How does Phoenix benefit from hardware caching?

You are runing on a box that depending on the plan might be runing several hundred VMs that’s not exactly a good enviroment for gathering CPU cache stats

True. :slight_smile: I thought that providing perf with the process id would screen out the noise, but maybe not.

I guess if I really want to test this, I need to set up a dedicated physical machine. Maybe I’ll get around to trying that eventually.

1 Like

I’m far behind you but I’m thinking of buying one RPi 3 and one RPi Zero for such purposes exactly. Create an app, compile it to a microSD card with Nerves, put the microSD card in the RPi, boot it and then start observing it through your main machine.

I don’t want to mislead you though. So far I am only having the same idea as yourself but haven’t even checked if RPi’s ARM CPU supports such monitoring…

1 Like

haven’t even checked if RPi’s ARM CPU supports such monitoring…

Ooo, I was thinking a Pi would be a good way to try this, but it didn’t even occur to me that it was ARM and might not support monitoring.

Anyway, if you do try this, maybe we can remote pair on trying the monitoring? Seems like it would be fun. :slight_smile: (Although basically the extent of my knowledge is to try running the perf command I showed.)

1 Like

Sure. I’d love to. But I am struggling a lot in my job and with my health recently and I am having a hard time motivating myself looking to the sides. :frowning:

Plus I am really torn between using Raspbian or a compiled Nerves image. Guess I’ll buy 1x RPi 3 and 2x RPi zero then… :smiley:

1 Like

So, a bit more constructive reply this time.

Right now I don’t have any RPi. I am thinking of buying a few but I have to count pennies recently and there are more important things on my table first. I could probably use my Macbook 12" (2015) for this but I am not sure if the desktop environment won’t skew the results. Plus macOS doesn’t support perf as you mentioned. And my main Windows 10 dev/gaming machine is having dozens of background processes that makes any proper measurement impossible.

I’d be more than willing to cooperate with you on such a monitoring project. I am very unhealthily curious about these things. But I’ll have to get back to you once I bought an RPi – or give me a shout when you do. That way we could compare what happens if the owner monitors the device through their local network, or if the other guy monitors through the internet.

Plus, that might be a good way to find any possible problems with Nerves as well.

You know you can just fire up a minimal live linux distro with elixir from USB Drive to check what you want to check? It’s much less hassle then buying and setting up RPi’s

IMO the point here is to have an independent and separate machine for more objective results. Plus I don’t want to lose all that I’m doing on my dev/gaming machine and stare at a console on my 34" monitor for several hours. :smiley:

Yeah, but if you work on your Windows machine I doubt you work on your Macintosh at the same time :wink: Just saying.

Fair enough, I’m just unsure if the background macOS stuff won’t disturb the measurements is all.

But I will try anyway in a few weeks. You might be right. :wink:

I would simply run Linux from USB on that mac :slight_smile: If you have the code that i can simply run and send you an output, I can run it on a 6 x64 cores with minimal linux setup (the setup might take some time though :P). And I can run circa 16 hours a day, every work day (that means weekends are off ;P)

If you need cache details go to: http://www.cpu-world.com/CPUs/K10/AMD-Phenom%20II%20X6%201100T%20Black%20Edition%20-%20HDE00ZFBK6DGR%20(HDE00ZFBGRBOX).html and scroll to cache.

1 Like

These two posts by Nathan Long could help…


Big Nerd Ranch is one of the big leaders in the iOS development world. Pretty nice to see them also start adopting Elixir.

These two posts by Nathan Long could help…

:joy::joy::joy::joy::joy::joy::joy:

2 Likes