• Log In
  • Register

EIST and C states informational.

Forum Search


  • Be respectful to others
  • No spam
  • No NSFW content
  • No piracy or key resellers
  • No link shorteners
  • Offensive content will be removed


PseeCarguy47 7 months ago

Hey forum members, just thought I would share some information about these features. Intel and AMD both use C states. Intel also has EIST. It is useful to sometimes disable C states or EIST and C states when searching for your particular processors maximum roof. Handy for overclocking. But there is really no need to leave them disabled one you have found and tuned for your specific chips roof. With Intel I have found its best to use EIST and not C states. EIST is basically the same but with more steps. I'll use this analogy, think of the CPU as and engine and the computer as a car. You want to move so you hit the throttle. You only need enough throttle for the desired speed. If you need to go 50mph why keep the throttle floored? Or, your traveling at 100 mph but want to stop, would you leave your foot on the throttle, put it in neutral and hit the brakes to stop and then sit there with your foot to the floor engine screaming? Why not let the CPU cool during less intensive parts of a game? Or save some cash in the process? C states and EIST with Intel and AMD are a good thing. Enjoy

Comments Sorted by:

Vareten 2 Builds 1 point 7 months ago


Also C States are different from EIST. Speedstep (EIST) allows a CPU that is in use to slow its speed instead of running at a fixed frequency. C-States take it a step further and nearly shut down entire cores to conserve even more power. If you've left the PC it'll shut down cores with C-States enabled. With just EIST it'll maybe drop down to the 600Mhz range.

PseeCarguy47 submitter 1 point 7 months ago

Thanks, more information, could have shared in the thread, I'm nkt offended. Just thought I would share since there are so many who just completely disable both in thinking there will be more performance gained while gaming. Shutting them down for more performance during gaming is widely spread while not really practical in terms of more performance in games. Don't think the CPU is gonna try to park cores during gaming unless you are playing a game not optimized for four or more cores and you have more. Like a two thread game on a quad. In that case, how would disabling core parking help. Parking and restarting takes less than hundreds of a second.

PseeCarguy47 submitter 1 point 7 months ago

I read quite a bit about my old Q6600 to get more from it. Much came from the operating system. Went from 32bit when donated to 64 on W7, but that explains why it only went down to 600mhz at idle. It didn't bother me but I wondered why I couldn't set its floor any lower than 600 on power settings. But for gaming I don't see gaining much from disabling core parking, would be hardly perceptible if at all.

Allan_M_Systems 1 point 7 months ago

More thoughts:

AMD has "cool n quiet" while Intel has "Speedstep."

Both offer the ability to drop down to lower multipliers and lower voltages at reduced compute load. Unfortunately, they can't always be enabled with overclocking. Depends on the motherboard. I've had some AMD motherboards (gigabyte/biostar boards X 5 examples on 4 different sockets) have no problems running CNQ with over-clocks and offset voltages, on the other hand, some are totally unstable while using CNQ at anything but all-auto settings (Asus boards X 2 examples).

No matter really, as CNQ doesn't save as much power as we might think. At full voltage and clock speeds, C states alone can still bring these CPU's down to a few watts at low/no load, which is fine for a desktop. Also CNQ causes a performance penalty in many conditions anyway, so is best avoided for peak-performing computers.

PseeCarguy47 submitter 1 point 7 months ago

Thanks for adding to the thread Varaten and Allen, that is the point of this thread. I'm all about as much performance as possible with the given equipment. But I also like saving energy. We all have different approaches to getting what we want. My approach is due to a little electrical background knowledge, but I haven't pushed for maximum clocks on a personal pc yet due to limitations in equipment. With my first build I was just too scared. With my second pc, I ran into a board limitation on my Socket 775. First pc was completely capable knowing what I know now and I even built for overclocking. Athlon 1800+ 2003 model. But at the time I had less experience and electrical knowledge. My approach is just a bit different.

I'll try to explain in as little text as possible and many overclocked understand much of what I will explain. So all chips have a designed tdp. We overclock until we reach a roof, at which point V must be increased. Overclocking and More voltage produce more heat. Heat becomes the determining factor besides the silicon lotto. So with what I understand it really comes down to cooling. Its an endless cycle with battling heat. Heat increases resistance. More resistance requires more voltage, which produces more heat. More resistance is more latency. So really, when overclocking, when you hit the chips max, before increasing voltage, you have actually run into a resistance problem. So you should actually should be able to push for higher clocks before increasing voltage by increasing cooling. But for all out performance or breaking records or benchmarking stats, it'll take a combo of both. But both of you have helped me understand chip feature specifics just that much more.