This was an interesting read, along with the companion article about the early development from last year [0].
I feel like this is a very interesting development, but I can't quite put my finger on why. It's also interesting to compare with previous comments about Nvidia have much less available [1]. I wonder if tooling like this (along with Brendan and his team) could prove to be a competitive advantage for Intel?
>A flame graph is a visualization I invented in 2011 for showing sampled code stack traces.
Another day, another lie on the internet.
I though Gregg was cool.
narism•9mo ago
Who is the real inventor in your opinion?
Veserv•9mo ago
Flame graph, as in a display of sampled call stacks growing upward organized per level to aggregate like call stacks, is almost certainly their invention.
Why they chose that over the much more common visualization of downward call stacks that has been standard for decades is a mystery. That is the form most commonly used to display call stacks in trace visualizers since at least the early 2000s (though likely earlier).
Nice tool. Looks like it uses what we now call a "flame chart" (not a flame graph). I don't know if it ever had an original name, but I've seen these in tools from at least the 1990's.
moralestapia•9mo ago
If I understand correctly, your plot is different because you aggregate all of the sampled data; which is in itself quite useful, btw.
While I was going through my B.Sc., I worked a lot with performance optimization, this was around 2008. I used many tools which were slight variations on the idea of visualizing stack traces. I don't recall if that specifically was a thing or not. I believed it was something that has been doing for decades perhaps, because I was touching gdb, perf, etc... I am now reading about it and turns out that the visualization side of it is relatively recent (not 80s but 00s).
Anyway, I apologize for my initial comment being that harsh, I still think you're cool! :D
brendangregg•9mo ago
The answer has always been in the source of flamegraph.pl:
# This was inspired by Neelakanth Nadgir's excellent function_call_graph.rb
# program, which visualized function entry and return trace events. As Neel
# wrote: "The output displayed is inspired by Roch's CallStackAnalyzer which
# was in turn inspired by the work on vftrace by Jan Boerhout".
Neel's did upward "flames" and I think CallStackAnalyzer did as well from memory, so based on the ones I was familiar with this was the precedent. We were also dealing with stacks that were commonly less than 30 frames deep, so it tended to fit on the screen. Also in my original flamegraph.pl implementation was an option to invert the flame graph (--inverted), so I've always let people choose. But I guess I had to pick something as the default.
Neel's was the biggest direct inspiration. I changed it to put the alphabet on the x-axis and reduced the color hues, but they look similar.
No, this samples the GPU as well. That's the novelty.
SleepyMyroslav•9mo ago
While novel it also very far removed from hardware. In sense that aggregating what actually going on with work submitted from multiple queues is hard. Even gathering timing events for start stop of each can be confusing and not adequate when GPUs execute more than one shader at the same time.
That's not to say its not useful I just dont really trust aggregates even on multithreaded CPU if I can't go check raw events.
kammerdiener•9mo ago
It's not using timing-based aggregation. The EU stall samples from hardware include the instruction pointer, which links them to the shaders mapped in the GPU's address space.
anitil•9mo ago
I feel like this is a very interesting development, but I can't quite put my finger on why. It's also interesting to compare with previous comments about Nvidia have much less available [1]. I wonder if tooling like this (along with Brendan and his team) could prove to be a competitive advantage for Intel?
[0] https://www.brendangregg.com/blog/2024-10-29/ai-flame-graphs... [1] https://news.ycombinator.com/item?id=41994469
moralestapia•9mo ago
>A flame graph is a visualization I invented in 2011 for showing sampled code stack traces.
Another day, another lie on the internet.
I though Gregg was cool.
narism•9mo ago
Veserv•9mo ago
Why they chose that over the much more common visualization of downward call stacks that has been standard for decades is a mystery. That is the form most commonly used to display call stacks in trace visualizers since at least the early 2000s (though likely earlier).
moralestapia•9mo ago
See [1], and many others.
1: https://web.archive.org/web/20091124115438/https://graphics....
brendangregg•9mo ago
moralestapia•9mo ago
While I was going through my B.Sc., I worked a lot with performance optimization, this was around 2008. I used many tools which were slight variations on the idea of visualizing stack traces. I don't recall if that specifically was a thing or not. I believed it was something that has been doing for decades perhaps, because I was touching gdb, perf, etc... I am now reading about it and turns out that the visualization side of it is relatively recent (not 80s but 00s).
Anyway, I apologize for my initial comment being that harsh, I still think you're cool! :D
brendangregg•9mo ago
Neel's was the biggest direct inspiration. I changed it to put the alphabet on the x-axis and reduced the color hues, but they look similar.