Current hardware development greatly influences the design decisions of modern database systems. For many performance-focused systems, query compilation emerged as an integral part and different approaches for code generation evolved, making use standard compilers, general-purpose compiler libraries, or domain-specific generators. However, primarily focused on dominating x86-64 server architecture; but neglected current developments towards other CPU architectures like ARM RISC...

Dynamic binary instrumentation frameworks are popular tools to enhance programs with additional analysis, debugging, or profiling facilities add optimizations translations without requiring recompilation access source code. They analyze the code, translate into a---typically low-level---intermediate representation, needed transformation and then generate new code on-demand at run-time. Most thereby focus on a fast rewriting process cost of lower quality leading significant slowdown in...

Emulation of other or newer processor architectures is necessary for a wide variety use cases, from ensuring compatibility to offering vehicle computer architecture research. This problem usually approached using dynamic binary translation, where machine code translated, on the fly, host during program execution. Existing systems, like QEMU, focus translation performance rather than overall execution, and extensions, HQEMU, are limited by their underlying implementation. Conversely,...

Providing new parallel programming models/abstractions as a set of library functions has the huge advantage that it allows for an relatively easy incremental porting path legacy HPC applications, in contrast to effort needed when novel concepts are only provided languages or language extensions. However, performance issues be expected with fine granular usage functions. In previous work, we argued binary rewriting can bridge gap by tightly coupling application and at runtime. We showed...

Fast compilation is essential for JIT-compilation use cases like dynamic languages or databases as well development productivity when compiling static languages. Template-based allows fast times, but in existing approaches, templates are generally handwritten, limiting flexibility and causing substantial engineering effort.

Low compilation times are highly important in contexts of Just-in-time compilation. This not only applies to language runtimes for Java, WebAssembly, or JavaScript, but is also crucial database systems that employ query as the primary measure achieving high throughput combination with low execution time. We present a performance comparison and detailed analysis compile JIT back-ends provided by GCC, LLVM, Cranelift, single-pass compiler context queries. Our results show LLVM achieves highest...

Data science workloads frequently include Python code, but Python's dynamic nature makes efficient execution hard. Traditional approaches either treat as a black box, missing out on optimization potential, or are limited to narrow domain. However, deep and integration of user-defined code into data processing systems requires extracting the semantics entire code. In this paper, we propose novel approach for high-level by transforming general functions program generators that generate...

In High Performance Computing (HPC) the performance of applications is paramount, which has led to a wide body work on optimizing compilers. However, compilers alone are naturally limited in their potential, as they cannot exploit run-time information specialize generated code. For this reason, components perform such specializations gaining traction. particular, application or use-case driven have high potential for optimizations, can benefit from explicit guidance user, allows targeted...