简体繁体 English

在多核服务器中使用 Z3

[英]Use Z3 in multicore server

原文 2016-10-05 02:28:17 9 1 c++/ multithreading/ z3/ multicore

I want to set up a system that will be fully dedicated to Z3.我想建立一个完全专用于 Z3 的系统。 Let's say it has 4 cores and I would like to use all the power of the machine.假设它有 4 个内核，我想使用机器的所有功能。

I will be solving large formulas that have around 1000 incremental asserts.我将解决具有大约 1000 个增量断言的大型公式。

I would like to solve the formulas in a parallel fashion.我想以并行方式解决这些公式。 I've read this question and I see that a unique Context should be created for every instance solving a formula.我已经阅读了这个问题，我看到应该为每个求解公式的实例创建一个唯一的Context 。

My question then is, what is the most optimal way to use the full system resources (4 cores) and solving formulas with incremental asserts?我的问题是，使用完整系统资源（4 个内核）并使用增量断言求解公式的最佳方法是什么？ Should I create a context per core and somehow sync the push and pops along them to incrementally solve the formulas?我应该为每个核心创建一个上下文，并以某种方式同步推送和弹出以逐步解决公式吗？

Thanks谢谢

1 个解决方案

Expressions created via one context can not be used in another context.通过一个上下文创建的表达式不能在另一个上下文中使用。 So, yes, if those cores/contexts need the same expressions, they will have to be copied and/or translated (see also Z3_translate ).因此，是的，如果这些核心/上下文需要相同的表达式，则必须复制和/或翻译它们（另请参阅Z3_translate ）。