The Official OC (OverClocking) Thread!

I think what may have happened is you've got one of the less overclock-friendly steppings, and it basically will respond to the same OC, but at a far higher Vcore. Only so much is good, and I'd recommend settling for what you can get from 1.4-1.45V, anything above it probably pushing it.

 

where have i seen this recommendation before?

http://forums.afterdawn.com/thread_view.cfm/25/87039#3157543

LOL, just messing w/ u sam~

 

Indeed, but it's relatively common knowledge now I hope, especially if more of us mention it!

 

fasfrank,

since you mentioned you had thoughts about going into water-cooling, this MAY be something you would want to consider for a COMPLETE (ie. includes CPU water block, SPP water block, MCP water block, VREG water block) basically ALL the essential parts on a mobo that could do well with water cooling...

for the $500 price, it's obviously NOT "budget friendly" and all these parts and accessories could be purchased separately for substantially less with even HIGHER quality components used, but i think it's a novel idea by eVGA to offer a mobo with ALL the goodies needed that you can hook up your existing water cooling to without the need to 'mod' the board yourself.

my own eVGA mobo has already been fitted with water blocks from various companies (ie. SWIFTECH APOGEE GTX EXTREME for cpu block, DANGER DEN MAZE 4 INTEL CHIPSET WATER BLOCK FOR NVIDIA 680I MOTHERBOARDS for chipset blocks, DANGER DEN GPU-6001 GPU WATER BLOCK FOR NVIDIA 8800GTX for my video cards...everything on 1/2" barbs/tubing~ )

i'm sure it's arguable that there are BETTER performance blocks available out there for custom water cooling, but this is what i currently have on my rig. i'm not one to fuss over "x brand is better than y brand" simply because the other "better" peformance parts were not available at the time for purchase when i built the water-cooling portion of it; and i'm sure within the next 6 months to a year, a newer "flagship" product will be introduced by the companies i chose to go with above and i will no longer have the "latest and greatest" (availability is a major factor of concern for me, i will NOT order something internationally because U.S. markets do not carry what i want instock)

just wanted to throw some ideas out there for ya regarding the overclocking aspect of things, the gains are significant going from air cooling to water cooling...and i expect the gains are exponentially multiplied when you go to phase change~ but unfortunately for me, i have zero experience dealing with phase change cooling and it's a bit too intimidating and daunting of a task for me to consider it an option in the near future. going to water from high end air cooling is one thing i've overcome in terms of initial hesitancy, but bringing refrigeration into the equation is way over my head.

docTY

edit: i forgot to put up the LINK for the Black Pearl, it's not even out yet, but it sure does look nice...

http://www.evga.com/articles/365.asp

 

Hi again,

I'm sort of continuing this from the Dream PC thread so here we go again!


I'm listing what I'm setting the Vcore at in the bios, and not what it's reporting. When I select and set a voltage, say 1.45, then reboot and then go back into bios, it will report the current Vcore as 1.43 while still showing the selected voltage as 1.45.

setting...reported
1.35 = 1.32
1.40 = 1.36
1.45 = 1.41
1.50 = 1.47
1.55 = 1.52
1.60 = 1.56
1.65 = 1.60

After updating the bios I get the same readings. I'm running Orthos again... CPU is at 3.612, Vcore is set to 1.575, reporting 1.54. Core temps are 39°C to 50°C for almost the entire test. On one portion though it does get up to 54°C. This lasts about a minute and then it drops back down.

I hear what everyone is saying about Vcore, my thoughts are that if I'm going to use Vcore as my overclocking limit then cooling (air or liquid) no longer becomes a factor... If I set the limit to 1.4- 1.45 actual Vcore then my max core temps won't exceed 38°C. That limits my overclock to about 3.2 Ghz.

Here's a screenshot of my last Orthos test:



Tell me if I'm wrong here....
When overclocking you are going to raise the cpu/fsb frequency to achieve a target speed.
Vcore is raised to provide stability as needed.
More Vcore means more heat.
Cooling is used to keep the cpu/core temps down to a reasonable level.
If heat is a limiting factor then an improvement in cooling will allow greater Vcore thus higher cpu speeds...
That's how I'm approaching this.

 

fasfrank,

This is true to a degree as there are electronic limits that don't concern heat. It's a bit like taking a light bulb and freezing it so you could run higher voltage! Not going to happen! Same thing with the cores as they have a limit other than heat too!

I don't know what you are using to monitor your core temps, but the core temps posted are not right! they should be and are much higher than that. I use Everest Ultimate Edition and I get core temps that high at idle with my 4300 with a CPU temp of 26-28C. Something isn't right as the core temp sensors are a part of the CPU internally and usually are more accurate because they are in direct contact with the cores. I suspect you have fried the sensors for the cores. I did the same thing to my D-940 where all of a sudden I was getting low, but false readings. The CPU worked all right but I RMA'd it because it made me nervous not knowing what the temps actually were. The MB eventually failed too, because of a cheap PS! Another lesson learned the hard way!

I was running my 4300 at 3.0 showing 32,000+ MIPS in Sandra 2007 Pro. In actual work it was slower than when set at 3.276 and only reporting 30,000+MIPS. I ran a test with the same DVD processed with DVDRB/CCE at both settings. the slower MIPS/higher CPU speed processed the DVD 7 minutes faster! Numbers don't always tell the whole story. my personal guess as to why this happens is "throughput"! I think the higher fsb enables the data to move faster both before and after processing and that's why it's faster when it shows slower numbers in Sandra! Here's what the benches looked like at 3.01GHz!







At my current settings the MIPS are about 2000 less, the MB about 5000MB/s higher and the MM about 5,000it/s slower, but it's a faster overall performer. As you can see by my current memory bandwidth, another part of the increased throughput is the higher bandwidth!



Clock On,
theone :}

 

Yep, up the Vcore massively, and with a mega cooling setup it'll run cool but it'll expire after 6 months. Electrical changes start happening inside the CPU when you up the vcore, and not very often good ones!

 

sammorris,

First thing that happened to mine was the temps went down for no good reason and never went to the same levels again! I remember when I posted about it, Sophocles came back and told me the thermal sensors were gone and to back it down some before I lost the CPU! The only good thing is that the sensors went blooey first! When I changed the MB and the CPU, it ran and benched at 4.01GHz, but being an older tech board and already having fried one, thanks to my cheap PS, I dialed it down to 3.71GHz. Still going strong today as my Roomies computer. Hey, it made me a member of the 4GHz club, if only for a short while! LOL!!

Clock On,
theone :}

 

Hey guys,
I downloaded the trial of Everst and ran Orthos, CoreTemp and Windows Task Manager. The thing I'm noticing more and more is that what I set in the bios for voltage is not what is getting reported.
I'm not all that impressed with my power supply so maybe I'm having a problem there. It's a CoolMax 700W unit.

I looked at the temps when I first installed the CPU and at the default speed/voltages they are still the same, about 20°C at idle and 25°C or so at full load. Ambient about 67°F

Here's the screenshot after running Orthos for about 30 minutes. The Vcore is set at 1.575 but getting reported here as 1.52.



I'll have to buy Everest. I've barely looked at it, but it appears excellent. Thanks!


Edit:
I ran it full load at the default settings for a few minutes:

 

fasfrank,
I'd like to point out a couple of things from your Everest screen shot. First, the CPU temp can never be higher than the cores as the CPU temp comes from the heat of the cores! Looking at a temp of 50C for the CPU means that the cores are more like 62-67C. The core temps are usually the most accurate because the sensors are attached directly to the individual cores. Right now my cores are at 37&40C but my CPU is only 29C. My guess with yours is that you've generated too much heat and the expansion has damaged the sensors. With mine it just read temps that were lower than room temp!

Funny you should mention CoolMax! I didn't have the 700 watt but I did have 3 500 watt replacements for it, the last one went in the trash last sunday! I won't use one again. The first one, the fan failed. The replacement for that went up in a beautiful 4th of July series of bangs, sparks and smoke. The afore mentioned last replacement was the one I threw in the trash! Get this, they didn't even want to old ones back! They used to be good PSs and a good company! Now I won't ever buy another one. It's a shame that a good company went to hell like it did! Must have been bought out or something The Thermaltake that replaced it weighed almost twice as much for the same 500 watts. I took the CoolMax apart and it looked impressive at first glance but had poor quality electronics and not near as much heat sink area as it could/should have had! Max load was an actual 330 watts! Newegg knew they were junk because when do they ever not want a defective anything back unless it's incredibly cheap like 80mm fans. Who's going to pay $5.47 to ship back a $5 fan I've owned some cheap power supplies in my life and have RMA'd a few. Not once did any of them tell me to just throw it away! This was a first! It's called cutting your losses! Newegg knew from the start they were junk! Probably couldn't return them because they were more than likely a clearence lot purchase. They do ask for each and every serial # though! It also got me the incredible price on the Thermaltake of $21 with free shipping! Great PS.

Clock On,
theone :}

 

Thermaltake PSUs greatly depend on which model they are, as far as I know. Since it's not actually Thermaltake that make the units, the quality of the unit depends on who produced the insides. In the case of the Toughpower it's Channelwell technology, who make power supplies for Antec if I'm not mistaken. In the case of the Purepower series and before, I'm not sure who it was. Thermaltake PSUs always seemed to be a bit of a "probably best not get one of those" area until the Toughpower arrived, and then they started getting good rep with the other brands. From what I've seen and heard, I will eventually get a Corsair unit, when I can afford to spend that much on just getting a quieter power supply!

 

looks like i fried my core sensor too when i tried to run my cpu on 1.55v o0. lol

 

fasfrank,

yeah, something just isn't right with your readings or sensors. note the higher core temps on mine. the highest I've ever had the vcore is 1.4500. It's at 1.4250 now.



theone,

yes, I know my temps aren't as good as yours, not many ppls are, but my vcore is higher than yours and the ambient temp around my pc is about 78 F with very little air movement around the outside of it.

 

Mort81,
I suspect that the Alandales run cooler because there is less physically inside them. on the 43 & 4400s there is only one cache, where the 63 & 6400s have both but one is disabled. I seem to remember that Gina's ran pretty cool too. There's nothing wrong with the temps you displayed, besides yours is clocked higher than mine. I do run a ceiling fan all the time, which helps too! I just made these!



fasfrank,
As you can see from both Mort81's and mine, something is not at all right with your temps.

Clock On,
theone

 

No picture proof, but at my LAN party (wide open space, cool room) I managed 36C idle and 44C load with the fan speed 100%, then with the fan disabled (not intentionally, oops!) it still ran at around 50C load. GPU was more impressive though, I got it under 40C idle!

 

sammorris,

The day your word isn't good enough for me, shoot me! I would deserve it!!

Clock On,
theone

 

Cool though those temps were, I now have an issue. I had to correct this once at Fragsoc, but it now seems to happen regularly - the contact for the fan on my Freezer 7 Pro seems to keep breaking. I have to go in the case and firmly push on the fan connector and the fan will run again for a few more hours before turning off. Either the connector for the fan, or the connector on the motherboard is presumably bent in some manner. This means I need to have my case fans on high to cool my CPU sufficiently, which is annoying.

 

sammorris.

This is the same MB that you have to re-set the bios to get working every couple of days? Might be something in the bios turning it off! The only intermittent connector problem I've ever had with a MB was with the molex plug for the X800GTO. Every once in a while I would get an error message telling me that there was no voltage to the video card!

Clock On,
theone :}

 

No, I went in there with the PC running and touched it, I had to hold my finger on it, or it would stop whenever i let go of the connector. I gave it a hard prod and it's running now, but there's something there I need to change.

 

sammorris,

More than likely either a bad wire connection at the plug or a bad female pin in the plug. On one side of the plug you will find 4 open slots. Use a very small (Jeweler's) screwdriver and push the flat pin inside the slot and you can remove them. Best way is to pull on each wire. If one is bad you will feel it. If so remove it and solder the wire back on. Spread the pin you pressed to get it out of the plug and put it back in the plastic plug till it locks and you should be good to go!

Clock On,
theone :}