VERY successful D700 OC

[Quaggoth]

Duron 700 @ 900 at 113F after playing F1-2000 for about 2 hours (1 Full Race at Suzuka with an average Laptime of about 1:42.1xx ) I think it was at 105 before I started playing. It posts and runs fine at 1000 for about 10-15 Min's. Haven't tried anything higher. BTW, this OC probably sounds pretty normal, BUT, I am running a stock HSF and only 1.675V on a KT7-RAID. I think at 1.85 I should be able to get about 1100 maybe, but I need a SupaOrb or something. Also, I am using Radio Shack Special thermal compound.:|:Q:disgust::frown:


Hmmm, sorry for the emotes, I just wanted to see all of them.

 

[MaJik]

Nice job
know the week?

 

[office boy]

Well that?s pretty good, but I'd have to say that in my experience the orbs and their fancy shmancy cooling ways work great for P3's and celeron2's but can't hold up on the duron/T-birds. (well the ArctiCooler can but good luck finding one.

However the FOP's and Alphas are better IMO just because they have so much more mass to help soak up the heat.

Yeah their a bit more expensive, but some might say you get what you pay for. (although AMD does a pretty good job of proving that wrong)

 

[IaPuP]

Radio shack thermal compound isn't too great assuming it's their silicon based stuff.

Has a thermal resistivity of about .6 W/mK

Aluminum based material is about 2.0 W/mK and Arctic Silver is about 5 W/mK

Eric

 

[Quaggoth]

Thanks. I am going to get some Arctic Silver, and I am still in the decision making process about what cooler to get. I like the LOOK of the Supa Orb, and my case DOES spend a good chunk of it's time open..... Anyway, I don't really care about the Money involved (Well, To a certain extent), so I may get an Alpha.

Majik - How do I tell what week it is? I know it's one of the numbers on probably the Bottom left corner of the Processor, but which one?

Office Boy - I just read an article about the Arcticooler, I think I may buy one of those, and for right now, I think 900 will "get by" until I can find one

IaPuP - I am DEFINATELY getting some Arctic Silver.

Heck, Even if thing only runs at 1000 stable after I get it cooled down..... I might just be happy with that. I mean KT7-RAID for 149, Duron 700 for 110 means 1Ghz for 275 or so after tax and shipping

 

[buzzby]

If you want some good cooling then get your self a

globalwin FOP38


its loud but it works very well

 

[ingenue007]

Quaggot - where did you get yours from? i bought ine from mwave about 3 weeks ago (always forgot to look at the week). mine runs at 1046 at 1.675 volts too. mine can run at 1100 at 1.85 volts, but that's too high for me.

you should get the TAISOL HS (dropped mine 11 C over stock HS) with arctic silver.

 

[buzzby]

What voltage r u runing and is there a website with a list of the best voltages 4 different settings

 

[DaddyG]

I gotta agree on the Radio Smack grease, it dries out faster than a keg on a Friday night. I am not at all impressed with Thermaltakes efforts in Socket A, none of the ORBs impress me. They look kewl but thats about all. The new Taisol is one kick-ass hsf. Made from Forged instaed of cast Aluminum, for higher density AND it has the correct Socket A clip to engage all three lugs on the socket.

 

[htown]

I total agree, so what is your reallife performance on your setup.

 

[Dan]

I've got to brag a little:

Duron 700 @ 1083MHz, 1.85V, 100F.

 

[Nevin]

Dan,

Unfortunately, your 100F temperature is probably more a testament to the inherent inaccuracies of trying to measure a CPU's temperature with an external thermistor than it is to the efficiency of your CPU cooling.

From some of your other posts I gather that you are using a Cooler Master DP5-6H11 heatsink. This heatsink uses a 21cfm fan and as a thermal resistance between .45 and .55 deg C/W.

Now we just do some math. Since you stated your temperature in Fahrenheit, I will convert all the temperatures to Fahrenheit for clarity.

Power out put of your CPU @ 1083MHz and 1.85 volts.... 65 watts.

Thermal resistance of heatsink assuming a perfect thermal interface... ~.50 deg C/W = .93 deg F/W

65 watts x .93 deg F/W = 60 degree CPU temperature over ambient. (For heatsink testing purposes, ambient temperature is defined as the air temperature 1 inch above the heatsink fan.)

So for your temperature to actually be 100F with your cooling setup, the air temperature inside you case, one inch above the heatsink fan would have to be 40F or below.

Dan, I'm not trying to pick on you. I just want to illustrate that the temperatures supplied by motherboards and other external measurement products for AMD processors are immensely inaccurate. From an engineers point of view, trying to judge the actual internal temperature of a CPU using a thermistor touching the bottom of a CPU or the side of a CPU core is as futile as trying to judge the temperature inside a house by taping a thermometer somewhere on the outside of the house.

In actuality, the common measurement temperature points for AMD processors are located in secondary heat paths between resistive junctions. Here is a link to a diagram that shows this problem in detail:
Why many temperature measurements are wrong diagram

If you have a motherboard that uses a thermistor under the CPU to measure the processor temperature, here is an interesting experiment. Remove the heatsink and CPU and put a bit of thermal grease on the end of the thermistor. This will allow it to make better thermal contact with the bottom of the CPU. Replace the CPU and heatsink and check your temperatures, they will be significantly higher than they were before.

Is the CPU actually running much hotter because you put a small dab of thermal grease on the thermistor measuring the temperature of the bottom of the CPU?

Of course not, you have changed the resistance in the secondary heatpath to the thermistor and it is probably actually giving you a more accurate representation of the actual CPU internal temperature.

Nevin

 

[IaPuP]

Most newer motherboards BIOS have a compensation for this.

Asus did some experiements- I read the whitepapers- and estimated that their temperature sensor on the A7V reports an average of (I think) 10.4 deg C less than actual core temperature. Their newest BIOS (and the Abit KT7) adds this 10 deg C to the "detected" value to report a more precise temperature.

I would still say it is incorrect +/-5 deg C or so.

Eric