benchmark-discuss

hello this is a heads up that in v1.9.0 (and currently at HEAD) the way we report Time for multi-

hello! v1.8.4 has been released. full changelog: https://github.com/google/benchmark/releases/tag/v1.

hello no, nothing is sent to anywhere. a cursory glance at the source should allow you to be

I have tried on my m2 and it works fine. Maybe you haven't mentioned the libraries . On Friday,

hello i think we caught up on the discord but to repeat here: the pause/resume will add some overhead

hello! v1.8.3 is now available: https://github.com/google/benchmark/releases/tag/v1.8.3 also, if you

hello previous attempts at having IRC or other channels haven't resonated well, so i've set

hello! can you share any efforts you've made? is it the work being done by the reader and writer

i don't have any a priori expectations about what would be faster or slower, and i assume that

hello! 15 months ago, we renamed the master branch to main. shortly afterwards, i was asked to create

i think you'd want to link to libpfm and define HAVE_LIBPFM. those two things are all the cmake

I wrote a benchmark with 10 iterations and 20 repetitions. The execution is wrapped with the unix

CPU Processors with a large number of transistors of microscopic dimensions produce electrical

My question arises from https://github.com/google/benchmark/blob/main/docs/user_guide.md#custom-

i remember seeing this before. i think it suggests the pthread you're linking in is not built for

(I mentioned this topic in https://github.com/google/benchmark/issues/1140 before I found this

step 7 should read: "steps 3 -> 5" are repeated On Friday, 26 February 2021 at 10:33:44

[simplesslife@ONESTEPONEDAY test_checktime]$ ./test 2021-01-25T16:35:52+08:00 Running ./test Run on (

Thanks Dominic :-) I've added a comment with a link to the repo in your linked open issue ( HTML

Hi, I'm experimenting with multithreaded benchmarks and trying to figure out what the numbers

I just wanted to post a final update to say that the benchmarks now show the expected results. Really

The atomic counter itself is a workaround to synchronize between threads (so that state.iteration()

KeepRunning is largely deprecated in favour of the auto loop over state. KeepRunningBatch is a

No, that's not what I am looking for. The linked example is unable to create a shared object. I

Hi, I wanted to ask if benchmark-master folder should be placed in the project one before or after

Hi Dominic, Thanks for your feedback. Happy to know it works well for you and on Linux too. I hope it

I see, so do I need to manully do the calculations time/interation when comparing different benchmark

Please do! On Wed, Aug 21, 2019 at 4:45 AM Sayan Bhattacharjee <aero....@gmail.com> wrote:

Did you see https://github.com/google/benchmark#manual-timing ? This was added largely for the GPU

Sorry, I didn't update the thread. Yes, v1.5.0 (as our build scripts require three sections in