Request for Help 4.6 on FX8350

[Priller]

Pretty good for 4.62.

Also run the Benchmark a few times in a row. I usually get the first score, run it again right after it finishes then the second score is higher. I've seen 30 point differences between 1st and 2nd runs.

 

[twocents]

Yep, got an extra 12 points by running a few times in a row. Thanks for all the help. Do you think running chip at this vcore 24/7is ok for chip or should I only really use it when needed?

 

[Priller]

I usually do the same thing. Then make a change then do the tests again to see the difference the change I made actually did and not anything program related.

Now that your multi for HT link and NB Freq are at 11 you can make one more step up and change both of them to x12.

Some may run at x13 but x12 is the limiting factor that I have found.

So make one more change and increase both to x12 and run Cinebench again to see what happens.

 

[twocents]

OK so the NB and HT are both now at 2577. Ran the test 4 times in a row. First time was lower, not unexpected but even after the 4th time the score is 4 points lower than previously. Am I doing this right?

 

[Priller]

It's possible you have something else running you didnt have running before.

It's also possible that you aren't using that much bandwidth across the northbridge due to the ram being slower. I would suspect to see an increase with that change if your ram was running faster. You saw in my test that speed picked up points but my ram is a tad over 1900 Mhz.

 

[Johan45]

OK so the NB and HT are both now at 2577. Ran the test 4 times in a row. First time was lower, not unexpected but even after the 4th time the score is 4 points lower than previously. Am I doing this right?

You may have the CPU_NB a bit low which can affect that. Try setting it to 1.25v see if you get an improvement.

 

[Priller]

You may have the CPU_NB a bit low which can affect that. Try setting it to 1.25v see if you get an improvement.

I always like your approach.

I always tell people to take it one step at a time. You just smash on the voltage and let it ride. lol

 

[Johan45]

I always like your approach.

I always tell people to take it one step at a time. You just smash on the voltage and let it ride. lol

I don't know if I'd call that smashing. Smashing is what I do to my stuff
That's just something I have observed with the NB when I was doing some memory testing. If the volts are a bit low it can really affect the performance. Depending on his CPU 1.25 to 1.28v would be what I would consider "normal for a frequency of 2600.

 

[Priller]

I don't know if I'd call that smashing. Smashing is what I do to my stuff
That's just something I have observed with the NB when I was doing some memory testing. If the volts are a bit low it can really affect the performance. Depending on his CPU 1.25 to 1.28v would be what I would consider "normal for a frequency of 2600.

I know. Just givin ya some guff.

If everything goes well then I will be doing some smashing Johan style in a few hours.

 

[Johan45]

I know. Just givin ya some guff.

If everything goes well then I will be doing some smashing Johan style in a few hours.

Oh it's always fun making them go fast. Sitting on the edge of your seat wondering if it's going to crash this time or blow up.

 

[Priller]

Oh it's always fun making them go fast. Sitting on the edge of your seat wondering if it's going to crash this time or blow up.

My board will do 1.536 volts steady but gets too hot. So far 1.520 volts got me 5.170 Ghz. I think if I get the temps just a tad lower with that bump in voltage I have 5.2Ghz in the bag. Trying to beat my personal R15 record.

What was/in your record in Cinebench R15 on water?

 

[twocents]

Hi Johan45, Thanks. I just checked and the CPU-NB is already at 1.25v manual. must have set it to that before. I cant bump the ram anymore otherwise it would be at over 1700 and is only rated for 1600...Anyway, I ran it again 4 more times. First time got an improvement to 739. Thought that was it but subsequent goes averaged 737 to 735. Ended up at 735. TBH, I didn't think I would get this far! Do you guys think that, now I know the vcore that will run P95 Blend stable for 2 hours (1.4125v), I would get any more out of it by leaving ram at 1600, fsb at 200 and just upping cpu muti to 23 as an experiment? Thanks again.

 

[RGone]

twocents, all of this overclocking stuff is trial and error. You get an idea for "your" system; try it and see if it works better or worse. Many of the things some of us do everyday were only "found" after trying something and the result worked for us and others came to try something and it worked for them. BUT it did not work for someone else. Noone but you will know what your particular group of parts will do. You have the keyboard and power switch.
RGone...

 

[Johan45]

Here you go 887, I blocked out my voltage for your own protection.

 

[twocents]

OK. CPU Multi at 23.0, a cheeky 201 on the FSB for a max Cinebench Score of 744. HT and Northbridge equated at 2400 which is 12*FSB. Ramfreq at full 1600 with fbs:ram at 1:4.

In the words of Scotty, "I'm giving her all she's got Captain!"

 

[Johan45]

Mice you're working your way up. The points don't climb very fast on this bench so that's a good gain.

 

[Mr.Scott]

In the words of Scotty, "I'm giving her all she's got Captain!"

You can send my royalty check later.

 

[twocents]

...she wont take much more of this captain! lol

 

[papadan]

Hi, a quick question. Why is twocents processor code name showing Zambezi ? I thought all 8350s were Visheras.

 

[RGone]

Hi, a quick question. Why is twocents processor code name showing Zambezi ? I thought all 8350s were Visheras.

Old version of CPUz most likely.
RGone...