[trents]

Yellow is okay as long as your core temps are not exceeding 55 C. If you are unsure of the impact of the offset you choose, just try it and then check in HWMontor in the CPU vcore
line to see what it did. Then readjust as necessary in bios. Just try things. One small increment in voltage is not likely to damage anything. Be bold! You are too tentative about all this. By the way, 20x is a place you ought to stop and make sure its stable there. 20x would be 4.0 ghz.

[domjam]

yes 20 is were it faild 19.5 was a pass and my vcore was 1.42 max cpu 47 c max and all 6 cores at 34c max this was with the stock voltage that u gave me orginaly to start with. i dident quite understand why it faild at 20 though since the prime window dropt out with only 5 mins remaining of the 20 mins and the temps on hardware moniter were actule lower then the 19.5 ones. i will try with 20 again with incress of volt and report back..little by little im gettin there. thank for ur time and pasients

[domjam]

just got a pass with cpu ratio at 20 and volt 1.389(0.056250)<blue. cpu 49 c,cores 36 c and vcore 1.43 still way within the limit. this is right after 20 mins of prime95 blend test 6 torture test. i can up the ratio more?

[trents]

Good. Just remember that whenever you go into bios and hit the "+" key it will add the smallest increment of voltage, whatever that is. You don't need to get caught up in the math because you can't change the size of the smallest increment. It is what it is and it will be small enough to not present any real danger of frying anything.

Yeah, with core temps at 36 C. (I'm assuming that's "max" and not idle temp) you are fine if you need to add some more vcore.

Please hold at 4.0 for now but give me some screenshots of the three CPU-z tabs and of the entire HWMonitor interface that shows the CPU vcore section all the way down through the core temps section. I want to see max CPU vcore (i.e, CPU voltage) as well as max CPU socket temps and max core temps so run that Prime95 test again at 20x ghz if you need to to capture that data. Thanks. You're doing good!

[domjam]

i have been saving logs of my progress so i wont have to run prime95 for this time.if some stuff is cut out of hwm that u need to view let me no and i wil lrun prime95 and snap a screneshot.otherwise here is the results i got

[trents]

Looking good!

Now, let's consolidate and confirm.

I want you to run the Prime95 blend test for two hours. If you fail then give a tad more vcore and retest for two hours. Passing 20 minutes of Prime is not enough to conclude you are solidly stable but it does mean you are probably close.

[domjam]

ok dam thats long. just out of curiosity i no that the cpu ratio incresses the speed of the cpu but dosent the bus speed to that aswell? u allso said to me continue to run the test for 20 mins till it fails and when it fails to increes the volt and up the ratio till the cpu exeeds the values u gave me..core 55,cpu 65 vcore 1.55 the wall..but i dident get to the wall yet..is it safe to contnue upin stuff till it exeeds the temps and then start runnign prime95 for 2 houres? sorry im stil la bit confused..ut alot more clued up then beffor

[trents]

4.0 ghz is going to be close to the limit of that CPU, no matter how much voltage you throw at it. Since you have good water cooling you might squeeze out a bit more. Since we are close to hitting the wall, all tests now will be two hours long. Start with where you are before you try squeezing any more out of it. Two hours is not long at all! A lot of guys on this forum would say you have to test for 8 hours to confirm stability. In my experience, two hours is enough. If you pass the two hr. test at 4.0 then you can try for more but the tests need to be the two hour ones from here on out.

When you overclock with the CPU multiplier the bus speed remains constant at 200 mhz. You can look at CPU-z (the CPU tab) and see that for yourself.

[domjam]

maybe i dident explain that well
if i have a cpu ratio of 16 and a bus of 231 or around that mark it makes the speed faster then 3.2ghz with a ratio of 16 and a bus of 200.. wat im trying to say is if i have a ratio of 20 but say a bus of 231 or more then 200.dosent that incresse of the bus speed incress the speed of the cpu aswell? so insted of trying to go beyond a ratio of 20. to get more speed,u can increse the bus insted so u dont have to increese voltage??i get the idea that increesing the ratio requires more volt when doing so..im gona run the test tomorow.its late here now.need to ge some sleep

[trents]

Increasing the speed of the bus increases the speed of the CPU as well. Yes it does. But increasing the speed of the bus also increases the speed of the ram . . . and of the HT Link . . . and of the CPUNB, all at the same time so that you are having to juggle all those things at once to keep any of them from becoming the source of instability. And that makes overclocking much more of a challenge, especially for a beginner. All those little separate frequencies I mentioned above are all tuned to the fsb (front side bus) and that's why they all rise in concert with the fsb. You have to keep adjusting each one of them to keep their frequencies within certain limits so that none of them causes instability. When overclocking with the CPU multiplier you don't touch the fsb and all those other buses stay put as well. However, once you get settled on your final CPU overclock speed you can go back and mess with the other things on an individual bases if you want. Some say increasing the CPUNB over stock improves memory performance but others say there is no real world gain.

[domjam]

i have yet to get a pass with the 2 houre test. im in yellow with volt now but no pass yet. last atempt was with a volt of 0.065625 i got blue screen around 40 mins in testing.. trying the next figure witch is 0.068750 will keep u posted

[domjam]

2 hour blend 6 torture tests passed with a cpu ratio of 20 and cpu offset voltage of 0.068750 heres the HWM and 3 cpu-z windows to review. let me no what to do next when u get this.thanks again

[RGone]

You are running the P95 two hour testing with HWMonitor running in the background so that the Mins and Maxs are of the system under full stress and 'then' back to idle after P95 test is over???

I ask because HWMonitor shows max CPU CORE temps of only about 38c under load. Just making sure what I see and others see is what we think it to be.

[domjam]

HWM has been running since i started prime 95. I took the HWM pic just as the 2 hour test finished.(before clicking stop)
Do u need me to post HWM without prime95 running so you can compare the 2?

[RGone]

Originally Posted by domjam

Do u need me to post HWM without prime95 running so you can compare the 2?

NO only trying to make sure what page of the book we were using. It really does not matter much what is happening when n0t stressed.

"trying the next figure witch is 0.068750" << Is what you said in earlier post. Bump it up one more and then go to HT Frequency and set to 205 and then test with P95. See if that passes since 'every' frequency including cpu, ram and CPUNB will be *raised*.

[domjam]

ok well provided i have don everything correctly having HWM running in background wast running prim 95 (blend test 6 torture test.start.leave it for 2 hours and then stop the test having no workers stop or crashes and takin picture of HWM just as it finshes)
then everything you see should be correct

[domjam]

another 2 hour test yes?

[domjam]

forgot to ask is the HT link frequancy the cpu bus frequency? i have HT link speed on my bios but its on auto and it only gos up in 100s

[trents]

No, tbey are not the same. Look at the lower left corner ofthe CPU-z CPU tab and you wl see their frequencies displayed. HT Reference is the bus, not HT Link. HT Reference is the official name for the main system bus but it's called "bus" in CPU-z. Leave the HT Link alone.

[domjam]

ok nevermind i misread.sorry i will try this and report back after 2 hours