Didn't really find a related topic, so I made one.
Today's challenge: E8200 @ 2660 mhz @0,976V. 0,99 was prime stable for almost 16 hours, now it's time to go lower.

Quite interesting, but why?

You could probably go very low indeed with the right settings.

GL with your lower-ness xD

Zero.

This is how.......

Quote from S14 DRIFT :

Quite interesting, but why?

Lower CPU voltage gives you a couple of major benefits:

1. Less heat produced by CPU -> lower case temps (less stress on components) and you can go for a slower fan speed (quieter).
2. Less energy required to power CPU -> Lower electricity bills.

This is the lowest I can run to get it stable but I'm getting around 48 C idle / 62 C load, which is a bit higher than I really want to run 24/7 so I'll probably drop it back down a bit.

The problem with running the CPU lower is you lose performance long before you actualy get low enough to be unstable.

For my 3.2ghz prescott HT I have to have the voltage increased to get optimum performance because the CPU is power hungry.

62º C under load is nothing, most CPUs run quite comfortably at that temperature these days. They make them to handel higher temps because most people wont bother to get realy good cooling.

As long as the case temp stays below 40º C the computer will be fine.

It's temperatures above 70º C you have to worry about, I had my prescott up to 100º C one time before I upgraded my CPU cooler. I was told the old one would be enough, my CPU didn't seem to think so.

As for less power consumption, thats negligable as far as power savings goes, especialy at the CPU.

Quote from DragonCommando :

The problem with running the CPU lower is you lose performance long before you actualy get low enough to be unstable.

Im only lowering the voltage, not the clock of the CPU. If it can do stock clocks at lower voltage, why not? Less heat, less power consumption.

And about the challenge.. at 0,976 superpi ran into an error, although it was prime stable until windows update restarted my PC. Trying 0,984 now.
Btw, it runs 7*500 at stock voltage (~1,22), but for now, I'm trying to find the lowest possible voltage for stock clocks.

I'm running an E8200 in a ASUS P5K Pro, with Thermalright HR-01 Plus in 'passive' mode. Case fans on the top and rear, but the heatsink has no fans. Load temps are around 40-42C in 18C room.

.8v@ 800mhz total system power 123W at the wall @ idle

4 hardrives + DVDRW + Spec you see below in sig

Even though the clock is staying the same, there will be internal errors because of how the CPU works.

CPU's use two voltages to do what they do, they call them Logic High, and Logic Low. these are 1 and 0 respectively.

If you run the CPU voltage too low it will start running into errors where it gets a low voltage (0) on a logic gate that should be high (1). you won't get instability right away because of internal error correction, but it will hurt the performance of the processor because it has to spend time to correct errors.

My prescott had majour problems with this before I upped the voltage, it got bad enough that I had loop errors that would completely hard lock the computer under load.

If the processor is designed to run at a set voltage, it should be run at that voltage.

Quote from DragonCommando :

Even though the clock is staying the same, there will be internal errors because of how the CPU works.

CPU's use two voltages to do what they do, they call them Logic High, and Logic Low. these are 1 and 0 respectively.

If you run the CPU voltage too low it will start running into errors where it gets a low voltage (0) on a logic gate that should be high (1). you won't get instability right away because of internal error correction, but it will hurt the performance of the processor because it has to spend time to correct errors.

My prescott had majour problems with this before I upped the voltage, it got bad enough that I had loop errors that would completely hard lock the computer under load.

If the processor is designed to run at a set voltage, it should be run at that voltage.

After all the tests I've ran, I haven't noticed any kind of slowdown under any tests yet, nor any games and applications.

Quote from DragonCommando :

If the processor is designed to run at a set voltage, it should be run at that voltage.

And never, ever overclock a CPU or graphics card...

If you haven't noticed any slowdowns you are very lucky then, or your CPU is realy good at error correction.

er you might want to read up on how cmos works

Quote from DragonCommando :

I had my prescott up to 100º C one time before I upgraded my CPU cooler.

prescotts (and most newer CPUs) have a thermal limiter in them... if the temperature is over 70C, it will downclock itself to prevent CPU burnout.

IIRC it's set at 85 or 90 degrees as standard.

Quote from S14 DRIFT :

IIRC it's set at 85 or 90 degrees as standard.

i don't think it's that high... (i haven't had my prescott 3.0E run that hot since it's last maintenance cleaning)

knowing my luck, it's probably 80C

edit: isn't the upper thermal limit of prescotts (the point past where they refuse to work, or the point where they break, i forget which) - 95C?

well my system said 100c and it's been pretty accurate, so I would assume that it actualy hit or was close to 100c.

My CPU doesn't seem to throttle at all, It never has. I've had it at 80c and it didn't even drop one Mhz.

CPU-z actualy says that my CPU doesn't support throttling.

lately though it hasn't been getting nearly that hot, I've added alot of extra cooling. Prescotts get realy hot no matter what, my computer used to actualy heat my room in the winter.

I still use my PC as a heater, but that's normally my GFX card which idles at 65 degrees.

I had a Prescott 3.06 before I had this Q6600, it would idle at around 45 and load about 60. This Q6600 idles between 35 and 45, and load is about 50-55.

100c is a bit high for a CPU to hit, besides, something HAS to be wrong if it's hitting that, and it's on the standard cooler at standard clockspeeds.

There was a test on Toms Hardware ages ago, it seems that AMD (older ones, anyway) sucked and would either blow the CPU, or the chip and the mobo together!

I was warned by a Computer Tech shortly after I got the computer that the 3.2 prescott HT generates retarded amounts of heat, but at the same time, I was told I HAD to increase my voltage because its also power hungry.

I upgraded my power supply not even a week after I got it because the original one coulden't keep up, and then a month later I upgraded my cooler and added three extra fans and a side fan just to keep the thing cool, otherwhise it would hit critical and my computer would hard lock.

That is the first and last computer I will ever buy pre built, I usualy build my own, but for that one I got lazy and had someone else do it, and I paid for it in the end. Problem is, I'm way behind in my knowlege of parts, the last CPUs I know the particulars on are the last P4s.

Wow, seems like Pentium really dicked up then, if thier standard cooler isn't efficient enough.

Although in my experience the cooler on the P4 was very 'dust happy', for some reason. I'd have to clean it out every week or it'd idle at 65 degrees. My room is even more messy now but I've yet to clean out the cooler on my Q.

yeah, prescotts are hot, but they aren't that hot... at least mine isn't...

normally idles around 45c, idles around 50-55c when the stock hsf is clogged.

i've only had it hit the throttling zone a few times... last time was last year.


i forget what i was doing at the time...

edit: for a comparison, my xp2000+ chip idles at 52c, even if i freeze out my apartment by leaving the windows open. (whereas my prescott's fan would spin so slow that the sensor registers 0rpm)

Quote from bunder9999 :

Mar 11 13:20:18 mythtv kernel: CPU1: Temperature above threshold
Mar 11 13:20:18 mythtv kernel: CPU0: Temperature above threshold
Mar 11 13:20:18 mythtv kernel: CPU0: Running in modulated clock mode
Mar 11 13:20:18 mythtv kernel: CPU1: Running in modulated clock mode


i forget what i was doing at the time...

My processor has never throttled at all, either it is defective, or its one odd prescott that doesn't do it. Like I said, CPU-z says it doesn't support throttling.

Well, It seems that even at 0,992, superpi always freezes. But, prime95 runs 24+ hours without error, games also, so I'll hang with 0.976 till It's error free.
Hm, I'm curious what's the lowest voltage it would run stable, if I let's say lower the multiplier and run it 6*333. Thats the next challenge

Quote from amp88 :

This is the lowest I can run to get it stable but I'm getting around 48 C idle / 62 C load, which is a bit higher than I really want to run 24/7 so I'll probably drop it back down a bit.

48 / 62? omg... with my old-as-hell chassie my e85 idles at 29°c and peaks at 37°C... and thats when it's overclocked :O

Quote from Feffe85 :

48 / 62? omg... with my old-as-hell chassie my e85 idles at 29°c and peaks at 37°C... and thats when it's overclocked :O

Btw you will find that his CPU is a Quad core + has a further 6mb cache. Which tends to add to a CPU's TDP .

Quote from Gener_AL (UK) :

Btw you will find that his CPU is a Quad core + has a further 6mb cache. Which tends to add to a CPU's TDP .

well thats true i know, but a friend of mine has a q66, which is slightly cooler then mine just thought it was a little high, not that it will kill it but anyway...

nice to see someone trying to use lower wattage :P