Linked by Thom Holwerda on Wed 28th Nov 2012 15:17 UTC
Thread beginning with comment 543457
To view parent comment, click here.
To read all comments associated with this story, please click here.
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE[6]: Comment by MOS6510
by tylerdurden on Wed 28th Nov 2012 23:47
in reply to "RE[5]: Comment by MOS6510"
XP was definitively designed with multiprocessing in mind. The NT kernel already supported natively SMP, MMP, and (cc)NUMA years before XP for that matter.
The scheduler and some subsystems have been tweaked through the years, so the newer kernels do appear to behave "faster" for most interactive tasks. Computers keep also getting faster as well.
There were some artificial limitations in the number of cores supported between the "home" (only 1 socket supported) and "pro" versions of XP. But I have used a 16 core workstation using XP, and it ran SMP workloads like a champ.
Edited 2012-11-28 23:50 UTC
RE[7]: Comment by MOS6510
by tylerdurden on Thu 29th Nov 2012 00:19
in reply to "RE[6]: Comment by MOS6510"




Member since:
2006-02-15
The kernel simply wasn't designed for multi-core systems as at the time such systems were first and foremost high-end devices and the situation was not anticipated to change. If you think back to those times people and companies were entirely focused on megahertz - race -- including Intel and AMD -- and people just assumed we'd be running into hundreds of gigahertz on a single core.
There are various kinds of fixes to make XP run slightly better on dual-core systems, but even these do not really fully fix the situation and XP's kernel still suffers a serious penalty on three cores or more.