An official website of the United States government
Concurrent kernel execution on GPU has proven an effective technique to improve system throughput by maximizing the resource utilization. In order to increase programmability and meet the increasing memory requirements of data-intensive applications, current GPUs support Unified Virtual Memory (UVM), which provides a virtual memory abstraction with demand paging. By allowing applications to oversubscribe GPU memory, UVM provides increased opportunities to share GPU resources across applications. However, in the presence of applications with competing memory requirements, GPU sharing can lead to performance degradation due to thrashing. NVIDIA's Multiple Process Service (MPS) offers the capability to space share bare metal GPUs, thereby enabling cluster workload managers, such as Slurm, to share a single GPU across MPI ranks with limited control over resource partitioning. However, it is not possible to preempt, schedule, or throttle a running GPU process through MPS. These features would enable new OS-managed scheduling policies to be implemented for GPU kernels to dynamically handle resource contention and offer consistent performance. The contribution of this paper is two-fold. We first show how memory oversubscription can impact the performance of concurrent GPU applications. Then, we propose three methods to transparently mitigate memory interference through kernel preemption and scheduling policies. To implement our policies, we develop our own runtime system (PILOT) to serve as an alternative to NVIDIA's MPS. In the presence of memory over-subscription, we noticed a dramatic improvement in the overall throughput when using our scheduling policies and runtime hints.
more »
« less
This content will become publicly available on May 8, 2026
GOLDYLOC: Global Optimizations & Lightweight Dynamic Logic for Concurrency
Modern accelerators like GPUs increasingly execute independent operations concurrently to improve the device’s compute utilization. However, effectively harnessing it on GPUs for important primitives such as general matrix multiplications (GEMMs) remains challenging. Although modern GPUs have significant hardware and software GEMM support, their kernel implementations and optimizations typically assume each kernel executes inisolationand can utilize all GPU resources. This approach is highly efficient when kernels execute in isolation, but causes significant resource contention and slowdowns when kernels execute concurrently. Moreover, current approaches often onlystaticallyexpose and control parallelism within an application, without considering runtime information such as varying input size and concurrent applications – often exacerbating contention. These issues limit performance benefits from concurrently executing independent operations. Accordingly, we propose GOLDYLOC, which considers theglobalresources across all concurrent operations to identify performant GEMM kernels, which we call globally optimized (GO)-Kernels. GOLDYLOC also introduces a lightweight dynamic logic which considers thedynamicexecution environment for available parallelism and input sizes to execute performant combinations of concurrent GEMMs on the GPU. Overall, GOLDYLOC improves performance of concurrent GEMMs on a real GPU by up to 2 × (18% geomean per workload) versus the default concurrency approach and provides up to 2.5 × (43% geomean per workload) speedup over sequential execution.
more »
« less
- Award ID(s):
- 2238608
- PAR ID:
- 10589449
- Publisher / Repository:
- Association for Computing Machinery
- Date Published:
- Journal Name:
- ACM Transactions on Architecture and Code Optimization
- ISSN:
- 1544-3566
- Subject(s) / Keyword(s):
- Concurrency-aware Execution, General Matrix Multiplication, GPGPUs
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Graphics Processing Units (GPUs) exploit large amounts of thread-level parallelism to provide high instruction throughput and to efficiently hide long-latency stalls. The resulting high throughput, along with continued programmability improvements, have made GPUs an essential computational resource in many domains. Applications from different domains can have vastly different compute and memory demands on the GPU. In a large-scale computing environment, to efficiently accommodate such wide-ranging demands without leaving GPU resources underutilized, multiple applications can share a single GPU, akin to how multiple applications execute concurrently on a CPU. Multi-application concurrency requires several support mechanisms in both hardware and software. One such key mechanism is virtual memory, which manages and protects the address space of each application. However, modern GPUs lack the extensive support for multi-application concurrency available in CPUs, and as a result suffer from high performance overheads when shared by multiple applications, as we demonstrate. We perform a detailed analysis of which multi-application concurrency support limitations hurt GPU performance the most. We find that the poor performance is largely a result of the virtual memory mechanisms employed in modern GPUs. In particular, poor address translation performance is a key obstacle to efficient GPU sharing. State-of-the-art address translation mechanisms, which were designed for single-application execution, experience significant inter-application interference when multiple applications spatially share the GPU. This contention leads to frequent misses in the shared translation lookaside buffer (TLB), where a single miss can induce long-latency stalls for hundreds of threads. As a result, the GPU often cannot schedule enough threads to successfully hide the stalls, which diminishes system throughput and becomes a first-order performance concern. Based on our analysis, we propose MASK, a new GPU framework that provides low-overhead virtual memory support for the concurrent execution of multiple applications. MASK consists of three novel address-translation-aware cache and memory management mechanisms that work together to largely reduce the overhead of address translation: (1) a token-based technique to reduce TLB contention, (2) a bypassing mechanism to improve the effectiveness of cached address translations, and (3) an application-aware memory scheduling scheme to reduce the interference between address translation and data requests. Our evaluations show that MASK restores much of the throughput lost to TLB contention. Relative to a state-of-the-art GPU TLB, MASK improves system throughput by 57.8%, improves IPC throughput by 43.4%, and reduces application-level unfairness by 22.4%. MASK's system throughput is within 23.2% of an ideal GPU system with no address translation overhead.more » « less
-
Alternating Least Square (ALS) is a classic algorithm to solve matrix factorization widely used in recommendation systems. Existing efforts focus on parallelizing ALS on multi-/many-core platforms to handle large datasets. Recently, an optimized ALS variant called eALS was proposed, and it yields significantly lower time complexity and higher recommending accuracy than ALS. However, it is challenging to parallelize eALS on modern parallel architectures (e.g., CPUs and GPUs) because: 1) eALS’ data dependence prevents it from fine-grained parallel execution, thus eALS cannot fully utilize GPU's massive parallelism, 2) the sparsity of input data causes poor data locality and unbalanced workload, and 3) its large memory usage cannot fit into GPU's limited on-device memory, particularly for real-world large datasets. This paper proposes an efficient CPU/GPU heterogeneous recommendation system based on fast eALS for the first time (called HEALS) that consists of a set of system optimizations. HEALS employs newly designed architecture-adaptive data formats to achieve load balance and good data locality on CPU and GPU. HEALS also presents a CPU/GPU collaboration model that can explore both task parallelism and data parallelism. HEALS also optimizes this collaboration model with data communication overlapping and dynamic workload partition between CPU and GPU. Moreover, HEALS is further enhanced by various parallel techniques (e.g., loop unrolling, vectorization, and GPU parallel reduction). Evaluation results show that HEALS outperforms other state-of-the-art baselines in both performance and recommendation quality. Particularly, HEALS achieves up to 5.75 x better performance than a state-of-the-art ALS GPU library. This work also demonstrates the possibility of conducting fast recommendations on large datasets with constrained (or relaxed) hardware resources, e.g, a single CPU/GPU node.more » « less
-
null (Ed.)Contemporary GPUs support multiple kernels to run concurrently on the same streaming multiprocessors (SMs). Recent studies have demonstrated that such concurrent kernel execution (CKE) improves both resource utilization and computational throughput. Most of the prior works focus on partitioning the GPU resources at the cooperative thread array (CTA) level or the warp scheduler level to improve CKE. However, significant performance slowdown and unfairness are observed when latency-sensitive kernels co-run with bandwidth-intensive ones. The reason is that bandwidth over-subscription from bandwidth-intensive kernels leads to much aggravated memory access latency, which is highly detrimental to latency-sensitive kernels. Even among bandwidth-intensive kernels, more intensive kernels may unfairly consume much higher bandwidth than less-intensive ones. In this article, we first make a case that such problems cannot be sufficiently solved by managing CTA combinations alone and reveal the fundamental reasons. Then, we propose a coordinated approach for CTA combination and bandwidth partitioning. Our approach dynamically detects co-running kernels as latency sensitive or bandwidth intensive. As both the DRAM bandwidth and L2-to-L1 Network-on-Chip (NoC) bandwidth can be the critical resource, our approach partitions both bandwidth resources coordinately along with selecting proper CTA combinations. The key objective is to allocate more CTA resources for latency-sensitive kernels and more NoC/DRAM bandwidth resources to NoC-/DRAM-intensive kernels. We achieve it using a variation of dominant resource fairness (DRF). Compared with two state-of-the-art CKE optimization schemes, SMK [52] and WS [55], our approach improves the average harmonic speedup by 78% and 39%, respectively. Even compared to the best possible CTA combinations, which are obtained from an exhaustive search among all possible CTA combinations, our approach improves the harmonic speedup by up to 51% and 11% on average.more » « less
-
Finite-state automata serve as compute kernels for many application domains such as pattern matching and data analytics. Existing approaches on GPUs exploit three levels of parallelism in automata processing tasks: 1)~input stream level, 2)~automaton-level and 3)~state-level. Among these, only state-level parallelism is intrinsic to automata while the other two levels of parallelism depend on the number of automata and input streams to be processed. As GPU resources increase, a parallelism-limited automata processing task can underutilize GPU compute resources. To this end, we propose AsyncAP, a low-overhead approach that optimizes for both scalability and throughput. Our insight is that most automata processing tasks have an additional source of parallelism originating from the input symbols which has not been leveraged before. Making the matching process associated with the automata tasks asynchronous, i.e., parallel GPU threads start processing an input stream from different input locations instead of processing it serially, improves throughput significantly and scales with input length. When the task does not have enough parallelism to utilize all the GPU cores, detailed evaluation across 12 evaluated applications shows that AsyncAP achieves up to 58× speedup on average over the state-of-the-art GPU automata processing engine. When the tasks have enough parallelism to utilize GPU cores, AsyncAP still achieves 2.4× speedup.more » « less
