PERFORMANCE MATTERS (joint work with Charlie Curtsinger, Grinnell College) emeryberger.com, @emeryberger Emery Berger College of Information and Computer Sciences UMASS AMHERSTA short time ago : un.bmp Ogle is too slow! OGLE’84 is too slow!Transistors (millions) Clock Speed (MHz) Performance used to be easy 0.001 0.01 0.1 1 10 100 1,000 10,000 1970 1975 1980 1985 1990 1995 gle loading… No mojitos for me… Back to the present…Transistors (millions) Clock Speed (MHz) Performance not easy anymore 0.001 0.01 0.1 1 10 100 1,000 10,000 1970 1975 1980 1985 1990 1995

Being Friendly to Your Hardware Performance Engineering A gentle introduction to hardware for software engineers 2Where does C++ run? 3On an abstract C++ machine 4On an abstract C++ machine? In most practical cases at boot time only Same capacity, different composition => different performance profile From JESD 79-4 DDR4 specificationMemory • Memory system is in the uncore • Cores act Multiple instructions resulting in fewer operations • ISA restrictions may have impact to performance Imaginary ARM mov r20, 0x123456789abcdef0Register renaming 52 Branching Fetch Decode Queue

Poster submission: Modern C++ for Parallelism in High Performance Computing Victor Eijkhout CppCon 2024 Introduction This poster reports on ‘D2D’, a benchmark that explores elegance of expression and context of a High Performance Computing ‘mini-application’. The same code has been implemented using a number of different approaches to parallelism. Implementations are discussed with performance results. Relevance multi-dimensional arrays through ‘mdspan’, it is interesting to explore what C++ can offer for lower level performance critical operations. Scientific computing is an interesting test cases since many algorithms are

Introduction Firsts steps Context Theoretical foundations Outline of an implementation Conclusion High-Performance Numerical Integration in the Age of C++26 Vincent Reverdy Laboratoire d’Annecy de Physique des past, other languages do far better in terms of everything: functionality, ease of use, and even performance This talk The goal is NOT to revolutionize everything or show a library that beats everything yn+1 = yn + h s � i=1 biki ki = f � tn + cih, yn + h s � j=1 aijkj � , i = 1, . . . , s Performance concerns The Butcher Tableau can be very sparse Null coefficients should be optimized away Compilers

volume in terabytes ● Program analysis research and functional programming in a past life ● Love performance, software abstractions, and clean APIsWhy Python? ● Writing extensive APIs in Python - low boilerplate We’re at CppCon :) Why Python? Why C++?● Why? ○ Avoid reimplementing complex code for Python ○ Performance ○ Back and forth with user’s python code ○ Interoperability with data structures in Python - things: ○ Deal with actual pointers and C++ data types ● The compiled program keeps most of the performance and dynamism of an interpreted language, and: ○ is now a C++ .so ○ is not an interpreted scriptCython

is moved into general-purpose registers for computations • Depending on the platform, may see a performance gain at this stageQuat Functions template inline this platform, as with Neon, we now build using SSE registers • We now have two platforms whose optimizations are completely independent • Except for include guards and checking PLT_SIMD is defined, there

Binding Constraints Architecture Substitution Construction Conclusion Symbolic Calculus for High-Performance Computing from Scratch using C++23 Vincent Reverdy Laboratoire d’Annecy de Physique des Particules all know about optimization, performance, parallelism, . . . What this talk is not about Complicated maths (you are smart people, you can do it yourself) High-performance computing (you all know about concepts Symbolic calculus (derivatives, integrals) Full blown custom rule-based rewriting High-performance Since formulas have the entire information on the mathematical AST, it’s possible to generate

ABI 3 . 4SOMETIMES JIT GIVES PERFORMANCE BENEFITS SOMETIMES JIT GIVES PERFORMANCE BENEFITS 3 . 5SOMETIMES JIT GIVES PERFORMANCE BENEFITS SOMETIMES JIT GIVES PERFORMANCE BENEFITS IF I DON'T KNOW WHERE ARCHITECTURE FLAGS WITH THE BEST ARCHITECTURE FLAGS 3 . 5SOMETIMES JIT GIVES PERFORMANCE BENEFITS SOMETIMES JIT GIVES PERFORMANCE BENEFITS IF I DON'T KNOW WHERE THE SOURCE IS GOING TO BE RUN, I CAN'T AOT-COMPILE NOT BE RUNNING THE CODE ON JUST ONE ARCHITECTURE 3 . 5SOMETIMES JIT GIVES PERFORMANCE BENEFITS SOMETIMES JIT GIVES PERFORMANCE BENEFITS IF I DON'T KNOW WHERE THE SOURCE IS GOING TO BE RUN, I CAN'T AOT-COMPILE

Time and space efficiency are where it’s at. How do these benefits break down?JiT Space-time Performance-time tradeoff Again, Alice knows about trading off space and time, what with all the bottles labeled “full upward funargs” allow passing a function with its closure. The paper dives into important optimizations for Smalltalk. It discussed “macro-expansion of v-code into n-code, with caching”, in other the literature. DPF’s performance is either equivalent to or, when it can exploit runtime information, superior to hand-coded demultiplexors. DPF achieves high performance by using a carefully-designed