From: Tåle Skogan
Subject: Locking in the runtime
Date: Wed, 15 Dec 2004 21:28:37 +0000
Message-ID: <opsi2i9zt53fve5s@news.online.no>

The Python runtime has a mutex called the "global interpreter lock" which  
reduces the benefit of native threads on machines with multiple CPUs.  
What’s the situation like for Common Lisp implementations with support for  
native threads? Does the locking scheme in the runtime significantly  
reduce the value of multiple CPUs? It would be interesting to hear about  
both commercial and non-commercial implementations.

Tåle
-- 
Using M2, Opera's e-mail client.

From: Frode Vatvedt Fjeld
Subject: Re: Locking in the runtime
Date: Sat, 18 Dec 2004 15:41:31 +0000
Message-ID: <2hu0qjr4us.fsf@vserver.cs.uit.no>

Tåle Skogan <·····@frisurf.no> writes:

> The Python runtime has a mutex called the "global interpreter lock"
> which reduces the benefit of native threads on machines with
> multiple CPUs.  What’s the situation like for Common Lisp
> implementations with support for native threads? Does the locking
> scheme in the runtime significantly reduce the value of multiple
> CPUs? It would be interesting to hear about both commercial and
> non-commercial implementations.

I'd be surprised if any CL implementation that claimed to support OS
threads had any major serialization issues like that. However, I
believe there are/have been a number of implementations with "threads"
that were not "OS threads" in the sense that actual concurrency is/was
not allowed.

-- 
Frode Vatvedt Fjeld