Breaking the Boundaries in Heterogeneous-ISA Datacenters

Barbalace, Antonio; Lyerly, Robert; Jelesnianski, Christopher; Carno, Anthony; Chuang, Ho-Ren; Legout, Vincent; Ravindran, Binoy

doi:10.1145/3037697.3037738

Cited by 37 publications

(61 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We augment this with a simple cross-ISA execution migration policy: when there is an idle big core with an empty job queue, applications are migrated from the small to big core. Cross-ISA migration is accomplished using Popcorn Linux [2]: for each application, the Popcorn compiler generates multiple binaries, one per ISA, wherein all symbols have the same addresses and sizes. During execution, when a migration decision is made, the Popcorn run-time transforms the application's state (registers and stack) between ISAs, and the Popcorn OS migrates memory pages between ISAs, lazily and on-demand.…”

Section: Scheduling Heuristicsmentioning

confidence: 99%

See 1 more Smart Citation

Scheduling HPC workloads on heterogeneous-ISA architectures

Karaoui

Carno

Lyerly

et al. 2019

Proceedings of the 24th Symposium on Principles and Practice of Parallel Programming

Self Cite

View full text Add to dashboard Cite

In this paper, we investigate the effectiveness of multiprocessor architectures with ISA-different cores for executing HPC workloads. Our envisioned design point in the heterogeneous architecture space is one with multiple cache-coherency domains, with each domain hosting cores of a different ISA and no coherency between domains. We prototype such an architecture using an Intel Xeon x86-64 server and a Cavium ThunderX ARMv8 server, interconnected using a high-speed network fabric. We design, implement, and evaluate policies for scheduling HPC applications with the goal of maximizing workload makespan. Our results reveal that such an architecture is most effective for workloads that exhibit diverse execution times on ISA-different CPUs, with gains exceeding 60% over ISA-homogeneous architectures. Furthermore, cross-ISA execution migration can yield gains up to 38%.

show abstract

Section: Scheduling Heuristicsmentioning

confidence: 99%

“…For all other uses, contact the owner/author(s). forms -shared-memory chip multiprocessors [2,7], multiprocessors with multiple cache-coherent domains (and no coherence between domains) [3], and composite-ISA cores [6]. Figure 1.…”

Section: Introductionmentioning

confidence: 99%

Scheduling HPC workloads on heterogeneous-ISA architectures

Karaoui

Carno

Lyerly

et al. 2019

Proceedings of the 24th Symposium on Principles and Practice of Parallel Programming

Self Cite

View full text Add to dashboard Cite

show abstract

“…pairs of machines. In particular, Popcorn Linux is able to migrate tasks between the x86 64 and the arm64 architectures with a sub-second latency [4]. Note that this number does not contain the address space transfer latency that happen on-demand a er migration time.…”

Section: System Modelmentioning

confidence: 99%

“…We propose to implement these concepts by augmenting the Popcorn Linux [4,6] OS. Based on Linux, Popcorn is currently capable of migrating tasks between the arm64 and x86 64 architectures (back and forth) in a two machine setup.…”

Section: Introductionmentioning

confidence: 99%

OS Support for Thread Migration and Distribution in the Fully Heterogeneous Datacenter

Olivier

Kim

Ravindran

2017

Proceedings of the 16th Workshop on Hot Topics in Operating Systems

Self Cite

View full text Add to dashboard Cite

e datacenter is becoming fully heterogeneous, integrating multiple OS-capable CPUs of di erent Instruction Set Architectures in separate machines. ese machines present diverse performance and power consumption pro les and we show that signi cant potential bene ts for both metrics can be expected, should these machines be able to cooperate in the processing of datacenter, multiprogrammed workloads. We advocate that this cooperation should be enabled at the level of the OS, relieving the programmer from any e ort related to the heterogeneity of the managed machines. We propose a distributed OS architecture running on a fully heterogeneous computer cluster, enabling this cooperation through three main components: the abstraction of the entire cluster in a single system image, a distributed shared memory system, and a heterogeneous scheduler. CCS CONCEPTS •So ware and its engineering →Operating systems; Distributed memory; Scheduling;

show abstract

“…However, the research community has been exploring experimental heterogeneous-ISA designs, showing that they provide better performance and energy efficiency than single-ISA heterogeneity. Exploration in this design space includes many forms -shared-memory chip multiprocessors [10,85,86], multiprocessors with multiple cachecoherent domains (and no coherence between domains) [53], and composite-ISA cores [84] -across many settings, ranging from cluster architectures [68] to mobile settings [50].…”

Section: Introductionmentioning

confidence: 99%

Cross-ISA execution of SIMD regions for improved performance

Pang

Lyerly

Ravindran

2019

Proceedings of the 12th ACM International Conference on Systems and Storage

Self Cite

View full text Add to dashboard Cite

We investigate the effectiveness of executing SIMD workloads on multiprocessors with heterogeneous Instruction Set Architecture (ISA) cores. Heterogeneous ISAs offer an intriguing clock speed/parallelism tradeoff for workloads with frequent usage of SIMD instructions. We consider dynamic migration of SIMD and non-SIMD workloads across ISA-different cores to exploit this trade-off. We present the necessary modifications for a general compiler/run-time infrastructure to transform the dynamic program state of SIMD regions at run-time from one ISA format to another for cross-ISA migration and execution. Additionally, we present a SIMD-aware scheduling policy that makes cross-ISA migration decisions that improve system throughput. We prototype a heterogeneous-ISA system using an Intel Xeon x86-64 server and a Cavium ThunderX ARMv8 server and evaluate the effectiveness of our infrastructure and scheduling policy. Our results reveal that cross-ISA execution migration within SIMD regions can yield throughput gains up to 36% compared to traditional homogeneous ISA systems.

show abstract

Breaking the Boundaries in Heterogeneous-ISA Datacenters

Cited by 37 publications

References 39 publications

Scheduling HPC workloads on heterogeneous-ISA architectures

Scheduling HPC workloads on heterogeneous-ISA architectures

OS Support for Thread Migration and Distribution in the Fully Heterogeneous Datacenter

Cross-ISA execution of SIMD regions for improved performance

Contact Info

Product

Resources

About