gpu-trace-perf
This is a rust rewrite of some tooling I built for comparing performance between different graphics driver settings on graphics traces. The goal is for a driver developer to be able to quickly experiment and find how their changes affect the performance of actual rendering.
Right now only apitrace traces are supported. Each draw call gets bracketed with GPU time elapsed queries, and we sum them across the last frame and compare that total between the two drivers. Thus, lower numbers in the percentage change column indicate that your driver change is better.
Installing
apt-get install cargo
cargo install gpu-trace-perf
Example usage
gpu-trace-perf run --traces $HOME/src/traces-db beforedriver afterdriver
This command will find all the traces in traces-db and run them in a loop printing stats until you feel ready to hit ^C.
The beforedriver and afterdriver arguments are scripts in your
path that set the environment to make you use your new driver, like
this:
#!/bin/sh
Requirements
Currently only apitrace traces are supported, and you'll need it installed.
Cross building for your embedded device
Add the following to ~/.cargo/config:
[target.armv7-unknown-linux-gnueabihf]
linker = "arm-linux-gnueabihf-gcc"
[target.aarch64-unknown-linux-gnu]
linker = "aarch64-linux-gnu-gcc"
And set up the new toolchain and build:
rustup target add aarch64-unknown-linux-gnu
cargo build --release --target aarch64-unknown-linux-gnu gpu-trace-perf
scp target/aarch64-unknown-linux-gnu/release/gpu-trace-perf device:bin/
TODO
- See if we can come up with something similar to the apitrace functionality using renderdoc's replay API
- Do per-draw-call stats as an option for drilling down into the effect of a change
- Parse and replay apitraces on our own instead of relying on external apitrace
- Parse and replay renderdoc traces
License
Licensed under the MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)