• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

FEATURED AMD ZEN Discussion (Previous Rumor Thread)

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Why wouldn't you? I still run all the intel junk on my 6850K/5960X no reason to have it punched out to 4.5 all of life, but no reason it shouldn't clock up that high if needed. This isn't 2004 they work a bit better now.

I wasn't aware cool and quiet didn't work in 2004. Not sure what that even has to do with anything here and now....

Your "Intel Junk" has nothing to do with my preference. I've done all the overclocking with power saving features on and off.

There's no super measurable power saving while I'm not exceeding the efficiency threshold. It always draws the most from being under a heavy load opposed to idle, even if the Cpu is pinned to a frequency or not. I like to see my overclock while idle or loaded, I'm always 100% sure the clocks are set at all times.
 
I've found in benchmarking particularly, if you let the cores use power saving features, and this is specifically my Intel 5960x, the result returned is lower. I assume this is because the cores become 'lazy', and have to scale up the frequency, which incurs an associated performance penalty.

This is just my experience. Obviously for real world workloads, daily usage, etc, this may not matter at all, but specifically for benchmarking, those few milliseconds of delay have an impact, so if you're going to benchmark, I would disable power saving features so your cores remain pegged at whatever frequency and voltage you overclock to.

I personally run with them off 24/7, and have been for almost 2 years with this 5960x. Maybe this is compromising longevity, so be it.

I have a plug the PSU is plugged into before the wall, and it draws the same wattage at idle whether pegged at 4.5 or downclocking itself to 1.3. Therefore, I see negligible 'power saving' advantages, at least in my case.

I also left Ryzen pinned to 4.1 24/7, have not tested power consumption, personally do not buy high end CPUs and motherboards to 'save power', bit like buying a Ferrari and trying to get 30 mpg, good luck. =)
 
DWTech, your observations may be accurate for that particular family, but its not the same across all x86 CPUs.

AMD uses a very smart tech of neural networks and predictions to accurately apply power and frequency needed. Same with Intel's Turbo 2.0 for their latest chips. Each iteration is better than the last. So a new round of tests would be needed. P-State OCing is on my list of a lot of crap I'd like to test. :)
 
DWTech, your observations may be accurate for that particular family, but its not the same across all x86 CPUs.

AMD uses a very smart tech of neural networks and predictions to accurately apply power and frequency needed. Same with Intel's Turbo 2.0 for their latest chips. Each iteration is better than the last. So a new round of tests would be needed. P-State OCing is on my list of a lot of crap I'd like to test. :)

Sure, I can appreciate that.

I think it's common sense that even though it doesn't appear to deviate much, it is using more power when the clock speed is set higher all of the time, but it just appears to be quite minimal in the case of my 5960x.

I haven't tested with even my 8350, Skylake, or Ryzen. It could be an entirely more noticable fluctuation on those.
 
I wasn't aware cool and quiet didn't work in 2004. Not sure what that even has to do with anything here and now....

Your "Intel Junk" has nothing to do with my preference. I've done all the overclocking with power saving features on and off.

There's no super measurable power saving while I'm not exceeding the efficiency threshold. It always draws the most from being under a heavy load opposed to idle, even if the Cpu is pinned to a frequency or not. I like to see my overclock while idle or loaded, I'm always 100% sure the clocks are set at all times.

Depends on the goal, my server runs 24/7 the prior CPU in it ran with an OC it's entire life. That chip ran with CnQ enabled and it made a consumption difference over a months time.

Now if we are talking just goofing around or clocking for a competition sure I turn off everything. It made a massive difference back in the day. Now Turbo for both companies is quite a bit smarter, voltage scaling also works quite a bit better.
 
Depends on the goal, my server runs 24/7 the prior CPU in it ran with an OC it's entire life. That chip ran with CnQ enabled and it made a consumption difference over a months time.

Now if we are talking just goofing around or clocking for a competition sure I turn off everything. It made a massive difference back in the day. Now Turbo for both companies is quite a bit smarter, voltage scaling also works quite a bit better.

I'm not sure even saying a goal..... or perhaps preference to do one or the other and final decision.

And just to be sure I'm clear, I have absolutely no objection to people running power savings with a final overclock at X P-state.

Just figured I'm not trying to save power while at the same time consuming more power at a load condition even if P-states are enabled or not While Overclocked. Ultimately, you are using more power if you actually USE the pc.

IE:
"I didn't buy a 220W Cpu to overclock and try and save power"

:shrug:
 
You must have cheap power in Chicago. The average electricity bill in Ontario is $250-300/ mth. Mine isn't that high but most of the power hogs are on natural gas and I have a ton of LED lights.
 
Yes we have cheap power in Virginia and Delaware where my two houses are. Both are served by Electric Cooperatives that provide cheaper rates. My Virginia home has natural gas also for heat and the water heater, the Delaware home is all electric. Even with the oven and dryer on electric, the Virginia bill only averages $80 in winter and $150 in summer with the AC on pretty much 24/7 so the average is probably under $125 a month. The Delaware house is near the beach and used in the summer only, but some heat is left on all winter. That bill averages around $75 per month. Even so, the Virginia home is now on pretty much all LED lighting so hopefully next year's bills will be a little lower. We also just replaced our 20-year old HVAC in Virgina with a higher SEER rated unit so this summer's bills should be a bit lower.
 
You must have cheap power in Chicago. The average electricity bill in Ontario is $250-300/ mth. Mine isn't that high but most of the power hogs are on natural gas and I have a ton of LED lights.

My PC is the least of power consumption. Probably always has been....

Wanna save power? Ditch your microwave.
 
Mine runs full tilt all the time too Shrimp just saying for some it's a valid concern or at least perceived as a concern. Just like sleep/ low power state. I don't use it but many do or at least try to.
 
Mine runs full tilt all the time too Shrimp just saying for some it's a valid concern or at least perceived as a concern. Just like sleep/ low power state. I don't use it but many do or at least try to.

And just to be sure I'm clear, I have absolutely no objection to people running power savings with a final overclock at X P-state.
 
I'm not sure even saying a goal..... or perhaps preference to do one or the other and final decision.

And just to be sure I'm clear, I have absolutely no objection to people running power savings with a final overclock at X P-state.

Just figured I'm not trying to save power while at the same time consuming more power at a load condition even if P-states are enabled or not While Overclocked. Ultimately, you are using more power if you actually USE the pc.

IE:
"I didn't buy a 220W Cpu to overclock and try and save power"

:shrug:

I would be curious what middle pstates showed consumption wise vs a half load on a max overclock.

As for the 220w CPU comment. I drive a cobra and a 450whp pickup truck, neither vehicle is exactly known for gas mileage, doesn't mean I leave every stop light pinned to the floor. There are throttle positions below 100%.
 
That analogy doesn't work. You always have 450 horse power. How much voltage you use might equate to the amount of fuel you use. But mine is 325 horse power and is always available.

You are correct, I don't have to use 1.35v for only 3.6ghz. No doubt.

But alas I'm gonna come back another time. This week people seem sensitive while my original comment wasn't really to turn into some serious debate on saving very small amounts of power.
 
That analogy doesn't work. You always have 450 horse power. How much voltage you use might equate to the amount of fuel you use. But mine is 325 horse power and is always available.

You are correct, I don't have to use 1.35v for only 3.6ghz. No doubt.

But alas I'm gonna come back another time. This week people seem sensitive while my original comment wasn't really to turn into some serious debate on saving very small amounts of power.

You don't know how motors build power do you? I make 450WHP@4700RPM with my foot flat on the floor. I don't always stand with my foot on the floor and let it hit that speed in the same way most people don't run their CPU flat out 24/7
 
Gah, you can throw your 450HP number around all you want. Means nothing if your only turning in the 350 foot pound range. I'd call it a weak 450HP. But I still don't see how HP would be anything to compare to a cpu.

While your using a 450HP gas motor (power saving on), I'll run the electric 450HP @ 0 rpm motor (power savings off) and have a race. But wait, that's not fair? A cpu goes from 800mhz to 3.8ghz in nano seconds. Still can't see a good way to compare this to your Gasser.

However the engine and processor are cooled very much the same.

So you can run air cooling and cool and quiet, I'll run full tilt with liquid cooling and just have 500WHP @ 0rpm

EDIT:

Thinking on power draw and comparing it to a car.

Car not under load at 60 mph uses less fuel than when under load (could be carrying more weight for example) it would use substantially more fuel.

Processor not under load uses less power when idle at 6000mhz than it would under a full load (using lots of transistors for compute).

Both still at the same speed.
 
Last edited:
Back