Run-time memory tracking and profiling.

Part of the rerun family of crates.

This is a library for tracking memory use in a running application. This is useful for tracking leaks, and for figuring out what is using up memory.

re_memory includes an opt-in sampling profiler for allocation callstacks. Each time memory is allocated there is a chance a callstack will be collected. This information is tracked until deallocation. You can thus get information about what callstacks lead to the most live allocations, giving you a very useful memory profile of your running app, with minimal overhead.

lib.rs:

Run-time memory tracking and profiling.

First steps

Add re_memory to your Cargo.toml:

cargo add re_memory

Install the AccountingAllocator in your main.rs:

use re_memory::AccountingAllocator;

#[global_allocator]
static GLOBAL: AccountingAllocator<std::alloc::System>
    = AccountingAllocator::new(std::alloc::System);

Checking memory use

Use MemoryUse::capture to get the current memory use of your application.

Finding memory leaks

Turn on memory tracking at the top of your main() function:

re_memory::accounting_allocator::set_tracking_callstacks(true);

Now let your app run for a while, and then call accounting_allocator::tracking_stats to get the statistics. Any memory leak should show up in TrackingStatistics::top_callstacks.

More

See also accounting_allocator.