Hi,

I'm looking for algorithms and usage of core.atomic in real world programs. Any programming language should be fine as long as they make use of the C++11/C11 Memory Model[0].

Thanks in advance!

[0] A non exhaustive list of PL: C, C++, D, Rust.
[1] https://dlang.org/phobos/core_atomic.html
[2] https://llvm.org/docs/Atomics.html
[3] http://en.cppreference.com/w/cpp/atomic/memory_order

On Thursday, 19 November 2020 at 08:42:32 UTC, rm wrote:
> Hi,
>
> I'm looking for algorithms and usage of core.atomic in real world programs. Any programming language should be fine as long as they make use of the C++11/C11 Memory Model[0].
>
> Thanks in advance!
>
> [0] A non exhaustive list of PL: C, C++, D, Rust.
> [1] https://dlang.org/phobos/core_atomic.html
> [2] https://llvm.org/docs/Atomics.html
> [3] http://en.cppreference.com/w/cpp/atomic/memory_order

Watch herb sutter's talks on atomics ("Atomic weapons" is the 3 hour version that goes into more depth)

On 19/11/2020 17:54, Max Haughton wrote:
> On Thursday, 19 November 2020 at 08:42:32 UTC, rm wrote:
>> Hi,
>>
>> I'm looking for algorithms and usage of core.atomic in real world programs. Any programming language should be fine as long as they make use of the C++11/C11 Memory Model[0].
>>
>> Thanks in advance!
>>
>> [0] A non exhaustive list of PL: C, C++, D, Rust.
>> [1] https://dlang.org/phobos/core_atomic.html
>> [2] https://llvm.org/docs/Atomics.html
>> [3] http://en.cppreference.com/w/cpp/atomic/memory_order
> 
> Watch herb sutter's talks on atomics ("Atomic weapons" is the 3 hour version that goes into more depth)

Hi,

Thank you for your reply. It took me long enough to watch the talk.
The talk is mostly correct. The major complaint I have is that you cannot think about weak memory behaviors in terms of instructions re-ordering. A counter example for this is called IRIW (independent reads of independent writes). Where two threads see writes from other two threads in different order.


I work on verifying the usage of these atomics hence looking for real usage to verify.


Thanks again,