[jkadlec]

As far as I know, the only way to have something close to Wireshark for CPUs is to do a simulation. Then you can measure network latencies, cache hits and so on, but the simulation will be extremely slow. There's GEM5 (http://www.gem5.org) - a simulator with couple of DSLs designed to do just that, but I'd bet that Intel has their own thing. I did my thesis on simulating MESIF in GEM5, here's a link: https://dip.felk.cvut.cz/browse/pdfcache/kadlej16_2013dipl.p... , that could get people started if they are interested (it also might be a good introduction into the subject of cache coherency protocols). I have the code laying around somewhere, I've been wanting to make it public, but it needs cleanup first.

[irishjohnnie]

I think the closest thing to a Wireshark for CPUs is Intel VTune (https://software.intel.com/en-us/intel-vtune-amplifier-xe)

[hendzen]

Actually, you can buy a JTAG debugger [0] for Intel CPUs. It's going to cost you though.

[0] - https://software.intel.com/en-us/articles/intel-system-studi...

[irishjohnnie]

Wouldn't the equivalent of that be a JTAG debugger for the NIC?

[fmstephe]

Although it isn't equivalent to wireshark, linux's perf command (no doubt osx/win/bsd have their own) is close enough and very easy to use.

It exposes a very large number of CPU counters, including cache misses etc.

Not quite as thorough as simulating. But probably more useful in practice.

[lallysingh]

There was one for powerpcs (PSIM?) that IBM put out during the mac/power days.

But generally you run benchmarks to figure out what the costs are and performance counters to see how often your code is paying those costs.