flat assembler
Message board for the users of flat assembler.

Index > Heap > Nehalem might not do what you tell it to

Author
Thread Post new topic Reply to topic
revolution
When all else fails, read the source


Joined: 24 Aug 2004
Posts: 17287
Location: In your JS exploiting you and your system
revolution
It seems the Intel engineers know better than you about what power states the Nehalem should be in. I just hope they got it right.
... the PCU monitors OS performance state requests, so it can actually make intelligent decisions about what power/performance state to go into, despite what the OS is telling it. ... Nehalem’s PCU can ... attempt to more intelligently decide what power/performance states it will instruct the CPU to go into, regardless of what the OS thinks it wants.
Post 12 Sep 2008, 03:08
View user's profile Send private message Visit poster's website Reply with quote
Madis731



Joined: 25 Sep 2003
Posts: 2141
Location: Estonia
Madis731
I think I need to protect this small fragment of text from any further bashes and flames Wink

If you have followed ALL through the AnandTech's article (and some others maybe) the problem is with OS trying to minimize the latency (response time etc.) and thus guiding the threads to both CPU cores or threads. What Intel CPUs previously did was "overclock" the die when the other one was idle - STILL remaining under the promised TDP. Its all good, but when in comes to Windows, even one process jumps from core to core and there is no idle core. Just both cores taking ½ the load.

This was bad because Windows didn't get performance and Intel's CPUs didn't get to use their powers. Nehalem fixes it making the probability of this happening ... higher. When one of 4 cores is idling, then the other three can be "overclocked".

The other problem (their talking about Windows again - I don't know why - maybe just a good example) they mentioned was that some OSs tend to make too much C-state switches and they loose performance due to penalties encountered while there's a better way: switch less, get more work done, and save more power Smile Intel actually does know better what's good for their CPU than OS-makers. Maybe its the secrets they're holding, but it may well be that an OS doesn't care about the best practices and its easier for the CPU to hold the drum-stick Very Happy
Post 12 Sep 2008, 06:24
View user's profile Send private message Visit poster's website Yahoo Messenger MSN Messenger Reply with quote
revolution
When all else fails, read the source


Joined: 24 Aug 2004
Posts: 17287
Location: In your JS exploiting you and your system
revolution
Actually I didn't intend it as a flame toward Intel. But my comment about hoping they got it right was meaning that I hope they have a setting to either turn it off or lower it's sensitivity when needed. In complex CPUs like these it is very difficult to predict all possible usage patterns.
Post 12 Sep 2008, 07:23
View user's profile Send private message Visit poster's website Reply with quote
Display posts from previous:
Post new topic Reply to topic

Jump to:  


< Last Thread | Next Thread >
Forum Rules:
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You can attach files in this forum
You can download files in this forum


Copyright © 1999-2020, Tomasz Grysztar. Also on YouTube, Twitter.

Website powered by rwasa.