Open the working folder
Pick a folder for scripts and results. The file tools work only inside it.
Workflows
Disk speed test: a repeatable run, step by step
03
Check a drive and a connection you own
A disk speed test measures sequential speed on large files and random speed on many small reads and writes, and these are different numbers. This guide checks your own drive and then your router with ping, logging both. A website for CPU benchmark comparisons shows only other people's hardware and settings, so your own logged runs matter more.
Describe the test
Say which drive, how large the test file is and how many runs you want. Name the settings to log.
Approve each step
In Ask first mode, read each card before you click OK. Every file write shows as a diff.
Check the result
Compare the logged numbers with the console output yourself, then repeat the run to see the spread.
An example: the bench.py timing pattern
Run bench.py and add the median to results.csv with today's settings.
- write_file
bench.pysaved - Shell
python bench.pyexit 0 - append_file
results.csvrow added - read_file
results.csvread
Waiting for your OK before the next stepRejectAccept
bench.py
import time
def work():
total = 0
for i in range(1000000):
total += i * i % 7
return total
work() # warm-up call, not timed
runs = []
for n in range(5):
start = time.perf_counter()
work()
runs.append(time.perf_counter() - start)
runs.sort()
print("median seconds:", runs[2])bench.py is a CPU timing example that shows the pattern you reuse for a disk test: one warm-up call, 5 timed runs, then the median. It uses no network and writes no files.
Two checks for a beginner
Test your drive with a large file
Before reading disk benchmark results, check the size of the test file: it must be larger than your RAM or cache, otherwise you measure the cache and not the drive. Tell the agent which disk test command you use, or ask it to write a short Python script, and read the diff before anything runs. Shell then runs it and reports the exit code. Log sequential and random speed as separate numbers, delete the test file afterwards and never test a drive that holds the only copy of important data. Repeat the run 5 times and check the median yourself.
Network latency test between your PC and your router
Latency is the round-trip time of a small message. Ask the agent to run ping -n 20 followed by your router or your own server address, and Shell prints the times. Then write_file saves them as latency.txt and grep pulls the lines that hold the times. Add tracert with the same address if you want to see the hops. Keep the traffic small and short, and never point it at servers you do not own. Repeating the same check over several days shows how latency changes. Ping shows the round-trip time only at the moment you run it, not whether a VPS stayed up, so a VPS uptime benchmark needs more than this check. Compare today's file with an earlier one, and check the numbers yourself.
