Parallel architectures are nowadays not only confined to the domain of high performance computing, they are also increasingly used in embedded time-critical systems. The ARGO H2020 project1provides a programming paradigm and associated tool flow to exploit the full potential of architectures in terms of development productivity, time-to-market, exploitation of the platform computing power and guaranteed real-time performance. In this paper we give an overview of the objectives of ARGO and explore the challenges introduced by our approach.

Most schedulability analysis techniques for multi-core architectures assume a single Worst-Case Execution Time (WCET) per task, which is valid in all execution conditions. This assumption is too pessimistic for parallel applications running on multi-core architectures with local instruction or data caches, for which the WCET of a task depends on the cache contents at the beginning of its execution, itself depending on the task that was executed before the task under study. In this paper, we propose two scheduling techniques for multi-core architectures equipped with local instruction and data caches. The two techniques schedule a parallel application modeled as a task graph, and generate a static partitioned non-preemptive schedule. We propose an optimal method, using an Integer Linear Programming (ILP) formulation, as well as a heuristic method based on list scheduling. Experimental results show that by taking into account the effect of private caches on tasks' WCETs, the length of generated schedules is significantly reduced as compared to schedules generated by cache-unaware scheduling methods. The observed schedule length reduction on streaming applications is 11% on average for the optimal method and 9% on average for the heuristic method.

A task can be preempted by several jobs of higherpriority tasks during its response time. Assuming the worst-casememory demand for each of these jobs leads to pessimistic worst-case response time (WCRT) estimations. Indeed, there is a bigchance that a large portion of the instructions and data associatedwith the preempting task Tj are still available in the cache when Tj releases its next jobs. Accounting for this observation allowsthe pessimism of WCRT analysis to be significantly reduced, which is not considered by existing work. The four main contributions of this paper are: 1) The conceptof persistent cache blocks is introduced in the context of WCRTanalysis, which allows re-use of cache blocks to be captured,2) A cache-persistence-aware WCRT analysis for fixed-prioritypreemptive systems exploiting the PCBs to reduce the WCRTbound, 3) A multi-set extension of the analysis that furtherimproves the WCRT bound and 4) An evaluation showing thatour cache-persistence-aware WCRT analysis results in up to 10%higher schedulability than state-of-the-art approaches.

Timing validation is a critical step in the design of real-time systems, that requires the estimation of Worst-Case Execution Times (WCET) for tasks. A number of different methods have been proposed, such as Static Deterministic Timing Analysis (SDTA). The advent of Probabilistic Timing Analysis, both Measurement-Based (MBPTA) and Static Probabilistic Timing Analyses (SPTA), offers different design points between the tightness of WCET estimates, hardware that can be analyzed and the information needed from the user to carry out the analysis. The lack of comparison among those techniques makes complex the selection of the most appropriate one for a given system. This paper makes a first attempt towards comparing comprehensively SDTA, SPTA and MBPTA, qualitatively and quantitatively, under different cache configurations implementing LRU and random replacement. We identify strengths and limitations of each technique depending on the characteristics of the program under analysis and the hardware platform, thus providing users with guidance on which approach to choose depending on their target application and hardware platform.

In this paper, we present EETCO: an estimation and exploration tool that provides qualitative assessment of data center design decisions on Total-Cost-of-Ownership (TCO) and environmental impact. It can capture the implications of many parameters including server performance, power, cost, and Mean-Time-To-Failure (MTTF). The tool includes a model for spare estimation needed to account for server failures and performance variability. The paper describes the tool model and its implementation, and presents experiments that explore tradeoffs offered by different server configurations, performance variability, MTTF, 2D vs 3D-stacked processors, and ambient temperature. These experiments reveal, for the data center configurations used in this study, several opportunities for profit and optimization in the datacenter ecosystem: (i) servers with different computing performance and power consumption merit exploration to minimize TCO and the environmental impact, (ii) performance variability is desirable if it comes with a drastic cost reduction, (iii) shorter processor MTTF is beneficial if it comes with a moderate processor cost reduction, (iv) increasing by few degrees the ambient datacenter temperature reduces the environmental impact with a minor increase in the TCO and (v) a higher cost for a 3D-stacked processor with shorter MTTF and higher power consumption can be preferred, over a conventional 2D processor, if it offers a moderate performance increase.

We propose a power-efficient many-core server-on-chip system with 3D-stacked Wide I/O DRAM targeting cloud workloads in datacenters. The integration of 3D-stacked Wide I/O DRAM on top of a logic die increases available memory bandwidth by using dense and fast Through-Silicon Vias (TSVs) instead of off-chip IOs, enabling faster data transfers at much lower energy per bit. We demonstrate a methodology that includes full-system microarchitectural modeling and rapid virtual physical prototyping with emphasis on the thermal analysis. Our findings show that while executing CPU-centric benchmarks (e.g. SPECInt and Dhrystone), the temperature in the server-on-chip (logic+DRAM) is in the range of 175-200°C at a power consumption of less than 20W, exceeding the reliable operating bounds without any cooling solutions, even with embedded cores. However, with real cloud workloads, the power density in the server-on-chip remains much below the temperatures reached by the CPU-centric workloads as a result of much lower power burnt by memory-intensive cloud workloads. We show that such a server-on-chip system is feasible with a low-cost passive heat sink eliminating the need for a high-cost active heat sink with an attached fan, creating an opportunity for overall cost and energy savings in datacenters.

This paper presents a first-order analytical model for determining the performance degradation caused by permanently faulty cells in architectural and non-architectural arrays. We refer to this degradation as the performance vulnerability factor (PVF). The study assumes a future where cache blocks with faulty cells are disabled resulting in less cache capacity and extra misses while faulty predictor cells are still used but cause additional mispredictions. For a given program run, random probability of permanent cell failure, and processor configuration, the model can rapidly provide the expected PVF as well as lower and upper PVF probability distribution bounds for an individual array or array combination. The model is used to predict the PVF for the three predictors and the last level cache, used in this study, for a wide range of cell failure rates. The analysis reveals that for cell failure rate of up to 1.5e-6 the expected PVF is very small. For higher failure rates the expected PVF grows noticeably mostly due to the extra misses in the last level cache. The expected PVF of the predictors remains small even at high failure rates but the PVF distribution reveals cases of significant performance degradation with a non-negligible probability. These results suggest that designers of future processors can leverage trade-offs between PVF and reliability to sustain area, performance and energy scaling. The paper demonstrates this approach by exploring the implications of different cell size on yield and PVF.

Estimating worst-case execution times (WCETs) for architectures with caches requires the worst-case number of cache misses to be upper bounded. Most existing static cache analysis methods use fixed-point computation and do not scale well with large code sizes. To address this scalability issue, we propose in this paper a new fast and scalable instruction cache analysis technique. In contrast to existing work, neither fixed point computation nor heavyweight interprocedural analysis are required. Thus, code sizes too long to analyze with existing techniques are then analyzable with lower analysis time and memory consumption, and with only a slight degradation of the analysis precision. Experimental results show a reduction of the analysis execution time of a factor 5 in average (with a peak near 30 for the largest and most complex code) with only a degradation of the analysis precision of 0.5% in average. The proposed technique is intended to be used in situations where the need for fast analysis outweighs the need for very tight results: early software development phases, timing analysis of large pieces of software, or iterative WCET-oriented compiler optimizations.

Multi-core chips have been increasingly adopted by the microprocessor industry. For real-time systems to exploit multi-core architectures, it is required to obtain both tight and safe estimates of worst-case execution times (WCETs). Estimating WCETs for multi-core platforms is very challenging because of the possible interferences between cores due to shared hardware resources such as shared caches, memory bus, etc. This paper proposes a compile-time approach to reduce shared instruction cache interferences between cores to tighten WCET estimations. Unlike, which accounts for all possible conflicts caused by tasks running on the other cores when estimating the WCET of a task, our approach drastically reduces the amount of inter-core interferences. This is done by controlling the contents of the shared instruction cache(s), by caching only blocks statically known as reused. Experimental results demonstrate the practicality of our approach.

There is a need for using virtual memory in real-time applications: using virtual addressing provides isolation between concurrent processes; in addition, paging allows the execution of applications whose size is larger than main memory capacity, which is useful in embedded systems where main memory is expensive and thus scarce. However, virtual memory is generally avoided when developing real-time and embedded applications due to predictability issues. In this paper we propose a predictable paging system in which the page loading and page eviction points are selected at compile-time. The contents of main memory is selected using an Integer Linear Programming (ILP) formulation. Our approach is applied to code, static data and stack regions of individual tasks. We show that the time required for selecting memory contents is reasonable for all applications including the largest ones, demonstrating the scalability of our approach. Experimental results compare our approach with a previous one, based on graph coloring. It shows that quality of page allocation is generally improved, with an average improvement of 30% over the previous approach. Another comparison with a state-of-the-art demand-paging system shows that predictability does not come at the price of performance loss.

With the advent of increasingly complex hardware in real-time embedded systems (processors with performance enhancing features such as pipelines, cache hierarchy, multiple cores), many processors now have a set-associative L2 cache. Thus, there is a need for considering cache hierarchies when validating the temporal behavior of real-time systems, in particular when estimating tasks' worst-case execution times (WCETs). In this paper, we propose a safe static instruction cache analysis method for multi-level non-inclusive caches. The proposed method is experimented on medium-size and large programs. We show that the method is reasonably tight. We further show that in all cases WCET estimations are much tighter when considering the cache hierarchy than when considering only the L1 cache. An evaluation of the analysis time is conducted, demonstrating that analyzing the cache hierarchy has a reasonable computation time.

Conventionally, the use of virtual memory in real-time systems has been avoided, the main reason being the difficulties it provides to timing analysis. However, there is a trend towards systems where different functions are implemented by concurrent processes. Such systems need spatial separation between processes, which can be easily implemented via the use of the memory management unit (MMU) of commercial processors. In addition, some systems have a limited amount of physical memory available. So far, attempts to provide real-time address spaces have focused on the predictability of virtual to physical address translation and do not implement demand-paging. In this paper we propose a compiler approach to introduce a predictable form of paging, in which page-in and page-out points are selected at compile-time. The problem under study can be formulated as a graph coloring problem, as in register allocation within compilers. Since the graph coloring problem is NP-complete for more than three colors, we define a heuristic, which in contrast to those used for register allocation, aim at minimizing worst-case performance instead of average-case performance. Experimental results applied on tasks code show that predictability does not come at the price of performance loss as compared to standard (dynamic) demand paging.