Post Your Overclock!

tranqueirex said:

Sure thing Mr. Skunksmash, I was enlightened:

I will simulate these two scenarios running a benchmark for each one just to figure out wich is better.

If you would kindly recommends me an effective benchmark software for that purpose I promisse to stop annoying you.

thanks again

lol....... your far from annoying mate

the general bench standards are set by...

3dmark06

3dmark vantage

any of these benchmarks will work well for overall system performance....run them on standard settings, so your able to make a comparison to other machines.

if you wanna just test RAM throughput, i'd suggest SISsandra, then run the RAM bandwidth bench....


:)

Wishmaster said:

looks good to me as well :)

I also agrre w/Skunk.
In the grand scheme of things, MEM bandwidth will not make a sig. diff. other than benchmarks.

I seem to recall reading somewhere that for AMD CPUS, they need to be kept at 50-55C MAX temp at full load ...
now i could be wrong here, can someone please confirm?


Couple other benchmarks i like: (beyond that thats already mentioned)

EVEREST is good. (has a 30 trial) (for MEM bandwidth -- prob. just as well off Sisandra tho))
SuperPi is another for calculations
Cinebench (CPU rendering)
less popular but good tests :)

These are basically limited to CPU testing though where as the ones Skunksmash reccomended are more all round, and generally the accepted "norm" for benchmarking.

ahh yes.....''superPI'' is a great little app, this will quickly show any slight increase in CPU performance.. :)

tranqueirex said:

Wishmaster said:

I seem to recall reading somewhere that for AMD CPUS, they need to be kept at 50-55C MAX temp at full load ...

Thanks for the warning Mr. Wishmaster, I am aware of that and that's exactly the limit I've imposed to myself to overclock in summer (worst case scenario). I've read the Phenoms became unstable when >60ºC and can be burned when >70º.

I am running SuperPi and 3dMark06 and now I am able to compare my stats with another overclockers. thanks for you and Mr. Skunksmash for the recommendations.

Unfortunately I am unable to go over 3.6GHz (1.475V) with my asus m4a78. 3.7 GHz is unstable no matter the HT / FSB / VCore / VNNB I choose. It is probably a limitation imposed by the Motherboard wich is too way limited in Vcore settings and has only 125W of TDP.

It's not a problem at all, running in 3.5GHz 1.425V (WEI 7.1) its a 7% of effective increase in CPU benchmark and that helps a lot to make my system follow the GPU (WEI 7.4) - anyway I just have a modest coolermaster TX-3 to breeze the CPU and 3.5GHz is a sweet spot for that humble setup

I know that its a crappy OC compared to your Intel Icores' OCs but it can serve as reference for another Phenoms + Asus owners, I mean if you want to go further with Phenoms II do NOT go ASUS M4A78 standard. you will be limited to 3.6GHz on phenom 720 because Vcore is 1.475 topped.

Here is a SS for your Knowledge base (stock value is 3463):

Thanks mr. skunksmash and wishmaster for your leads.

that's pretty good mate, i scored less with my 8800GT (i scored 12,008 with an E2180 @ 3.8ghz) & the 8800 outperforms the 4850 by quite a bit.. so your CPU is carrying the score pretty well..

you cant compare AMD to Intel in terms of how far they clock b/c intel have always been pretty easy to crank past stock...the new phenoms are very good clockers though

its a shame your board has run out of VCORE adjustment, you could of gone further.

still im sure its plenty fast enough as it is

skunksmash said:

not too great with the AMD OC's you may have to wait for someone running a phenom for a definitive answer, but that looks pretty good....:)

dont worry about your idle temps they're not that important.....& a 50c load out temp is VERY GOOD!! :), maybe you can squeeze it for 3.8ghz

if your ''prime stable'' for 8hrs+ then its safe to say the OC is solid.

Skunksmash I disagree. Prime 95 is not a adequte test by itself. I have seen many times system that run prime 95 for hours yet still will fail tougher tests. Try it yourself rume prime 95 at full blast and the open another program. System performance is hardly impacted. Then run Linpac and try it!

The measure is not "good enough" its all or nothing. If your PC can not run with ALL subsystems at 100 percent with out silent corruption or heat failure then it is not stable.

There are other CPU tests Linpac for one, that can blue screen Prime 95 Stable PC's in minuets.

In short Prime 95 for a OC test is not enough because:

1. It simply does not stress CPUS hard enough. Other test includeing Linpac will produce temps 10-15c higher.
2. It does not stress any other subsystem at the same time. Video, Disk I/O, audio, are not impacted.
3. Due to the limited nature of Prime 95 testing, max system temps are not reached.
4. Power supply is not test at maximum load.

I have battle tested my OC at 100 percent load of all of all system drawing nearly 730 watts at the wall plug for hours! To do this I had to limit my OC to 3.8 Running only that weak sister of a test Prime 95 I could OC to 4.2 and still control heat.

The point of all this is that OC of PC's is risky in terms of possible system corruption and data loss. Great back up and far more testing than most people do is justified.

Lens Pirate said:

skunksmash said:

not too great with the AMD OC's you may have to wait for someone running a phenom for a definitive answer, but that looks pretty good....:)

dont worry about your idle temps they're not that important.....& a 50c load out temp is VERY GOOD!! :), maybe you can squeeze it for 3.8ghz

if your ''prime stable'' for 8hrs+ then its safe to say the OC is solid.

Skunksmash I disagree. Prime 95 is not a adequte test by itself. I have seen many times system that run prime 95 for hours yet still will fail tougher tests. Try it yourself rume prime 95 at full blast and the open another program. System performance is hardly impacted. Then run Linpac and try it!

The measure is not "good enough" its all or nothing. If your PC can not run with ALL subsystems at 100 percent with out silent corruption or heat failure then it is not stable.

There are other CPU tests Linpac for one, that can blue screen Prime 95 Stable PC's in minuets.

In short Prime 95 for a OC test is not enough because:

1. It simply does not stress CPUS hard enough. Other test includeing Linpac will produce temps 10-15c higher.

crap...... its been used & is STILL being used by OCers worldwide


2. It does not stress any other subsystem at the same time. Video, Disk I/O, audio, are not impacted.

lol..... when testing for my max OC im not looking to stress these sub-systems anyway, why the hell would i stress test a soundcard..??, & my GPU is completely separate from these tests

3. Due to the limited nature of Prime 95 testing, max system temps are not reached.

prime is plenty good enough for the average user, 99% of OCers upon reaching max ''stable'' OC wont even be stressing the system anywhere near where these apps take the components..... ive yet to play a game or run an app which will utilize my CPU @ 100% load for the duration of use.

4. Power supply is not test at maximum load.

well...... if your PSU is a concern while OCing, then you really shouldn't be doing it, considering it only adds a few watts...

I have battle tested my OC at 100 percent load of all of all system drawing nearly 730 watts at the wall plug for hours! To do this I had to limit my OC to 3.8 Running only that weak sister of a test Prime 95 I could OC to 4.2 and still control heat.

i admire your thorough testing acumen...... but its really not necessary, if there's a problem with the clocks the machine will soon let you know.... it wont BLOW UP!!

The point of all this is that OC of PC's is risky in terms of possible system corruption and data loss. Great back up and far more testing than most people do is justified.

there are many reasons you could end up with data corruption, & yes OCing is one of them....... but in 5 years of OCing ive yet to loose / corrupt Data on my HDD

dont get me wrong, some of what you've said has merit....but its a little off for you to completely disregard ''prime'' as a viable app, its a tried & tested method.... proven to determine ALL OC's stability.

i personally dont use prime, i use Orthos...... this app is UBER intensive on ''priority 8 / blend & stress''

IMO, all in all....taking your rig to the brink of destruction, is not necessary for the majority of clockers....

mooneyham said:

Got my 720BEx3 at 3.6ghz 1.55volts 52hours prime stable, 30hours linpack stable with my xigmatek s1284 w/scythe ultra kaze. It gets about 36c load too . i can push it to 4.1ghz stable with 1.65volts but i see not enough performance in games to go that high.



on my video card(ati 4870) i can push it 832mhz on core and 1100mhz on memory and it gets me 4 more fps in crysis. it gets about 57c load on core and vrm's get only to 70c. its 3 hours furmark stable and countless hours of gaming stable.

edit: forgot about my memory and cpu-nb. i got my memory from 1066mhz at 7-7-7-25 to 1600mhz 7-7-7-20 with 1.74volts. with my cpu-nb i got 2764mhz with 1.5volts on it.

Mr. Mooneyham, thanks for the update. So far I can see setting 1.49V vcore to get 3.6 is OK in my case. I was wondering if that Vcore wasn't too much... but you had reached 1.65V? oh boy, you're really brave!

you got really impressive NB and temperature values. I will look for your cooler review.
I am supposing you can get these high NB values because you're choose DDR3 instead DDR2, is that correct?

BTW don't you have a 3dmark06 score of your 3.6GHz setup to post us? or your WEI scores?

Do you think its ok to get 55ºC fully load on the hottest ambient temp - 32C?
Thanks a lot!