Log-based Performance Counters — Handmade Hero — Episode Guide

0:11Recap and plan for today

2:10Our performance counters are inexpensive, convenient and thread-safe

3:39They report the amount of time an operation takes, but not _when_ it happens

4:18In order to improve our profile view, we need to collect more data

6:02The action of collecting the data should be cheap. Any real work should be deferred till the end of the frame

7:36A log-based system could fit our requirements

8:18Is the amount of logged events going to be too large?

10:27Let's give the log-based approach a try

10:55Can we call rdtscp?

12:15We seem to be in luck^α

13:50rdtsc tells us about elapsed processor time

14:51rdtscp should also report which processor/core is running our code

19:00Trying to see if rdtscp returns the core identity

20:20We don't have a way of disambiguating threads at the moment

20:58"It's kinda nice to get stuff for free"^β

21:13We'll use a lighter version of debug_record to record debug entries

23:22Disambiguating threads, cores, debug_records and debug_record_arrays inside debug_event

24:19We only need a single big DebugEventArray

25:26Keeping track of the position inside the debug event array

25:55Double-buffering the debug event arrays

27:12Filling in the debug event records

28:22The array index will be determined by a preprocessor symbol defined inside build.bat

29:49Defining AtomicAddU32

32:34Telling apart the beginning and end of a timed block using entry types

33:46Pulling together duplicated event recording code into the RecordDebugEvent macro

37:30What is the 32-bit interlocked add-exchange instruction? It's _InterlockedExchangeAdd

40:38We can't do a synchronous exchange of pointers and clear the event index at the same time...

42:08... unless we pack the position and the event index together into a single 64-bit variable

45:41Exchanging debug event arrays at the end of the frame

46:30Transforming the macros into inline functions temporarily for easy stepping

47:40Our debug array is not large enough for the amount of entries we're recording

49:04Bump that number temporarily just to see if that really is happening for reals^γ

49:30The information we log will allow more in-depth profiling operations

50:05Recap on the bundling together of array and event indices

50:49Switching event array indices correctly

53:29CollateRecords() can reproduce the behavior of our old non-log-based event system

57:12We have two event arrays (one for each compilation unit) and that makes the code uglier than it would be if we had better tools

58:25Linearizing the access to the debug_record arrays

1:02:21Fixing compilation errors

1:03:59Accessing the filenames, function names and line numbers of debug_records

1:06:50The log-based system seems to be working

1:07:37Q&A

🗩

1:07:37Q&A

🗩

1:07:37Q&A

🗩

1:08:04elxenoaizd You mentioned that you use a known base address for your memory management. Could you talk a bit more about that? Does that mean I can now find things by just offsetting from that address, and does it mean that if I fwrite this whole block I'll essentially be fwriting the whole game?

🗪

1:08:04elxenoaizd You mentioned that you use a known base address for your memory management. Could you talk a bit more about that? Does that mean I can now find things by just offsetting from that address, and does it mean that if I fwrite this whole block I'll essentially be fwriting the whole game?

🗪

1:08:04elxenoaizd You mentioned that you use a known base address for your memory management. Could you talk a bit more about that? Does that mean I can now find things by just offsetting from that address, and does it mean that if I fwrite this whole block I'll essentially be fwriting the whole game?

🗪

1:09:02butwhynot1 You can specify which functions to optimize by enabling optimization on the command line and surrounding code you don't want optimized with #pragma optimize("", off) ..... #pragma optimize("", on)

🗪

1:09:02butwhynot1 You can specify which functions to optimize by enabling optimization on the command line and surrounding code you don't want optimized with #pragma optimize("", off) ..... #pragma optimize("", on)

🗪

1:09:02butwhynot1 You can specify which functions to optimize by enabling optimization on the command line and surrounding code you don't want optimized with #pragma optimize("", off) ..... #pragma optimize("", on)

🗪

1:09:51elxenoaizd Do you ever find use size_t or do you just use u32, u64, etc?

🗪

1:09:51elxenoaizd Do you ever find use size_t or do you just use u32, u64, etc?

🗪

1:09:51elxenoaizd Do you ever find use size_t or do you just use u32, u64, etc?

🗪

1:10:20elxenoaizd Do you keep track of struct padding when you add / remove fields, or is it something you don't think about too much, so order of fields doesn't matter much?

🗪

1:10:20elxenoaizd Do you keep track of struct padding when you add / remove fields, or is it something you don't think about too much, so order of fields doesn't matter much?

🗪

1:10:20elxenoaizd Do you keep track of struct padding when you add / remove fields, or is it something you don't think about too much, so order of fields doesn't matter much?

🗪

1:10:56butwhynot1 Also, what's the point of the core number in the debug info? It seems the thread ID is the important part

🗪

1:10:56butwhynot1 Also, what's the point of the core number in the debug info? It seems the thread ID is the important part

🗪

1:10:56butwhynot1 Also, what's the point of the core number in the debug info? It seems the thread ID is the important part

🗪

1:11:54Look forward to tomorrow

1:12:35JamesWidman How will we avoid collecting or displaying stats on debug-rendering code?

🗪

1:12:35JamesWidman How will we avoid collecting or displaying stats on debug-rendering code?

🗪

1:12:35JamesWidman How will we avoid collecting or displaying stats on debug-rendering code?

🗪

1:12:53plain_flavored Are profilers as bad as debuggers?

🗪

1:12:53plain_flavored Are profilers as bad as debuggers?

🗪

1:12:53plain_flavored Are profilers as bad as debuggers?

🗪

1:12:58inliferty What do you think about checked exceptions?

🗪

1:12:58inliferty What do you think about checked exceptions?

🗪

1:12:58inliferty What do you think about checked exceptions?

🗪

1:13:29Thanks and a few words on preordering the source and using the GitHub repositories

🗩

1:13:29Thanks and a few words on preordering the source and using the GitHub repositories

🗩

1:13:29Thanks and a few words on preordering the source and using the GitHub repositories

🗩

Keyboard Navigation

Global Keys

Menu toggling

In-Menu and Index Controls

Quotes and References Menus and Index

Quotes, References and Credits Menus

Filter Menu

Filter and Link Menus

Credits Menu