arXiv subjects

Muhammad Osama

Publications and source records attributed to Muhammad Osama.

At least 19 recordsLinked to original sources

arXiv2026

Lindblad Multiproduct Formulas

We introduce Lindblad Multiproduct Formulas: a quantum error mitigation technique that uses two-dimensional tensor networks contracted with loop-corrected belief propagation. The quantities required to implement the error mitigation scheme that are evaluated with tensor networks can be less computationally expensive to calculate than the expectation values themselves, thus allowing for the possibility of applying our method to certain systems for which tensor network methods may struggle to calculate the observable quantities of interest. The workflow incorporates Clifford rescaling techniques and outputs an estimated error bar. We apply our method to a model of two-dimensional discrete time crystals studied previously and implement it on $65$ qubits arranged in a $3\!\times\!3$ heavy-hexagonal topology on the quantum computer ibm_basquecountry. We show that a GPU implementation of the classical part of our workflow achieves a speedup of up to $5.6\times$.

quant-ph↗

arXiv2026

The Einsum-Enabled Design Space for Graph Algorithms: A BFS Case Study

We propose a principled approach to reasoning about various graph algorithm implementations. We leverage the extended general Einsum notation (EDGE) which allows us to factor complexity along four axes: algebraic manipulation, mapping, format, and low-level implementations. Using breadth-first search (BFS) as a driving example and case study, we apply our methodology to explore over 90 variations across 26 categories of optimization choices for our GPU-based implementations. In addition to showing that our approach is general enough to represent previously discovered algorithmic techniques such as the pull variant of BFS, we discover novel variants that lead to geomean performance benefits ranging from 1.2x to 1.7x over the best Gunrock baseline variation for graphs with mid- to high- normalized degree variance.

cs.DS↗

arXiv2026

Trust but Verify: Mitigating Medical Hallucinations via Post-Hoc Adversarial Auditing and Multi-Agent Feedback Loops

Large Language Models (LLMs) are increasingly deployed in healthcare settings, yet their tendency to hallucinate poses risks when clinical decisions are involved. This study examine whether LLMs recommend recently banned or withdrawn pharmaceuticals when answering clinical questions and tests an agent-based method for reducing such errors. We developed a five-agent "Trust but Verify" system using a single LLM backbone. To measure regulatory knowledge obsolescence, we created an adversarial dataset of 103 clinical MCQs where historically correct answers now refer to banned substances. This scale ensures statistical significance across various therapeutic classes. We evaluated three open-access model families (GPT-OSS, Llama-3, Falcon-3) under vanilla and agentic conditions. Performance was measured via pointwise score, label accuracy, Hallucination Error Rate (HER), and Component Fidelity (CF) score. We also observed clinical safety regression in proprietary models. In default configurations, all models showed high hallucination rates, consistently selecting banned drugs that matched training data patterns. Our proposed agentic architecture reduced HER by approximately 53% across models. Pointwise scores shifted from -0.25 (unsafe recommendation) toward 0.0 (appropriate refusal). The safety audit intercepted dangerous outputs even when models' parametric knowledge favored the banned substance. The proposed multi-agent framework offers a model-agnostic method for enforcing regulatory compliance that prioritizes patient safety over fluent text generation. Our work demonstrates a practical approach for deploying autonomous AI systems in safety-critical healthcare settings. It shows how real-time regulatory data can be integrated into LLM pipelines to support clinical decision-making.

cs.LG↗

arXiv2026

Quokka#: Quantum Computing with #SAT

We present Quokka#, a versatile, open-source Python library for quantum circuit analysis. Quokka# reduces various simulation, verification, and synthesis tasks to weighted model counting (#SAT). It supports universal quantum circuits and a wide variety of gates. Quokka# provides multiple encodings based on different algebraic bases and equivalence-checking methods, enabling key performance trade-offs. Moreover, the new version of Quokka# adds approximate equivalence checking, which is crucial in its synthesis algorithms, since it enables translation between arbitrary gate sets. Its synthesis engine is depth-optimal, making it well-suited to real-world quantum computing. This paper demonstrates the design, extensibility, and use of Quokka#.

quant-ph↗

arXiv2026

Fleet: Hierarchical Task-based Abstraction for Megakernels on Multi-Die GPUs

Modern GPUs adopt chiplet-based designs with multiple private cache hierarchies, but current programming models (CUDA/HIP) expose a flat execution hierarchy that cannot express chiplet-level locality or synchronization. This mismatch leads to redundant memory traffic and poor cache utilization in memory-bound workloads such as LLM inference. We present Fleet, a multi-level task model that maps computation to memory scopes. Fleet introduces Chiplet-tasks, a new abstraction that binds work and data to a chiplet and enables coordination through its shared L2 cache. Wavefront-level, CU-level, and device-level tasks align with existing abstractions, while Chiplet-tasks expose a previously unaddressed level of the hierarchy. Fleet is implemented as a persistent kernel runtime with per-chiplet scheduling, allowing workers within a chiplet to cooperatively execute tasks with coordinated cache reuse. On AMD Instinct MI350 with Qwen3-8B, Fleet achieves 1.3-1.5x lower decode latency than vLLM at batch sizes 1-8 through persistent kernel execution and per-chiplet scheduling. At larger batch sizes, cooperative weight tiling increases L2 hit rate (from 12% to 54% at batch size 32 and from 39% to 61% at batch size 64), reducing HBM traffic by up to 37% and delivering 1.27-1.30x speedup over a chiplet-unaware megakernel baseline.

cs.AR↗

arXiv2026

GPU-Accelerated Quantum Simulation of Stabilizer Circuits

We introduce new parallel algorithms for efficiently simulating stabilizer (Clifford) circuits on GPUs, with a focus on data-parallel tableau evolution and scalable handling of projective measurements. Our approach reformulates key bottlenecks in stabilizer simulation -- such as Gaussian elimination and measurement updates -- into GPU-tailored primitives that eliminate sequential dependencies and maximize memory coalescing. We implement these techniques in QuaSARQ, a GPU-accelerated stabilizer simulator designed for large qubit counts and many-shot sampling. Across a broad benchmark suite reaching 180,000 qubits and depth 1,000 (roughly 130M gates), QuaSARQ shows substantial runtime improvements, with up to 105$\times$ speedup, and over 80% energy reduction on demanding instances. Moreover, QuaSARQ consistently outperforms Stim, a state-of-the-art CPU-optimized stabilizer simulator, as well as Qiskit-Aer (CPU/GPU), Qibo, Cirq, and PennyLane. Finally, QuaSARQ exhibits a significant advantage in many-shot sampling on large workloads. These results demonstrate that our parallel algorithms can significantly advance the scalability of stabilizer-circuit simulation, particularly for workloads involving extensive measurements and sampling.

quant-ph↗

arXiv2025

tritonBLAS: Triton-based Analytical Approach for GEMM Kernel Parameter Selection

We present tritonBLAS, a fast and deterministic analytical model that uses architectural parameters like the cache hierarchy, and relative code and data placement to generate performant GPU GEMM kernels. tritonBLAS explicitly models the relationship between architectural topology, matrix shapes, and algorithmic blocking behavior to predict near-optimal configurations without runtime autotuning. Based on this model, we developed and implemented a lightweight GEMM framework entirely within Triton. We evaluate the performance of tritonBLAS across a diverse set of GEMM problem sizes on modern GPUs. tritonBLAS achieves over 95% of the performance of autotuning solutions, while reducing autotuning time to zero. This makes tritonBLAS a practical drop-in replacement for empirical tuning in production HPC and ML workloads.

cs.DC↗

arXiv2025

Iris: First-Class Multi-GPU Programming Experience in Triton

Multi-GPU programming traditionally requires developers to navigate complex trade-offs between performance and programmability. High-performance implementations typically rely on low-level HIP/CUDA communication libraries that demand substantial engineering effort for even basic overlap patterns, while simpler abstractions often sacrifice performance. We present Iris, a multi-GPU communication library implemented entirely in Python and Triton that eliminates this trade-off. Iris provides tile-based symmetric memory abstractions that naturally align with Triton's programming model, enabling developers to write single-source kernels that seamlessly interleave computation and communication. We demonstrate a taxonomy of compute-communication overlap patterns--from bulk-synchronous to fine-grained workgroup specialization--that can be implemented with minimal code changes in Iris, often requiring just a few additional lines within the same Triton kernel. Our evaluation shows that Iris achieves near-optimal bandwidth utilization in microbenchmarks and delivers up to 1.79x speedup over PyTorch and RCCL for GEMM+All-Scatter workloads, demonstrating that high-level implementations can match or exceed heavily-optimized libraries while dramatically simplifying multi-GPU programming.

cs.DC↗

arXiv2025

HipKittens: Fast and Furious AMD Kernels

AMD GPUs offer state-of-the-art compute and memory bandwidth; however, peak performance AMD kernels are written in raw assembly. To address the difficulty of mapping AI algorithms to hardware, recent work proposes C++ embedded and PyTorch-inspired domain-specific languages like ThunderKittens (TK) to simplify high performance AI kernel development on NVIDIA hardware. We explore the extent to which such primitives -- for explicit tile-based programming with optimized memory accesses and fine-grained asynchronous execution across workers -- are NVIDIA-specific or general. We provide the first detailed study of the programming primitives that lead to performant AMD AI kernels, and we encapsulate these insights in the HipKittens (HK) programming framework. We find that tile-based abstractions used in prior DSLs generalize to AMD GPUs, however we need to rethink the algorithms that instantiate these abstractions for AMD. We validate the HK primitives across CDNA3 and CDNA4 AMD platforms. In evaluations, HK kernels compete with AMD's hand-optimized assembly kernels for GEMMs and attention, and consistently outperform compiler baselines. Moreover, assembly is difficult to scale to the breadth of AI workloads; reflecting this, in some settings HK outperforms all available kernel baselines by $1.2-2.4\times$ (e.g., $d=64$ attention, GQA backwards, memory-bound kernels). These findings help pave the way for a single, tile-based software layer for high-performance AI kernels that translates across GPU vendors. HipKittens is released at: https://github.com/HazyResearch/HipKittens.

cs.LG↗

arXiv2025

Eliminating Multi-GPU Performance Taxes: A Systems Approach to Efficient Distributed LLMs

As large language models (LLMs) continue to scale, their workloads increasingly rely on distributed execution across multiple GPUs. However, the conventional bulk synchronous parallel~(BSP) model used in such settings introduces significant performance inefficiencies. To characterize these bottlenecks, we introduce the ''Three Taxes'' (Bulk Synchronous, Inter-Kernel Data Locality, and Kernel Launch Overhead) as an analytical framework. We propose moving beyond the rigid BSP model to address key inefficiencies in distributed GPU execution. By exploiting libraries like Iris for Triton, we gain access to in-kernel communication primitives that enable the design of novel fine-grained programming patterns, offering greater flexibility and performance than traditional BSP-based approaches. These patterns systematically eliminate the three taxes by creating direct, tile-level producer-consumer pipelines and replacing global barriers with fine-grained dataflow synchronization. Applying this methodology to critical kernels, from the foundational All-Gather + general matrix multiplication operation to the complex Flash Decode algorithm, we observe a 10-20% speedup in end-to-end latency over BSP-based approaches, establishing a more programmable and efficient paradigm for distributed LLM workloads.

cs.DC↗

arXiv2025

Optimizing Attention on GPUs by Exploiting GPU Architectural NUMA Effects

The rise of disaggregated AI GPUs has exposed a critical bottleneck in large-scale attention workloads: non-uniform memory access (NUMA). As multi-chiplet designs become the norm for scaling compute capabilities, memory latency and bandwidth vary sharply across compute regions, undermining the performance of traditional GPU kernel scheduling strategies that assume uniform memory access. We identify how these NUMA effects distort locality in multi-head attention (MHA) and present Swizzled Head-first Mapping, a spatially-aware scheduling strategy that aligns attention heads with GPU NUMA domains to exploit intra-chiplet cache reuse. On AMD's MI300X architecture, our method achieves up to 50% higher performance over state-of-the-art attention algorithms using conventional scheduling techniques and sustains consistently high L2 cache hit rates of 80-97%. These results demonstrate that NUMA-aware scheduling is now fundamental to achieving full efficiency on next-generation disaggregated GPUs, offering a path forward for scalable AI training and inference.

cs.AR↗

arXiv2024

Stream-K++: Adaptive GPU GEMM Kernel Scheduling and Selection using Bloom Filters

General matrix multiplication (GEMM) operations are the fundamental building blocks of computational domains including artificial intelligence (AI). As GPU architectures evolve and high-performance AI becomes increasingly important, optimizing GEMM performance becomes a fundamental problem that needs to be addressed. This paper introduces Stream-K++, an enhancement to the promising Stream-K GEMM scheduling algorithm for workload balancing. We expand Stream-K's scheduling policies from three to seven and implement an efficient solution selection mechanism using Bloom filters. Our approach rapidly eliminates up to 95.8% of unsuitable configurations while maintaining a 100% true-negative rate. Implemented using the AMD Composable Kernel library and evaluated on AMD Instinct MI250X GPUs, Stream-K++ demonstrates significant performance gains (up to 43%) in select scenarios. It remains competitive (within 20% of optimal) for 60-97.6% of problem sizes. Our flexible framework, implemented in the Open-sieve C++ library, allows for easy adaptation to new problem sizes, scheduling policies, or additional tuning parameters, paving the way for future optimizations in GPU-based GEMM operations.

cs.DC↗

arXiv2024

Seer: Predictive Runtime Kernel Selection for Irregular Problems

Modern GPUs are designed for regular problems and suffer from load imbalance when processing irregular data. Prior to our work, a domain expert selects the best kernel to map fine-grained irregular parallelism to a GPU. We instead propose Seer, an abstraction for producing a simple, reproduceable, and understandable decision tree selector model which performs runtime kernel selection for irregular workloads. To showcase our framework, we conduct a case study in Sparse Matrix Vector Multiplication (SpMV), in which Seer predicts the best strategy for a given dataset with an improvement of 2$\times$ over the best single iteration kernel across the entire SuiteSparse Matrix Collection dataset.

cs.DC↗

arXiv2023

The Sparsity Roofline: Understanding the Hardware Limits of Sparse Neural Networks

We introduce the Sparsity Roofline, a visual performance model for evaluating sparsity in neural networks. The Sparsity Roofline jointly models network accuracy, sparsity, and theoretical inference speedup. Our approach does not require implementing and benchmarking optimized kernels, and the theoretical speedup becomes equal to the actual speedup when the corresponding dense and sparse kernels are well-optimized. We achieve this through a novel analytical model for predicting sparse network performance, and validate the predicted speedup using several real-world computer vision architectures pruned across a range of sparsity patterns and degrees. We demonstrate the utility and ease-of-use of our model through two case studies: (1) we show how machine learning researchers can predict the performance of unimplemented or unoptimized block-structured sparsity patterns, and (2) we show how hardware designers can predict the performance implications of new sparsity patterns and sparse data formats in hardware. In both scenarios, the Sparsity Roofline helps performance experts identify sparsity regimes with the highest performance potential.

cs.CV↗

arXiv2023

A Programming Model for GPU Load Balancing

We propose a GPU fine-grained load-balancing abstraction that decouples load balancing from work processing and aims to support both static and dynamic schedules with a programmable interface to implement new load-balancing schedules. Prior to our work, the only way to unleash the GPU's potential on irregular problems has been to workload-balance through application-specific, tightly coupled load-balancing techniques. With our open-source framework for load-balancing, we hope to improve programmers' productivity when developing irregular-parallel algorithms on the GPU, and also improve the overall performance characteristics for such applications by allowing a quick path to experimentation with a variety of existing load-balancing techniques. Consequently, we also hope that by separating the concerns of load-balancing from work processing within our abstraction, managing and extending existing code to future architectures becomes easier.

cs.DC↗

arXiv2023

Stream-K: Work-centric Parallel Decomposition for Dense Matrix-Matrix Multiplication on the GPU

We introduce Stream-K, a work-centric parallelization of matrix multiplication (GEMM) and related computations in dense linear algebra. Whereas contemporary decompositions are primarily tile-based, our method operates by partitioning an even share of the aggregate inner loop iterations among physical processing elements. This provides a near-perfect utilization of computing resources, regardless of how efficiently the output tiling for any given problem quantizes across the underlying processing elements. On GPU processors, our Stream-K parallelization of GEMM produces a peak speedup of up to 14$\times$ and 6.7$\times$, and an average performance response that is both higher and more consistent across 32,824 GEMM problem geometries than state-of-the-art math libraries such as CUTLASS and cuBLAS. Furthermore, we achieve this performance from a single tile size configuration per floating-point precision, whereas today's math libraries employ complex kernel-selection heuristics to select from a large ensemble of kernel variants.

cs.DS↗

arXiv2022

GPU Load Balancing

Fine-grained workload and resource balancing is the key to high performance for regular and irregular computations on the GPUs. In this dissertation, we conduct an extensive survey of existing load-balancing techniques to build an abstraction that addresses the difficulty of scheduling computations on the GPU. We propose a GPU fine-grained load-balancing abstraction that decouples load balancing from work processing and aims to support both static and dynamic schedules with a programmable interface to implement new load-balancing schedules. Prior to our work, the only way to unleash the GPU's potential on irregular problems has been to workload-balance through application-specific, tightly coupled load-balancing techniques. With our open-source framework for load-balancing, we hope to improve programmers' productivity when developing irregular-parallel algorithms on the GPU, and also improve the overall performance characteristics for such applications by allowing a quick path to experimentation with a variety of existing load-balancing techniques. Using our insights from load-balancing irregular workloads, we build Stream-K, a work-centric parallelization of matrix multiplication (GEMM) and related computations in dense linear algebra. Whereas contemporary decompositions are primarily tile-based, our method operates by partitioning an even share of the aggregate inner loop iterations among physical processing elements. This provides a near-perfect utilization of computing resources, regardless of how efficiently the output tiling for any given problem quantizes across the underlying processing elements. On GPU processors, our Stream-K parallelization of GEMM produces a peak speedup of up to 14x and 6.7x, and an average performance response that is both higher and more consistent across 32K GEMM problem geometries than state-of-the-art math libraries such as CUTLASS and cuBLAS.

cs.DC↗

arXiv2022

Essentials of Parallel Graph Analytics

We identify the graph data structure, frontiers, operators, an iterative loop structure, and convergence conditions as essential components of graph analytics systems based on the native-graph approach. Using these essential components, we propose an abstraction that captures all the significant programming models within graph analytics, such as bulk-synchronous, asynchronous, shared-memory, message-passing, and push vs. pull traversals. Finally, we demonstrate the power of our abstraction with an elegant modern C++ implementation of single-source shortest path and its required components.

cs.DC↗