platforms with the Intel compiler.

Wende, Florian↗

Logically Parallel Communication for Fast MPI+Threads Applications

Supercomputing applications are increasingly adopting the MPI+threads programming model over the traditional "MPI everywhere" approach to better handle the disproportionate increase in the number of cores compared with other on-node resources. In practice, however, most applications observe a slower performance with MPI+threads primarily because of poor communication performance. Recent research efforts on MPI libraries address this bottleneck by mapping logically parallel communication, that is, operations that are not subject to MPI's ordering constraints to the underlying network parallelism. Domain scientists, however, typically do not expose such communication independence information because the existing MPI-3.1 standard's semantics can be limiting. Researchers had initially proposed user-visible endpoints to combat this issue, but such a solution requires intrusive changes to the standard (new APIs). The upcoming MPI-4.0 standard, on the other hand, allows applications to relax unneeded semantics and provides them with many opportunities to express logical communication parallelism. Here in this article, we show how MPI+threads applications can achieve high performance with logically parallel communication. Through application case studies, we compare the capabilities of the new MPI-4.0 standard with those of the existing one and user-visible endpoints (upper bound). Logical communication parallelism can boost the overall performance of an application by over 2x.

97 MATHEMATICS AND COMPUTING↗

Prediction of long-term (> 10 year) cardiovascular outcomes in heart transplant recipients: Value of stress technetium-99m tetrofosmin myocardial perfusion imaging

Background: Myocardial perfusion imaging (MPI) using single-photon emission computed tomography (SPECT) is useful in the evaluation of cardiac allograft vasculopathy (CAV) in heart transplant (HTx) recipients. The current study evaluated the long-term prognostic value of stress SPECT MPI for predicting all-cause mortality and cardiac events in HTx recipients. Methods: The study population consisted of 166 HTx recipients (mean age 54 ± 10 years, 84% male) who underwent exercise or dobutamine stress {sup 99m}Tc-tetrofosmin SPECT MPI for the assessment of CAV. An abnormal SPECT MPI was defined as the presence of a fixed or a reversible perfusion defect. Endpoints were all-cause mortality, cardiac mortality, and non-fatal myocardial infarction (MI). Results: MPI abnormalities were detected in 55 patients (33%), including fixed defects in 28 patients (17%), partially reversible in 17 patients (10%), and completely reversible defects in 10 patients (6%). During a median follow-up of 12.8 years (range 0-15, mean follow-up 9.5 years), 109 (66%) patients died (all-cause mortality), of which 67 (40%) were due to cardiac causes. A total of 5 (3%) patients experienced a non-fatal MI. HTx recipients with a normal stress {sup 99m}Tc-tetrofosmin SPECT MPI had a significantly better prognosis as compared with those with an abnormal study, up to 5 years after the initial test. The presence of a reversible perfusion defect was a significant predictor of all-cause mortality, cardiac mortality, and major cardiac events, during the entire follow-up period. Conclusions: Stress {sup 99m}Tc-tetrofosmin SPECT MPI provides valuable prognostic information for the prediction of long-term outcome in HTx recipients. Patients with a normal stress {sup 99m}Tc-tetrofosmin SPECT MPI have a significantly better prognosis as compared with those with an abnormal study, up to 5 years after initial testing.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

The prognostic value of heart rate response during vasodilator stress myocardial perfusion imaging in patients with end-stage renal disease undergoing renal transplantation

Background: In asymptomatic end-stage renal disease (ESRD) patients undergoing vasodilator stress myocardial perfusion imaging (MPI) prior to renal transplantation (RT), the impact of pre-transplant heart rate response (HRR) to vasodilator stress on post-RT outcomes is unknown. Methods: We analyzed a retrospective cohort of asymptomatic patients with ESRD who underwent a vasodilator stress SPECT-MPI and subsequently received RT. Blunted HRR was defined as HRR <28% for regadenoson stress and <20% for adenosine stress. The primary endpoint was major adverse cardiac events (MACE), defined as cardiac death or myocardial infarction. Clinical risk was assessed using the sum of risk factors set forth by the AHA/ACCF consensus statement on the assessment of RT candidates. Results: Among 352 subjects, 140 had an abnormal pre-transplant HRR. During a mean follow-up of 3.2 ± 2.0 years, 85 (24%) MACEs were observed. Blunted HRR was associated with increased MACE risk (hazard ratio 1.72; 95% confidence interval 1.12-2.63, P = 0.013), and remained significant after adjustment for gender, sum of AHA/ACCF risk factors, summed stress score, baseline heart rate, and β-blocker use. HRR was predictive of MACE in patients with normal MPI and irrespective of clinical risk. Blunted HRR was associated with a significant increase in post-operative (30-day) MACE risk (17.9% vs 8.5%; P = 0.009). Conclusion: In asymptomatic ESRD patients being evaluated for RT, a blunted pre-transplant HRR was predictive of post-RT MACE. HRR may be a valuable tool in the risk assessment of RT candidates.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Complete revascularization determined by myocardial perfusion imaging could improve the outcomes of patients with stable coronary artery disease, compared with incomplete revascularization and no revascularization

Objectives: To compare the outcomes among patients treated by complete coronary revascularization (CCR) or incomplete coronary revascularization (ICR) and no coronary revascularization (NCR) by myocardial perfusion imaging (MPI), as well as to evaluate the impact of severity of ischemia on patients with coronary artery disease (CAD) by different therapy strategies. Background: Using myocardial ischemia severity determined by MPI guiding treatment strategies for CAD patients still lacks strong clinical evidences. Methods: Consecutive patients (N = 286) underwent clinical stress-rest SPECT MPI and were retrospectively followed-up. For assessment of outcome of treatment, all patients were classified into three groups (CCR, ICR, and NCR), and further divided into two subgroups as mild ischemia (< 10% ischemic myocardium) and moderate-severe ischemia (≥ 10% ischemic myocardium). All-cause death was defined as the primary endpoint, and the composite of deaths, nonfatal myocardial infarction, and repeat revascularization (MACE) as the secondary endpoint. Results: Two-hundred eighty-six patients were followed-up for 46 ± 21 months. Thirty deaths and 65 MACEs were recorded. Patients treated by revascularization had significantly lower MACE (P < .001) but not mortality (P = .158) than patients treated by NCR. Outcomes of CCR related to mortality rate were greater than ICR and NCR (death: P = .019, MACE: P < .001). In patients with moderate-severe ischemia, CCR showed improved outcomes than ICR and NCR (death: P = .034; and MACE: P < .001). In patients with mild ischemia, the outcomes of CCR, ICR, and NCR had no significant difference (P > .05). Multivariate regression Cox analysis revealed that summed difference score [death: HR 1.09 (1.03, 1.15), P = .004] was an independent risk factor and CCR was an independent negative predictor [death: HR 0.31 (0.12, 0.81), P = .017; MACE: HR 0.30 (0.16, 0.57), P < .001]. Conclusions: Outcomes of patients treated by CCR were most likely more promising in comparison with treatment of ICR and NCR, especially when patients had over 10% ischemic myocardium.

62 RADIOLOGY AND NUCLEAR MEDICINE↗