Automatic generation of custom SIMD instructions for Superword Level Parallelism

SIMD instructions are often implemented in modern multimedia oriented processors. Although SIMD instructions are useful for many digital signal processing applications, most compilers do not exploit SIMD instructions. The difficulty in the utilization of SIMD instructions stems from data parallelism in registers. In assembly code generation, the positions of data in registers must be noted. A technique of generating pack instructions which pack or reorder data in registers is essential for exploitation of SIMD instructions. This paper presents a code generation technique for SIMD instructions with pack instructions. SIMD instructions are generated by finding and grouping the same operations in programs. After the SIMD instruction generation, pack instructions are generated. In the pack instruction generation, Multi-valued Decision Diagram (MDD) is introduced to represent and to manipulate sets of packed data. Experimental results show that the proposed code generation technique can generate assembly code with SIMD and pack instructions performing repacking of 8 packed data in registers for a RISC processor with a dual-issue coprocessor which supports SIMD and pack instructions. The proposed method achieved speedup ratio up to about 8.5 by SIMD instructions and multiple-issue mechanism of the target processor.

SIMD (single-instruction multiple-data) instruction set extensions are quite common today in both high performance and embedded microprocessors, and enable the exploitation of a specific type of data parallelism called SLP (Superword Level Parallelism). While prior research shows that significant performance savings are possible when SLP is exploited, placing SIMD instructions in an application code manually can be very difficult and error prone. In this paper, we propose a novel automated compiler framework for improving superword level parallelism exploitation. The key part of our framework consists of two stages: superword statement generation and data layout optimization. The first stage is our main contribution and has two phases, statement grouping and statement scheduling, of which the primary goals are to increase SIMD parallelism and, more importantly, capture more superword reuses among the superword statements through global data access and reuse pattern analysis. Further, as a complementary optimization, our data layout optimization organizes data in memory space such that the price of memory operations for SLP is minimized. The results from our compiler implementation and tests on two systems indicate performance improvements as high as 15.2% over a state-of-the-art SLP optimization algorithm.

Post-quantum cryptography is considered to provide security against both traditional and quantum computer attacks. Dilithium is a digital signature algorithm that derives its security from the challenge of finding short vectors in lattices. It has been selected as one of the standardizations in the NIST post-quantum cryptography project. Hardware-software co-design is a commonly adopted implementation strategy to address various implementation challenges, including limited resources, high performance, and flexibility requirements. In this study, we investigate using compact instruction set extensions (ISEs) for Dilithium, aiming to improve software efficiency with low hardware overheads. To begin with, we propose tightly coupled accelerators that are deeply integrated into the RISC-V processor. These accelerators target the most computationally demanding components in resource-constrained processors, such as polynomial generation, Number Theoretic Transform (NTT), and modular arithmetic. Next, we design a set of custom instructions that seamlessly integrate with the RISC-V base instruction formats, completing the accelerators in a compact manner. Subsequently, we implement our ISEs in a chip design for the Hummingbird E203 core and conduct performance benchmarks for Dilithium utilizing these ISEs. Additionally, we evaluate the resource consumption of the ISEs on FPGA and ASIC technologies. Compared to the reference software implementation on the RISC-V core, our co-design demonstrates a remarkable speedup factor ranging from 6.95 to 9.96. This significant improvement in performance is achieved by incorporating additional hardware resources, specifically, a 35% increase in LUTs, a 14% increase in FFs, 7 additional DSPs, and no additional RAM. Furthermore, compared to the state-of-the-art approach, our work achieves faster speed performance with a reduced circuit cost. Specifically, the usage of additional LUTs, FFs, and RAMs is reduced by 47.53%, 50.43%, and 100%, respectively. On ASIC technology, our approach demonstrates 12, 412 cell counts. Our co-design provides a better tradeoff implementation on speed performance and circuit overheads.

Customized processor performance generally increases as additional custom instructions are added. However, performance is not the only metric that modern systems must take into account; die area and energy efficiency are equally important. Resource sharing during synthesis of instruction set extensions (ISEs) can significantly reduce the die area and energy consumption of a customized processor. This may increase the number of custom instructions that can be synthesized with a given area budget. Resource sharing involves combining the graph representations of two or more ISEs which contain a similar subgraph. This coupling of multiple subgraphs, if performed naively, can increase the latency of the extension instructions considerably, and yet, as we show in this paper, an appropriate level of resource sharing provides a significantly simpler design with modest increases in average latency for ISEs. Our main contributions are the introduction of a parametric method for exploring the tradeoffs that can be achieved between instruction latency and implementation complexity, and the coupling of design-space exploration with fast area-delay models for the operators comprising each ISE. We present experimental evidence that our heuristic exposes a broad range of design points, allowing advantageous tradeoffs between die area and latency to be found and exploited.

Use of single instruction multiple data (SIMD) functional units enables multimedia systems to exploit parallelism to a higher degree resulting in significant system performance improvements. While implementation of whole SIMD system functionality for an application results in wastage of area resources, we have observed that for a specific multimedia application, we only need to implement a customized SIMD unit that is a subset of whole SIMD standard implementation. Based on this study, we have proposed an extension to the traditional system design and synthesis flow by integrating a methodology of SIMD unit synthesis. Our system synthesizes a customized SIMD unit along with an extended instruction set and generates an equivalent version of assembly code for the application using the extended instruction set. The results of area and performance obtained by experimenting over our implementation of AltiVec compatible customized SIMD units show the effectiveness of our approach.

RISC-V is a modern open source Instruction Set Architecture (ISA) and designed in a very extendable manner, which allows for highly application specific solutions. However, the identification of suitable instruction set extensions usually requires a significant manual effort and therefore is a very challenging process. In this paper we propose a lightweight alternative methodology to find suitable application-specific RISC-V extensions, using a Virtual Prototype (VP). This is done, purely by observing the used instructions during the execution of the targeted application on the VP. Therefore, no further information about the application itself are needed. In this context the advantages and the flexibility of a VP and the straightforward extendability of the RISC-V ISA are demonstrated.

The Advanced Encryption Standard is used in many embedded devices to provide security. In the last years, several researchers have proposed to enhance general-purpose processors with custom instructions to increase the efficiency of cryptographic algorithms. In this work we have evaluated the impact of such instruction set extensions on the implementation security of AES. We have compared several AES implementation options which incorporate state-of-the-art software countermeasures against power-analysis attacks—with and without the use of instruction set extensions. For both scenarios we provide a thorough analysis for different countermeasures with regard to security, performance, and memory. We have found that even a moderate level of protection requires a considerable overhead both in terms of speed and memory. The instruction set extensions, which have been solely designed to increase performance, help to reduce this overhead, but it still remains high. An implementation with proper protection through software countermeasures is only feasible in a setting where the need for resistance against power analysis outweighs the need for performance.

The correctness of processors is a key for their application. Although some verification methods were developed and successfully applied to conventional microprocessors, only a few of them were used in the context of application specific devices. This work introduces a formal verification approach for a reconfigurable microprocessor with extendable instruction set. The application of this approach is demonstrated using register transfer description of the CoMPARE processor and the Stanford Validity Checker as prover. Some undesired side effects of different instructions that were not discovered during the simulation were found by the verification process. In addition some deficiencies of the hardware description notation as specification formalism were shown.

When it comes to performance, embedded systems share many problems with their higher-end counterparts. The growing gap between top processor frequency and memory access speed, the memory wall, is one such problem. Driven, in part, by low energy consumption and low cost requirements, embedded systems are often customized to a single application, or a very small set of applications. In addition, time-to-market requirements and the increasing complexity of embedded systems drives the need for fully or partially automated design tools and also to the extensive use of caches and cache hierarchies. The introduction of multi-processor-based embedded platforms has accelerated this trend; as the design space for embedded systems has grown, designers have become unclear as to whether automatic processor customization tools can cope with this increased complexity. The recent introduction of new techniques addressing the automatic customization, such as Architecturally Visible Storage (AVS) memory-enhanced Instruction Set Extension (ISE) identification algorithms, has also created new challenges. AVS memories are distinct from the cache hierarchy and rely on Direct Memory Access (DMA) transfers to communicate with main memory. In an embedded system containing hardware-managed caches, these extra AVS memories, in combination with their corresponding DMA transfers, cause coherence and consistence problems. Although the problems of coherence and consistence are well known in multi-processor systems, conventional solutions may be expensive in terms of area and power consumption, rendering them unacceptable for use in embedded systems. This thesis presents two low cost coherence mechanisms that solve these two problems. The first mechanism addresses embedded systems that already contain a hardware coherence protocol, like many high-end embedded multi-processor systems. Traditionally, the DMA transactions are transparent to the hardware coherence protocol. By ensuring visibility of these DMA transactions to the hardware coherence protocol, coherence can be guaranteed between AVS memories and data cache(s). As a result, minor changes to the DMA engine are required. Moreover, by forcing the processor pipeline to stall if a DMA transfer is active, memory consistence can be guaranteed. This mechanism provides significant speedup when compared to the execution of a non-ISE-enhanced system; however, due to the increase in bus traffic, this speedup comes at the expense of an increase in energy consumption. Coherent and Speculative DMA are both implementations of this mechanism. Single-processor systems do not contain hardware coherence protocols, and would therefore benefit from a lower-cost solution to the coherence and consistence problems than a hardware coherence protocol. By tightly coupling the AVS memories to the hardware cache, coherence and consistence for the complete system can be guaranteed. This coupling requires insignificant changes to the hardware cache's hit detection circuitry and state machine without influencing its critical path, thus it is inherently inexpensive. This mechanism provides significant speedups and reduces the energy consumed if compared to the execution on non-ISE-enhanced systems. Furthermore, the tight coupling enables direct communication between the AVS memories and the data cache, making this mechanism independent from the processor-to-memory distance. Virtual Ways and Way Stealing are both implementations of this mechanism. Besides enforcing coherence and consistence, the ability to integrate the architectural changes into an automated design flow is important. This thesis shows the influence of Coherent DMA, Speculative DMA, Virtual Ways, and Way Stealing on the ISE-identification algorithm. It shows the architectural requirements and the cost for enforcing coherence and consistence that need to be taken into account when applying these mechanisms in an automated flow without formulating new algorithms.

In both hardware and software, masking can represent an effective means of hardening an implementation against side-channel attack vectors such as Differential Power Analysis (DPA). Focusing on software, however, the use of masking can present various challenges: specifically, it often 1) requires significant effort to translate any theoretical security properties into practice, and, even then, 2) imposes a significant overhead in terms of efficiency. To address both challenges, this paper explores the use of an Instruction Set Extension (ISE) to support masking in software-based implementations of a range of (symmetric) cryptographic kernels including AES: we design, implement, and evaluate such an ISE, using RISC-V as the base ISA. Our ISE-supported first-order masked implementation of AES, for example, is an order of magnitude more efficient than a software-only alternative with respect to both execution latency and memory footprint; this renders it comparable to an unmasked implementation using the same metrics, but also first-order secure.

Way Stealing is a simple architectural modification to a cache-based processor that increases the data bandwidth to and from application-specific instruction set extensions (ISEs), which increase performance and reduce energy consumption. Way Stealing offers higher bandwidth than interfacing the ISEs the processor's register file, and eliminates the need to allocate separate memories called architecturally visible storage (AVS) that are dedicated to the ISEs, and to ensure coherence between the AVS memories and the processor's data cache. Our results show that Way Stealing is competitive in terms of performance and energy consumption with other techniques that use AVS memories in conjunction with a data cache.

Java bytecodes can be executed with the following three methods: a Java interpreter running on a particular machine interprets bytecodes; a Just-in-Time (JIT) compiler translates bytecodes to the native primitives of the particular machine and the machine executes the translated codes; and a Java processor executes bytecodes directly. The first two methods require no special hardware support for the execution of Java bytecodes and are widely used currently. The last method requires an embedded Java processor, picoJavaI or picoJavaII for instance. The picoJavaI and picoJavaII are simple pipelined processors with no ILP (instruction level parallelism) and TLP (thread level parallelism) supports. A so-called MAJC (microprocessor architecture for Java computing) design can exploit ILP and TLP by using a modified VLIW (very long instruction word) architecture and vertical multithreading technique, but it has its own instruction set and cannot execute Java bytecodes directly. In this paper, we investigate a processor architecture which can directly execute Java bytecodes meanwhile can exploit Java ILP and TLP simultaneously. The proposed processor consists of multiple slots implementing horizontal multithreading and multiple functional units shared by all threads executed in parallel. Our architectural simulation results show that the Java processor could achieve an average 20 IPC (instructions per cycle), or 7.33 EIPC (effective IPC), with 8 slots and a 4-instruction scheduling window for each slot. We also check other configurations and give the utilization of functional units as well as the performance improvement with various kinds of working loads.

In order to improve the processing speed of highly complex video coding that involves high data rate and large amount of computation, a parallel video coding system optimization (PVCSO) method is proposed in this paper. By using PC cluster and general H.264 video coding standard for parallel coding experiments, the important factors that are critical to parallel coding efficiency are thoroughly examined and analyzed, such as the number of computation nodes, performance of the network and cluster, parallel coding mechanism, algorithm complexity and raw data distribution mode. Based on the experimental results and analysis, the PVCSO method is proposed and used to design and build practical high efficiency parallel video coding system. For illustration, four high efficiency application-oriented PC cluster based parallel video coding systems are designed and shown. Experimental results show that the PVCSO method can effectively balance the conflicting efficiency factors to achieve the best parallel efficiency and cost performance of the parallel coding system. It provides an effective and practical solution for designing high efficiency parallel video coding system which can significantly improve the processing speed of video coding with huge amount of computation.

The Berlekamp–Massey algorithm finds the shortest linear feedback shift register for a binary input sequence. A wide range of applications like cryptography and digital signal processing use this algorithm. This research proposes novel parallel mechanisms offered by heterogeneous CPU and GPU hardwares in order to achieve the best possible performance for BMA. The proposed bitwise implementation of the BMA algorithm is almost 35 times faster than state of the art implementations. This further improvement is achieved by using SIMD instructions which provides data level parallelism. This new approach can be 4.6 and 35 times faster than a bitwise CPU and state of the art implementations, respectively. In order to achieve the highest possible speedup over a multi-core structure, a multi-threading implementation is introduced in this research. By leveraging on OpenMP we were able to obtain a speedup of 10 times over 12 cores server. The GPU device with thousands of processing cores can bring great speedup over the best CPU implementation. Two other parallel mechanisms offered by GPU are concurrent kernel execution and streaming. They achieve 14.5 and 2.2 times of speedup compared to CPU serial and typical CUDA implementations, respectively. Also, the performance of the openMP code with SIMD instructions is compared with GPU stream implementation. The effectiveness of the proposed method is evaluated in a real world error correction application and it achieves 6.8 times of speedup.

GPUs offer a powerful bulk synchronous programming model for exploiting data parallelism; however, branch divergence amongst executing warps can lead to serious performance degradation due to execution serialization. We propose a novel profile guided approach to optimize branch divergence while transforming a serial program to a data-parallel program for GPUs. Our approach is based on the observation that branches inside some data parallel loops although divergent, exhibit repetitive regular patterns of outcomes. By exploiting such patterns, loop iterations can be aligned so that the corresponding iterations traverse the same branch path. These aligned iterations when executed as a warp in a GPU, become convergent. We propose a new metric based on the repetitive pattern characteristics that indicates whether a data-parallel loop is worth restructuring. When tested our approach on the well-known Rodinia benchmark, we found that it is possible to achieve upto 48% performance improvement by loop restructuring suggested by the patterns and our metrics.