Linked by Thom Holwerda on Tue 10th Jun 2008 06:12 UTC
Earlier this month, we reported that The Unofficial Apple Weblog's as well as Ars Technica's sources said that Apple was working on the next version of Mac OS X, dubbed Snow Leopard. The news was that the new release wouldn't focus on new features, but on performance. During yesterday's WWDC 2008 keynote, Steve Jobs confirmed this rumour, and now Apple has published a preview page.
Thread beginning with comment 318027
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[3]: Comment by Kroc
by google_ninja on Tue 10th Jun 2008 17:20
in reply to "RE[2]: Comment by Kroc"
Member since:2006-02-05
50%-100% would be extreme corner cases, the vast majority of the time threads just don't get used enough to make multi-cores make a difference. Dual Core does make one, as you (more or less) are using one core for what you are doing, the other for the os. Quad Core is a complete waste for 95% of uses that a home user computer is used for.
Aparently all support for single core was dumped from 2k8, and honestly I don't notice much in the way of perf difference between it and vista 64 as a workstation machine.
Member since:2006-01-01
50%-100% would be extreme corner cases, the vast majority of the time threads just don't get used enough to make multi-cores make a difference. Dual Core does make one, as you (more or less) are using one core for what you are doing, the other for the os. Quad Core is a complete waste for 95% of uses that a home user computer is used for.
Not all OS follow the MS School of threading. A true SMP based OS would run any thread on any processor/core. In this scenario, the quad core looks like a good idea again, no?
RE[3]: Comment by Kroc
by sakeniwefu on Tue 10th Jun 2008 17:31
in reply to "RE[2]: Comment by Kroc"
Member since:2008-02-26
Recent Original Stories
Recent Comments
Headlines
Random Comments
Random Stories
Random OS Link
Member since:
2006-01-02
Well it's a question of what you're doing.
Likely on average the machine is going to run 0% faster because most client machine time is spent waiting on a user. But for the latency of certain actions, which is what users care about, it is indeed possible to extract huge performance gains, depending on how optimized the particular scenario is.
Also when scaling up, if you're at a cpu count where your scaling becomes poor due to lock contention, you can easily extract 50-100% gains by breaking the locks.