3.2. Multiload Analysis

Multichase - a pointer chaser benchmark.
Multiload - a superset of multichase which runs latency, memory bandwidth, and loaded-latency.

AWS Configurations -

• Intel machine (32 VCPU): m5.8xlarge
• ARM machine (32 VCPU): m6g.8xlarge
• Machine disk size (gp2): 8 GB
• AWS region: us-west-1b
• Run iterations: 5

Analysis -

Analyzing the latency and bandwidth effects of the machines we employed is crucial for gaining a deeper insight into the impediments that hinder the enhancement of their performance.

Multichase: Memory Read Latency by Byte/Thread and Architecture

In this benchmark the Multichase “simple” test is running, in which a single thread with variable amount of bytes compute, and measured is the average of ChaseNS (Multichase nano seconds <=> latency) out of 5 iterations.

Multiload: Memory BW by Load Threads and Architecture

The results are for Byte/Thread = 1,073,741,824.

• memcpy-libc: 1:1 rd:wr ratio - glibc memcpy()

• memset-libc: 0:1 rd:wr ratio - glibc memset() non-zero data

• memsetz-libc: 0:1 rd:wr ratio - glibc memset() zero data

• stream-sum: 1:0 rd:wr ratio - lmbench stream sum instructions: s+=a[i] (actual binary depends on compiler & -O level)

• stream-triad: 2:1 rd:wr ratio - lmbench stream triad instructions: a[i]=b[i]+(scalar*c[i])

Back

Next