🚀 go-pugleaf

warr wrote:

> George Bell <MasterofTimeADim@netscape.net> wrote in message news:<3E1797D0.4050907@netscape.net>...
> 
> 
>>
>>No, it couldn't possibly be any slower.  Software that was specifically 
>>designed for multiprocessor systems(i.e. by implementing more threads) 
>>might run slower than if it had been designed for a single processor in 
>>mind, because on a single processor the multiple threads would create 
>>more overhead. But this only means the software can only possible run 
>>faster with more CPUs. A single thread process wouldn't run any slower 
>>on a multiprocessor system, it just couldn't take advantage of the 
>>resources as well as the other.
>>
> 
> if a program is written for single CPU only, even though your OS is
> coordinating the 2 cpus and this program may be running on 2 cpus, but
> they are not running concurrently. they are still sequential
> processing.
> 


How do you know this is true for Linux?  I don't believe that the load 
balancing algorithm implemented for SMP by the Linux kernel will make 
such bad decisions when allocating CPU time to the processes.


> and the kernel has to switch between the 2 CPUs for 1 single process.
> it is definitely slower than the single cpu without interprocessor
> communication.
> 
> I didn't see any benchmarks on Linux. 
> 
> but in windows, there are lots of benchmarks that shows faster
> execution on a single CPU than a dual cpu if you use a program that
> isn't multi threaded.



That just tells me that Windows sucks, which I already knew.


> 
> OS is Dual CPU complient doesn't optimize your single threaded
> process.
> 


It may not be optimized but I don't believe it will run slower.  I think 
at worst it will run the same and you'll have unwisely spent money on 
more processing power when you'd been better off investing in SCSI hard 
disks, raid, etc.