Search

This work is based on our philosophy of providing interlayer system-level power awareness in computing systems. Here, we couple this approach with our vision of multi-partitioned memory systems, where memory accesses are separated based on their static predictability and memory footprint and managed with various compiler controlled techniques. We show that media applications are mapped more efficiently when scalar memory accesses are redirected to a mini-cache. Our results indicate that a partitioned 8K cache with the scalars being mapped to a 512 byte mini-cache can be more efficient than a 16K monolithic cache from both performance and energy point of view for most applications. In extensive experiments, we report 30% to 60% energy-delay product savings over a range of system configurations and different cache sizes.

In an earlier chapter about the FlexCache project [24], we described our vision of a multipartitioned cache where memory accesses are separated based on their static predictability and memory footprint, and managed with various compiler-controlled techniques supported by instruction set architecture extensions or with traditional hardware control.

Embedded parallel and distributed computing systems for real-time applications are becoming commonplace. Many such real-time applications are life-critical and require extensive fault-tolerance capabilities in order to ensure very high reliability. At the same time, cost, power, weight, and volume constraints require that any introduced redundancy must be efficiently used. Thus, failure-recovery strategies must be implemented to allow the system to most efficiently manage its resources in the presence of one or more failures, while attempting to continue the execution of the current tasks so as to not miss any deadlines. We have developed such a resource-management algorithm, which selects the optimal failure recovery procedure to be employed when a processor failure is detected. Further, in order to test this and other possible recovery algorithms, we have developed the RAPIDS simulator testbed. This paper describes the testbed, and shows how it provides a framework in which to simulate distributed real-time systems. In addition we describe how RAPIDS allows the user to validate fault recovery algorithms.

In this paper, we study the problem of positioning copies of shared data structures to reduce pow er consumption in real-time systems. Power-constrained real-time systems are of increasing importance in defense, space, and consumer applications. We describe our energy consumption model and present numerical results linking the placement of data structures to energy consumption.

Boosted transactions offer an attractive method that enables programmers to create larger transactions that scale well and offer deadlock-free guarantees. However, as boosted transactions get larger, they become more susceptible to conflicts and aborts. We describe a linear-time algorithm to detect transactions that cannot make progress, which transactions need to be aborted, and when . The algorithm guarantees zero false positives with minimal aborts. Our proposals, as implemented in DSTM2, increase the transactional throughput of the system, often by more than 30%.

Transactional Memory (TM) has been proposed as an alternative concurrency mechanism for the shared memory parallel programming model. Its main goal is to make parallel programming for Chip Multiprocessors (CMPs) easier than using the traditional lock synchronization constructs, without compromising the performance and the scalability. This topic has received substantial research attention and several TM designs have been proposed using various TM benchmarks. We believe that the evaluation of TM proposals would be more solid if it included realistic applications, that address on-going TM research issues, and that provide the potential for straightforward comparison against locks. In this paper, we introduce RMS-TM, a Transactional Memory benchmark suite composed of seven real-world applications from the Recognition, Mining and Synthesis (RMS) domain. In addition to featuring current TM research issues such as nesting and I/O and system calls inside transactions, the RMS-TM applications also provide a mix of short and long transactions with small/large read and write sets with low/medium/high contention rates. These characteristics, as well as providing lock-based versions of the applications, make RMS-TM a useful TM tool. Current TM benchmarks do not explore all these features. In our evaluation with selected STM and HTM systems, we find that our benchmark suite is also scalable, which is useful for evaluating TM designs on high core counts.

Hardware-assisted watchpoint systems enhance the execution of numerous dynamic software techniques, such as memory protection, module isolation, deterministic execution, and data race detection. In this paper, we show that previous hardware proposals may introduce significant energy overheads, and propose WatchPoint Filtering (WPF), a novel filtering mechanism that eliminates unnecessary watchpoint checks. We evaluate WPF on two state-of-the-art proposals for hardware-assisted watchpoints using two common memory checkers. WPF eliminates 83% of the watchpoint checks (up to 99.7%) and reduces 57% of the dynamic energy overhead (up to 78%) on average, without introducing additional performance execution overhead.

In this paper we investigate the benefits of turning the concept of transactional conflict from its traditionally fixed definition into a variable one that can be dynamically controlled in software. We propose the extension of the atomic language construct with an attribute that specifies the definition of conflict, so that programmers can write code which adjusts what kinds of conflicts are to be detected, relaxing or tightening the conditions according to the forms of interference that can be tolerated by a particular algorithm. Using this performance-motivated construct, specific conflict information can be associated with portions of code, as each transaction is provided with a local definition that applies while it executes. We find that defining conflicts in software makes possible the removal of dependencies which arise as a result of the coarse synchronization style encouraged by the TM programming model. We illustrate the use of the proposed construct in a variety of use cases with real applications, showing how programmers can take advantage of their knowledge about the problem and other global information not available at run-time. We describe how to implement a hardware TM design that utilizes this software construct. Our experiments reveal that leveraging software-defined conflicts, the programmer is able to achieve significant reductions in the number of aborts--over 50% for most applications. At 16 threads, our system with software-defined conflicts outperforms LogTM-SE in nearly all benchmarks, reaching an average reduction in execution time of 18%.

Deep Neural Networks (DNNs) are inherently computation-intensive and also power-hungry. Hardware accelerators such as Field Programmable Gate Arrays (FPGAs) are a promising solution that can satisfy these requirements for both embedded and High-Performance Computing (HPC) systems. In FPGAs, as well as CPUs and GPUs, aggressive voltage scaling below the nominal level is an effective technique for power dissipation minimization. Unfortunately, bit-flip faults start to appear as the voltage is scaled down closer to the transistor threshold due to timing issues, thus creating a resilience issue.This paper experimentally evaluates the resilience of the training phase of DNNs in the presence of voltage underscaling related faults of FPGAs, especially in on-chip memories. Toward this goal, we have experimentally evaluated the resilience of LeNet-5 and also a specially designed network for CIFAR-10 dataset with different activation functions of Rectified Linear Unit (Relu) and Hyperbolic Tangent (Tanh). We have found that modern FPGAs are robust enough in extremely low-voltage levels and that low-voltage related faults can be automatically masked within the training iterations, so there is no need for costly software-or hardware-oriented fault mitigation techniques like ECC. Approximately 10% more training iterations are needed to fill the gap in the accuracy. This observation is the result of the relatively low rate of undervolting faults, i.e.

Database management systems (DBMS) have become an essential tool for industry and research and are often a significant component of data centres. As a result of this criticality, efficient execution of DBMS engines has become an important area of investigation. This work takes a top-down approach to accelerating decision support systems (DSS) on x86-64 microprocessors using vector ISA exten- sions. In the first step, a leading DSS DBMS is analysed for potential data-level parallelism. We discuss why the existing multimedia SIMD extensions (SSE/AVX) are not suitable for capturing this parallelism and propose a complementary instruction set reminiscent of classical vector architectures. The instruction set is implemented using unin- trusive modifications to a modern x86-64 microarchitecture tailored for DSS DBMS. The ISA and microarchitecture are evaluated using a cycle-accurate x86-64 microarchitectural simulator coupled with a highly-detailed memory simulator. We have found a single oper- ator is responsible for 41% of total execution time for the TPC-H DSS benchmark. Our results show performance speedups between 1.94x and 4.56x for an implementation of this operator run with our proposed hardware modifications.

In this paper, we take a MIPS-based open-source uniprocessor soft core, Plasma, and extend it to obtain the Beefarm infrastructure for FPGA-based multiprocessor emulation, a popular research topic of the last few years both in the FPGA and the computer architecture communities. We discuss various design tradeo s and we demonstrate superior scalability through experimental results compared to traditional software instruction set simulators. Based on our experience of designing and building a complete FPGA-based multiprocessor emulation system that supports runtime and compiler infrastructure and on the actual executions of our experiments running Software Transactional Memory (STM) benchmarks, we comment on the pros, cons and future trends of using hardware-based emulation for research.

Next generation architectures will require innovative solutions to reduce energy consumption. One of the trends we expect is more extensive utilization of compiler information directly targeting energy optimizations. As we show in this paper, static information provides some unique benefits, not available with runtime hardware-based techniques alone. To achieve energy reduction, we use IPC information at various granularities, to adaptively adjust voltage and speed, and to throttle the fetch rate in response to changes in ILP. We evaluate schemes that are based on static IPC, runtime IPC and also combined, hybrid approaches.We show that IPC-based adaptive voltage scaling schemes can reduce energy consumption significantly, but the approach that also uses static IPC information in combination with runtime IPC, better captures program ILP burstiness and helps meet applications' target performance: an important criterion in the real-time domain. We have found that static IPC-based fetch-throttling works very well, in most cases performing similarly or better than hardware-only runtime IPC-based schemes. Overall, static IPC based resource throttling alone can save up to 14% energy in the processor with less than 5\% IPC degradation. The hybrid scheme saves somewhat more energy but at the expense of higher performance degradation than the static-only approach. In fact, we obtain the lowest IPC degradation with the static IPC-based scheme.

Traditionally, system designers have attempted to improve system performance by scheduling the processing cores and by exploring different memory system configurations and there is comparatively less work done scheduling the accesses at the memory system level and exploring data accesses on the memory systems. In this paper, we propose a memory access pattern based controller (MAPC). MAPC organizes data accesses in descriptors, prioritizes them with respect to the number and size of transfer requests. When compared to the baseline multicore system, the MAPC based system achieves between 2.41× to 5.34× of speedup for different applications, consumes 28% less hardware resources and 13% less dynamic power.

Memory reliability will be one of the major concerns for future HPC and Exascale systems. This concern is mostly attributed to the expected massive increase in memory capacity and the number of memory devices in Exascale systems. For memory systems Error Correcting Codes (ECC) are the mostcommonly used mechanism. However state-of-the art hardware ECCs will not be sufficient in terms of error coverage for future computing systems and stronger hardware ECCs providing more coverage have prohibitive costs in terms of area, power and latency. Software-based solutions are needed to cooperate with hardware. In this work, we propose a Cyclic Redundancy Checks (CRCs) based software mechanism for task-parallel HPC applications. Our mechanism incurs only 1.7% performance overheadwith hardware acceleration while being highly scalable at large scale. Our mathematical analysis demonstrates the effectiveness of our scheme and its error coverage. Results show that our CRC-based mechanism reduces the memory vulnerability by 87% on average with up to 32-bit burst (consecutive) and 5-bit arbitrary error correction capability. This work was supported by FI-DGR 2013 scholarship and the European Community’s Seventh Framework Programme [FP7/2007-2013] under the Mont-blanc 2 Project (www.montblanc-project.eu), grant agreement no. 610402 and TIN2015-65316-P.

The power consumption of digital circuits, e.g., Field Programmable Gate Arrays (FPGAs), is directly related to their operating supply voltages. On the other hand, usually, chip vendors introduce a conservative voltage guardband below the standard nominal level to ensure the correct functionality of the design in worst-case process and environmental scenarios. For instance, this voltage guardband is empirically measured to be 12%, 20%, and 16% of the nominal level in commercial CPUs [1], Graphics Processing Units (GPUs) [2], and Dynamic RAMs (DRAMs) [3], respectively. However, in many real-world applications, this guardband is extremely conservative and eliminating it can result in significant power savings without any overhead. Motivated by these studies, we aim to extend the undevolting technique to commercial FPGAs. Toward this goal, we will practically demonstrate the voltage guardband for a representative Xilinx FPGA, with a preliminary concentration on on-chip memories, or Block RAMs (BRAMs). We thank Pradip Bose, Alper Buyuktosunoglu, and Augusto Vega from IBM Watson for their contribution to this work. The research leading to these results has received funding from the European Union’s Horizon 2020 Programme under the LEGaTO Project (www.legato-project.eu), grant agreement nº 780681.

The ever-increasing complexity of high-performance computing applications limits performance due to memory constraints in FPGAs. To address this issue, we propose the Advanced Pattern based Memory Controller (APMC), which supports both regular and irregular memory patterns. The proposed memory controller systematically reduces the latency faced by processors/accelerators due to irregular memory access patterns and low memory bandwidth by using a smart mechanism that collects and stores the different patterns and reuses them when it is needed. In order to prove the effectiveness of the proposed controller, we implemented and tested it on a Xilinx ML505 FPGA board. In order to prove that our controller is efficient in a variety of scenarios, we used several benchmarks with different memory access patterns. The benchmarking results show that our controller consumes 20% less hardware resources, 32% less on chip power and achieves a maximum speedup of 52× and 2.9× for regular and irregular applications respectively. The research leading to these results has received funding from the European Research Council under the European Unions 7th FP (FP/2007-2013) / ERC GA n. 321253. It has been partially funded by the Spanish Government (TIN2012-34557).

Transactional Memory (TM) is being studied widely as a new technique for synchronizing concurrent accesses to shared memory data structures for use in multi-core systems. Much of the initial work on TM has been evaluated using microbenchmarks and application kernels; it is not clear whether conclusions drawn from these workloads will apply to larger systems. In this work we make the first attempt to develop a large, complex, application that uses TM for all of its synchronization. We describe how we have taken an existing parallel implementation of the Quake game server and restructured it to use transactions. In doing so we have encountered examples where transactions simplify the structure of the program. We have also encountered cases where using transactions occludes the structure of the existing code. Compared with existing TM benchmarks, our workload exhibits non-block-structured transactions within which there are I/O operations and system call invocations. There are long and short running transactions (200– 1.3M cycles) with small and large read and write sets (a few bytes to 1.5MB). There are nested transactions reaching up to 9 levels at runtime. There are examples where error handling and recovery occurs inside transactions. There are also examples where data changes between being accessed transactionally and accessed nontransactionally. However, we did not see examples where the kind of access to one piece of data depended on the value of another.

Simulators are key tools for computer architecture research. However, multicore architectures represent a highly complex challenge for software simulators, which may suffer from fidelity loss and long execution times. FPGAs can simulate multicore architectures with scalable performance and high accuracy, but the difficulty of debugging could hinder their adoption. In this paper we propose several techniques for inspection, debugging and verification of multicore architectures, both for software-based and FPGA-based simulations. These debugging extensions are cycle-accurate and unobtrusive. As a proof of concept, we have developed a 24-core RISC multiprocessor that runs the Linux Kernel, for which we provide three simulation modes: a fast, functional simulation; a detailed, cycle-accurate simulation; and a FPGA-based simulation. Our platform can run up to 24 cores and perform full-system verification at 17 million instructions per second.

Fault tolerance has become a fundamental concern in computer design, in addition to performance and power. Although several error detection schemes have been proposed to discover a faulty core in the system, these proposals could waste the whole core, including many error-free structures in it after error detection. Moreover, many fault-tolerant designs require additional hardware for data replication or for comparing the replicated data. In this study, we provide a low-cost, fine-grained error detection scheme by exploiting already existing comparators and data replications in the several pipeline stages such as issue queue, rename logic, and translation lookaside buffer. We reduce the vulnerability of the source register tags in IQ by 60%, the vulnerability of instruction TLB by 64%, the vulnerability of data TLB by 45%, and the vulnerability of the register tags of rename logic by 20%.

Deterministic execution of a multithreaded application guarantees the same output as long as the application runs with the same input parameters. Determinism helps a programmer to test and debug an application and to provide fault-tolerance in the systems based on replicas. Additionally, Transactional Memory (TM) greatly simplifies development of multithreaded applications where applications use transactions (instead of locks) as a concurrency control mechanism to synchronize accesses to shared memory. However, deterministic systems proposed so far are not TM-aware. They violate the main properties of TM (atomicity, consistency and isolation of transactions), and execute TM applications incorrectly. In this paper, we present DeTrans, a runtime system for deterministic execution of multithreaded TM applications. DeTrans executes TM applications deterministically, it executes nontransactional code serially in round-robin order, and transactional code in parallel. Also, we show how DeTrans works with both eager and lazy software TM. We compare DeTrans with Dthreads, a state-of-the-art deterministic execution system. Unlike Dthreads, DeTrans does not use memory protection hardware nor facilities of the underlying operating system (OS) to execute multithreaded applications deterministically. DeTrans uses properties of software TM to ensure deterministic execution. We evaluate DeTrans using the STAMP benchmark suite and we compare DeTrans and Dthreads performance costs. DeTrans incurs less overhead because threads execute in the same address space without any OS system calls overhead. According to our results, DeTrans is 3.99x, 3.39x, 2.44x faster on average than Dthreads for 2, 4 and 8 threads, respectively.

This paper introduces Atomic Dataflow Model (ADF) - a programming model for shared-memory systems that combines aspects of dataflow programming with the use of explicitly mutable state. The model provides language constructs that allow a programmer to delineate a program into a set of tasks and to explicitly define input data for each task. This information is conveyed to the ADF runtime system which constructs the task dependency graph and builds the necessary infrastructure for dataflow execution. However, the key aspect of the proposed model is that it does not require the programmer to specify all of the task’s dependencies exp licitly, but only those that imply logical ordering between tasks. The ADF model manages the remainder of inter-task dependencies automatically, by executing the body of the task within an implicit memory transaction. This provides an easy- to -program optimistic concurrency substrate and enables a task to safely share data with other concurrent tasks. In this paper, we describe the ADF model and show how it can increase the programmability of shared memory systems.

The use of FPGAs in computational workloads is becoming increasingly popular due to the flexibility of these devices in comparison to ASICs, and their low power consumption compared to GPUs and CPUs. However, scientific applications run for long periods of time and the hardware is always subject to failures due to either soft or hard errors. Thus, it is important to protect these long running jobs with fault tolerance mechanisms. Checkpoint-Restart is a popular technique in high-performance computing that allows large scale applications to cope with frequent failures. In this work we approach the fault tolerance of CPU-FPGA heterogeneous applications from a high level by using OmpSs@FPGA environment and a multi-level checkpointing library. We analyse the performance of several different applications and we understand what kind of overheads we can expect from checkpointing computational workloads running on FPGAs. Our results demonstrate overheads as low as 0.16% and 0.66% when checkpointing very frequently, indicating that this technique is efficient and does not add a significant amount of overhead to the system. In addition, we showcase a proof of concept for checkpointing partial data of the FPGA task itself. This can prove useful for workloads in which most data is offloaded to the FPGA memory at once and do not constantly move all the data between the accelerator and the CPU. This research has received funding from the European Union’s Horizon 2020 research and innovation programme under projects EuroEXA (grant agreement nº 754337) and eProcessor (grant agreement nº 956702).

Using low-power symmetric multi-cores on FPGAs are becoming ubiquitous in embedded computing. This is due to the emergence of power and energy as key design metrics, as important as performance. This leads to the requirement of powerful and reliable tools, which will be used for the Design Space Exploration (DSE) based on power and energy at an early stage of the design flow. In this paper, we propose a simulation based virtual platform power and energy estimation tool for heterogeneous Multiprocessor System-on-Chip (MPSoC) based platforms. There are two steps involved in this tool development. The first step is power model generation. For the power model development, we used functional parameters to set up generic power models for different parts of the system. This is a one-time activity. In the second step, a simulation based virtual platform framework is developed to accurately grab the activities used in the related power models generated in the first step. The combination of the two steps leads to a hybrid power estimation, which gives a better trade-off between accuracy and speed. The proposed tool is automated and also scalable for exploring complex embedded multi-core architectures. The efficiency of the proposed tool is validated through multi-cores/processors designed around the FPGAs and extended to accommodate futuristic multi-processors/cores for a reliable energy based DSE. The obtained power/energy estimation results provide less than 4% of error for single core processor, 8% for dual-core processor and 9% for heterogeneous MPSoC based systems when compared to real board measurements.

Due to the growing computational requirements of mobile applications, using a heterogeneous Multiprocessor System-on-Chip becomes an incontrovertible solution to meet the service requirements. Today, Electronic System-Level design is considered as a vital premise to explore design trade-offs for such devices in the early stage of the design flow. This paper proposes a novel system-level power/energy estimation methodology and optimization techniques for heterogeneous CPU-GPU based platforms. There are two parts involved in this methodology. First, we developed the power models by using functional parameters to set up generic power models for different parts of the platform. Second, we designed a simulation based system-level prototype using SystemC (JIT) and Cycle-Accurate simulators to accurately evaluate the activities used in the related power models. The combination of the two parts leads to a novel power estimation methodology at system-level, which gives a good trade-off between accuracy and speed. Moreover, leveraging our methodology, we introduce novel power optimization techniques such as inter-task DVFS and workload balancing at the system-level for CPU-GPU platforms. The efficiency of our proposed methodology and optimization techniques are validated through a CARMA kit, which consists of an ARM quad-core processor and a NVIDIA GPU processor (96 cores). Estimated power and energy values are compared to real board measurements. Our obtained power/energy estimation results provide less than 2.5% of error for single core processor, 4% for dual-core processor, 4% for quad-core, 4% for GPU and 6% multi-processor based systems. By using the proposed optimization techniques, we achieved significant power and energy savings of up to 45% and 70% respectively for various industrial benchmarks.

Recently proposed hybrid dataflow and shared memory programming models combine these two underlying models in order to support a wider range of problems naturally. The effectiveness of such hybrid models for parallel implementations of dense and sparse algebra problems is well known. In this paper, we show another real world example for which hybrid dataflow models provide better support than traditional shared memory models. Specifically, we compare these models using the game engine parallelization as a case study. We show that hybrid dataflow models decrease the complexity of the parallel game engine implementation by eliminating or restructuring the explicit synchronization that is necessary in shared memory implementations. The corresponding implementations also exhibit good scala-bility and better speedup than the shared memory parallel implementations, especially in the case of a highly congested game world that contains a large number of game objects. Ultimately, on an eight core machine we were able to obtain 4.72x speedup compared to the sequential baseline, and to improve 49% over the lock-based parallel implementation based on work-sharing.

In this work, we propose an efficient scheduler and intelligent memory manager known as AMMC (Advanced Multi-Core Memory Controller), which proficiently handles data movement and computational tasks. The proposed AMMC system improves performance by managing complex data transfers at run-time and scheduling multi-cores without the intervention of a control processor nor an operating system. AMMC has been coupled with a heterogeneous system that provides both general-purpose cores and application specific accelerators. The AMMC system is implemented and tested on a Xilinx ML505 evaluation FPGA board. The performance of the system is compared with a microprocessor based system that has been integrated with the Xilkernel operating system. Results show that the AMMC based multi-core system consumes 48% less hardware resources, 27.9% less on-chip power and achieves 6.8x of speed-up compared to the MicroBlaze-based multi-core system.

The current trend in development of parallel programming models is to combine different well established models into a single programming model in order to support efficient implementation of a wide range of real world applications. The dataflow model has particularly managed to recapture the interest of the research community due to its ability to express parallelism efficiently. Thus, a number of recently proposed hybrid parallel programming models combine dataflow and traditional shared memory. Their findings have influenced the introduction of task dependency in the recently published OpenMP 4.0 standard. In this paper, we present DaSH - the first comprehensive benchmark suite for hybrid dataflow and shared memory programming models. DaSH features 11 benchmarks, each representing one of the Berkeley dwarfs that capture patterns of communication and computation common to a wide range of emerging applications. We also include sequential and shared-memory implementations based on OpenMP and TBB to facilitate easy comparison between hybrid dataflow implementations and traditional shared memory implementations based on work-sharing and/or tasks. Finally, we use DaSH to evaluate three different hybrid dataflow models, identify their advantages and shortcomings, and motivate further research on their characteristics.

Transactional memory (TM) is a new optimistic synchronization technique which has the potential of making shared memory parallel programming easier compared to locks without giving up from the performance. This thesis explores four aspects in the research of transactional memory. First, it studies how programming with TM compares to locks. During the course of work, it develops the first real transactional application ¿ AtomicQuake. AtomicQuake is adapted from the parallel version of the Quake game server by replacing all lock-based synchronization with atomic blocks. Findings suggest that programming with TM is indeed easier than locks. However the performance of current software TM systems falls behind the efficiently implemented lock-based versions of the same program. Also, the same findings report that the proposed language level extensions are not sufficient for developing robust production level software and that the existing development tools such as compilers, debuggers, and profilers lack support for developing transactional application. Second, this thesis introduces new set of debugging principles and abstractions. These new debugging principles and abstractions enable debugging synchronization errors which manifest at coarse atomic block level, wrong code inside atomic blocks, and also performance errors related to the implementation of the atomic block. The new debugging principles distinguish between debugging at the language level constructs such as atomic blocks and debugging the atomic blocks based on how they are implemented whether TM or lock inference. These ideas are demonstrated by implementing a debugger extension for WinDbg and the ahead-of-time C# to X86 Bartok-STM compiler. Third, this thesis investigates the type of performance bottlenecks in TM applications and introduces new profiling techniques to find and understand these bottlenecks. The new profiling techniques provide in-depth and comprehensive information about the wasted work caused by aborting transactions. The individual profiling abstractions can be grouped in three groups: (i) techniques to identify multiple conflicts from a single program run, (ii) techniques to describe the data structures involved in conflicts by using a symbolic path through the heap, rather than a machine address, and (iii) visualization techniques to summarize which transactions conflict most. The ideas were demonstrated by building a lightweight profiling framework for Bartok-STM and an offline tool which process and display the profiling data. Forth, this thesis explores and introduces new TM specific optimizations which target the wasted work due to aborting transactions. Using the results obtained with the profiling tool it analyzes and optimizes several applications from the STAMP benchmark suite. The profiling techniques effectively revealed TM-specific bottlenecks such as false conflicts and contentions accesses to data structures. The discovered bottlenecks were subsequently eliminated with using the new optimization techniques. Among the optimization highlights are the transaction checkpoints which reduced the wasted work in Intruder with 40%, decomposing objects to eliminate false conflicts in Bayes, early release in Labyrinth which decreased wasted work from 98% to 1%, using less contentions data structures such as chained hashtable in Intruder and Genome which have higher degree of parallelism.