gs0510

+--------+------------------------+------+------+-----+-------+----------+
|        |        filename        | text | data | bss |  dec  |   hex    |
+--------+------------------------+------+------+-----+-------+----------+
|        |                        |      |      |     |       |          |
| BEFORE | kernel/pid.o           | 8447 | 3894 |  64 | 12405 | 3075     |
| AFTER  |                        | 7667 | 2018 |  64 |  9749 | 2615     |
| BEFORE | kernel/pid_namespace.o | 5692 | 1842 | 192 |  7726 | 1e2e     |
| AFTER  |                        | 5682 | 1842 | 192 |  7716 | 1.00E+24 |
+--------+------------------------+------+------+-----+-------+----------+

gs0510

+--------+------------------------+------+------+-----+-------+----------+
|        |        filename        | text | data | bss |  dec  |   hex    |
+--------+------------------------+------+------+-----+-------+----------+
|        |                        |      |      |     |       |          |
| BEFORE | kernel/pid.o           | 8447 | 3894 |  64 | 12405 | 3075     |
| AFTER  |                        | 3397 | 304  |   0 | 3701  |  e75     |
| BEFORE | kernel/pid_namespace.o | 5692 | 1842 | 192 |  7726 | 1e2e     |
| AFTER  |                        | 2854 | 216  | 16  |  3086 | c0e      |
+--------+------------------------+------+------+-----+-------+----------+

gs0510

Database server, to compile and run, run ./run.sh

cache.go has the cache store. server.go creates the server and serves request. main.go is the driver file.

gs0510

    //Hashmap to store the count call, can compare to strace for numbers!
    let mut map = HashMap::new();

    //allow the child to be traced
    let output = cmd.before_exec(traceme);

    let mut child = cmd.spawn().expect("child process failed");

    let pid = nix::unistd::Pid::from_raw(child.id() as libc::pid_t);

gs0510

gs0510

gs0510

gs0510

How does sandmark work and the context behind the decisions:

• codespeed - looked at how the folks python, Rust and LLVM were visualizing benchmarks, and there wasn’t a straightforward way to use the Rust benchmarks visualizer (also it wasn’t very stable at the time) or the LLVM benchmarks visualizer. Codespeed was being used by a couple of projects, Python’s pypy for example and seemed like a good choice.
  • Within codespeed, there’s also an artifacts directory where you can push data (for example the perf data!). There's also sandmark-analyze where the data from artifacts is put into a jupyter notebook if someone wants to graph it differently.
  • ocamlspeed is a fork because there were bugs in codespeed and getting the patches upstream takes time.
• sandmark is not relevant to our use case since even though it does provide the infrastructure, it moves the compiler and not the code. So you specify the opam package you want to run the compiler benchmarks for, and sandmark builds those package

gs0510

Fill in random order
index: 5.788160 micros/op; 	 172766.472026 op/s; 	 7.774491 MB/s; 	 total time = 57.881601s.
	write amplification in bytes = 10.454486; in nb of writes = 0.000639; 

 RW Read in random order
index: 5.057182 micros/op; 	 197738.594206 op/s; 	 8.898237 MB/s; 	 total time = 50.571817s.
	read amplification in bytes = 70.304486; in nb of reads = 1.016557 

 RO Read in random order
index: 8.763912 micros/op; 	 114104.301487 op/s; 	 5.134694 MB/s; 	 total time = 87.639115s.

gs0510

Fill in random order
index: 5.607651 micros/op; 	 178327.778691 op/s; 	 8.024750 MB/s; 	 total time = 56.076513s.
	write amplification in bytes = 10.454486; in nb of writes = 0.000639; 

 RW Read in random order
index: 4.744557 micros/op; 	 210767.815771 op/s; 	 9.484552 MB/s; 	 total time = 47.445574s.
	read amplification in bytes = 70.292040; in nb of reads = 1.016339 

 RO Read in random order