Question about determining 24/7 Temp

[Mangoose]

Hey guys,

Newbie on this forum here. Somewhat novice at OC... Been doing it for a while, mostly just tweaking multiplier + vcore and keeping at below technical max temperature. Working on a FX6300 right now and you guys seem very experienced with Bulldozer/Piledriver. Trying to do something different now though.

Right now I'm trying to find a voltage that gives me a safe 24/7 temperature.. because (A) I'm in FL and (B) my family can't tolerate anything under ~81 degrees Fahrenheit in the room (and my window AC unit is ****ing loud with no medium settings so I hate keeping it on). I'm using stock 2.5 GHZ for now, but after I find a low enough undervolt I'm going to slowly push multiplier to see if I can get stable with the voltage/temperature I want. I know, this is weird... but I'm experimenting here lol.

Anyways, I have a few questions about benching a safe 24/7 temperature:

1) From a bunch of research I've heard that 60C is the max safe temp. As in, AMD says 70C is the technical max, but 60C makes for a good safety gap before reaching 70C. I've also read that 55C is a good "sweet spot."

My question here is, what kinda of benchmarking am I looking at to "match" 55C or 60C? Is, say, running 30 minutes of P95 at Blend, and taking the max Temp of that, sufficient for seeing how my own temp compares to 55-60C? Or should I run the test for 24 hours and then see what the max temp is?

1b) Just to clarify, it seems 60C is the subjective safe 24/7 based on AMD's spec of 70C. I was going through http://www.overclockers.com/forums/...ormance-Scaling-Charts-max-OCs)LN2-Results-co which suggested 45C. Seems like AMD's spec for the 8 core is 61C, so 45C is a 15C threshold. Am I correct in assuming that with 70C on the FX 6 core, the same 15 threshold applies, as in ~70-15 = ~55C?

2) I will be keeping the computer on 24/7, but I will only intermittently run at full load. In other words should I really be careful at keeping under 55C or does intermittent full load less require the "hard" limit of 55-60C?

3) I'm thinking of keeping it around or under 50C because it's cold season right now, and when summer arrives in Florida ambient temps are going to rocket up a bit. Thus I'm going for ~50C bench so that when the ambient skyrockets I'll hopefully get no higher than 55C. Good idea?

Thanks everyone!

 

[RGone]

Straight up answer 'mangoose'? Never heard of the pablum you seem to have read or seen somewhere. We suggest as seen in HWMonitor Free version WHILE under the load of P95 Blend mode that the temp usually labled as cpu temp aKa Socket temp never exceed 72c ish and that the temp seen as Package temp in HWMonitor Free, not exceed 62c and that again while under the stress load generated by P95 in blend mode. This does not vary. Has not overall varied in nearly 3.3 years of use of the AMD FX series released in fall of 2011.

If you want to experiment with some lesser temp variant so that you can run a computer in a room that is hot for most of us at 81F, then adjust for a speed and temp that lets the cpu run within the temp range as mentioned above while under load. Job done.
RGone...ster.

 

[Mangoose]

Straight up answer 'mangoose'? Never heard of the pablum you seem to have read or seen somewhere. We suggest as seen in HWMonitor Free version WHILE under the load of P95 Blend mode that the temp usually labled as cpu temp aKa Socket temp never exceed 72c ish and that the temp seen as Package temp in HWMonitor Free, not exceed 62c and that again while under the stress load generated by P95 in blend mode. This does not vary. Has not overall varied in nearly 3.3 years of use of the AMD FX series released in fall of 2011.

If you want to experiment with some lesser temp variant so that you can run a computer in a room that is hot for most of us at 81F, then adjust for a speed and temp that lets the cpu run within the temp range as mentioned above while under load. Job done.
RGone...ster.

Appreciate the quick and great response RGone. I guess I wanted to mainly clarify should I be looking at the temp in HWMonitor (definitely using that and P95 already) after 30 minutes - 2 hours, or should I bench for 24 hours to get a very accurate result? I probably over-described the situation, but this was my main question.

Oh yeah, I guess the other main thing I wanted to ask is, is 62C (under P95 blend) good for 24/7 use or should I be looking for a smaller temp?

 

[RGone]

We suggest 2 hours of P95 Blend mode as the 'standard' for 24/7 stability. I generally do 'more' time for my ownself, but friends that are not new and use their clocked up rigs for daily use in business often go only 2 hours. So that choice is up to you.

Normally the 62c is not a constant temp for the entire time the computer is turned on. I now use Offset Vcore mode and have my system idle at 1640Mhz (0.996V) and yet when I begin video editting it jumps to 4.6Ghz (1.43V) and off to do real work. My actual temp doing full on video editting for even hours at a time is about 56c on the package temp. But remember I do not video edit 24/7, in fact I am typing this at 1.64Ghz at 0.996Vcore.
RGone...ster.