table-1-and-3

ScopeAdvice: Finding Over-synchronization and Memory Overheads

We provide the application binaries used in ScopeAdvice for generating results in Table 1 and Table 3.

Evaluation and test environment

We evaluated ScopeAdvice on the below hardware and software specification:

• SM compute capability: sm_86 (NVIDIA RTX 3090)
• Host CPU: AMD Ryzen 5950x
• OS: Ubuntu 20.04 with linux v 5.4.0-42-generic
• GCC version : 9.4.0
• CUDA version: cuda_11.2.r11.2
• CUDA driver version: 470.256.02

Steps to setup and reproduce results

The following are the steps required to reproduce the results. The commands should be run in this folder.

Reproducing results (Tables 1 and 3)

The following command runs compiles ScopeAdvice, runs every application against ScopeAdvice and generates result contained in Table 1 and Table 3. Each subfolder in this folder is dedicated to each application used in the evaluation.

Run the following command in the current folder:

./run.sh

The outputs of ScopeAdvice are parsed to count the number of unique cases caught. The scripts responsible for parsing can be found in table-1-and-3, called process-table-1-and-3.py.

Raw outputs for ScopeAdvice will be contained in table-1-and-3/ABC/fence.out and table-1-and-3/ABC/mem.out where ABC is the application name (e.g., CU for cuML). The file fence.out will report case of over-synchronization and the file mem.out will report the memory overhead.

Final parsed results will be outputted in the terminal and are also contained at table-1-and-3/table-1-result.csv and table-1-and-3/table-3-result.csv in comma-separated format.