FX-6300 Or Phenom II x4 965 BE Or Phenom II x6 1045T

[CAT-THE-FIFTH]

Intel has the advantage of having the resources to spend on cutting edge process technology and more importantly gets it out much earlier than anyone else. AMD uses the best alternatives available and yet they are 18 months to 24 months later. AMD is moving to 28NM next year for Kabini and Richland. AMD may get to 20NM by 2014 I suspect.

[mikeo01]

Think I have a stable oc setting for 3.8Ghz

vcore 1.320
cpu nb 2500mhz @ 1.2500v

I managed to post and run a quick cinebench at 4.05Ghz and got score of 6.89. I will try out proper settings soon, that was with RAM under rated specs. So will loosen the timings and get it up to 1600mhz because that gave better results at 3.8Ghz when I tested it. Didn't run a proper stability test but will do soon.

My wall seems to be 310. Still, I have some headroom with voltages, 1.320 is pretty damn good for 3.8Ghz

I have to be honest, this Phenom is a lot easier to overclock than my FX chip. May of been the chip, but this chip plays ball whereas my FX needed specifics!

[mikeo01]

Update!

Hows this looking?

4.0Ghz (297 x 13.5) @ 1.368 -1.380
CPU NB 2673mhz @ 1.12500
HT Link 2085mhz
RAM 9-9-9-24 794 Mhz

Temps:
CPU 51 - 55c tops
MB 18c (its cold day ) can get up to 23c

Also, in AMD OverDrive it only has the option to set vcore to 1.3, why is that? I don't use software to overclock I do it all from BIOS, but I use AMD OverDrive to double check settings.

I am not considering this OC stable just yet, need some more runs by at the moment it is OK.

Also, I have heard on some forums (been looking around) running 1.375vcore and over for Thuban is dangerous?! This true?

Apparently it shouldn't go above that for 24/7 use?

I have a 4.0Ghz overclock, and I sure can get it stable but what if I do require over 1.375vcore? Is that the same as say running 1.5 on an FX chip (in terms of overvolting and how dangerous it is).

I sure want to keep this chip along the rated voltage line, and definitely not go over to make sure the chip doesn't degrade.

[mikeo01]

Been running Prime for about 2-3 hours now, and lost all dignity, I have lost the world to live

On the up side, seems alright and heres the current settings that'll keep me happy for the next 1-2 years

297fsb 13.5x @ 1.368 (Cpuz and AI suite reported that number for ages now, so seems stable there) LLC on so vcore is exact to the value I input
CPU NB 2673Mhz @ 1.1250
RAM @ 1598Mhz 9-9-9-24 (Eudar RAM rated at 1333mhz)
HT link 2085Mhz

Temp: CPU 52c avg, 56 max
MB: 20 avg, 24c max

I assume my northbridge and VRMs are sweating by now I will get some cooling on that soon.

So that's it, looks like my OC.. until it crashes one day!

Will this chip degrade so much? vcore feels low but the 1055T has a rated of 1.375 according to AMDs detail sheets I am close to that limit

EDIT:
I threw my values into Extreme Outervisions power supply calculator, and it says my chip will only consume about 140W? I heard Phenoms can eat up to 200W?

[CAT-THE-FIFTH]

When I was overclocking I did more stability tests than that. I would run OCCT on its most stressful CPU setting for at least 8 to 12 hours,and then try Prime95 for a similar amount of time. I would also run the OCCT CPU and RAM test for around the same time too. I would also run IBT for a reasonable number of iterations too.

Do it overnight - IBT is a good way to get a good ballpark stable figure,before the other tests as about 10 to 20 iterations will weed out most unstable overclocks. You can then run the other tests overnight if required.

This way,I knew that even if I was doing a gaming session for a few hours,the CPU would most likely to more than stable enough,for quite a few years.

[mikeo01]

Stress testing is good if you are going to be folding, but it misses fluctuations. I usually call it stable if prime is fine for 2-3 hours, then I try out OCCT which puts some load on and off the CPU which kind of tests fluctuating loads, then IBT to double check.

I think 8 - 12 hours each is extreme I think as long as it can hold up for 2-3 hours on high load, and can run daily on light loads you're good to go.

That's my opinion though.

[CAT-THE-FIFTH]

Stress testing is good if you are going to be folding, but it misses fluctuations. I usually call it stable if prime is fine for 2-3 hours, then I try out OCCT which puts some load on and off the CPU which kind of tests fluctuating loads, then IBT to double check.

I think 8 - 12 hours each is extreme

That's my opinion though.

I ran overclocks for years with no issue. I consider 8 hours the minimum level TBH,as I can see myself running something that would constantly stress for the CPU for that time,even if it is a long MP RTS game. Moreover,I can run the stress tests overnight,so it is not really a problem.

[mikeo01]

Suppose it depends what you do, I am not really an extreme user and most games I run are not really a burden on the CPU. Actually apart from gaming the rest is Microsoft office tools, the rest is physical that I don't necessarily do at home

As long as it works for games, and daily basis with no BSODs then its all good

I don't feel comfortable putting that much a stress on the chip for that long period of time. It surely must degrade the chip by a bit.

As long as it doesn't blow up I am happy!

[mikeo01]

OCCT froze after 1 hour! I think it is because I have a low CPU NB voltage.

Could someone confirm what voltages I should be under for 24/7 safety? I don't want to be pushing more than what is safe and rated by AMD.

Thanks

[mikeo01]

Just noticed something, if I leave turbocore on it boosts the chip up to 4.0Ghz. You can see where I am going with this
If I can get my FSB that high so the core clock is 4.0Ghz would the chip boost itself up to 4.7Ghz?

I think I will have to configure offset mode, I highly doubt it could get to 4.7ghz maybe 4.2Ghz though.

Is there a way to control how much voltage is pumped in when turbocore kicks in? AMD OverDrive only allows 1.3v for turbocore which is way too low. Can offset voltage mode regulate this?

[Terbinator]

Just noticed something, if I leave turbocore on it boosts the chip up to 4.0Ghz. You can see where I am going with this
If I can get my FSB that high so the core clock is 4.0Ghz would the chip boost itself up to 4.7Ghz?

I think I will have to configure offset mode, I highly doubt it could get to 4.7ghz maybe 4.2Ghz though.

Is there a way to control how much voltage is pumped in when turbocore kicks in? AMD OverDrive only allows 1.3v for turbocore which is way too low. Can offset voltage mode regulate this?

I think it really depends. If you hit the TDP limit at 4GHz alone then ti won't boost up, i don't think. The whole boosting thing can vary as some boards allow you to set how much exactly the CPU can pull as well as having limits hardwired in - on Intel, at least.

Secondly, keeping the Turbo Speed isn't ideal. You have to set voltage higher than needed for 4GHz to compensate for any boost above 4GHz. Aside from the inherent extra testing required, you're going to be using more Vcore than you need for non-boost clocks.

[mikeo01]

I assumed I could set the offset value higher so that it boosts up? Or won't it work like that?

Yeah my motherboard doesn't allow a specific vcore for turbocore frequencies.

Although wouldn't the chip downclock three cores to allow a higher frequency on the other cores?

I will probably bench it just to see the results, but if it requires way too much vcore I think I'll have to leave it alone. My chip is reasonable at 4Ghz only requiring 1.380vcore, so there's kind of a bit of headroom

EDIT: Usually upping the multiplier (turbocore) doesn't require that much voltage than HTT alone?

[Terbinator]

I assumed I could set the offset value higher so that it boosts up? Or won't it work like that?

I can't say i've ever found messing with Offset that fruitful, even for downclocking/volting, so i can't really say.

Yeah my motherboard doesn't allow a specific vcore for turbocore frequencies.

The TDP limit isn't a specific Vcore for a given multiplier - it's merely jsut how much your CPU is allowed to pull under certain scenarios.

I will probably bench it just to see the results, but if it requires way too much vcore I think I'll have to leave it alone. My chip is reasonable at 4Ghz only requiring 1.380vcore, so there's kind of a bit of headroom

By all means do so, that isn't the major issues, it's more the extending trial and error i mentioned originally.

EDIT: Usually upping the multiplier (turbocore) doesn't require that much voltage than HTT alone?

I'm not massively knowledgeable where AMD overclcoking is concerned - but i do know that generally, clocking the HTT up nets you better results/stability - the unlock multiplier is just there to help you get the best of both worlds, and as you've found out, can net you that extra speed every now and again.

[mikeo01]

The TDP limit isn't a specific Vcore for a given multiplier - it's merely jsut how much your CPU is allowed to pull under certain scenarios.

So it is a limit of the actual chip itself? Also ASUS AI Suite Probe gives me "apparently" accurate power consumption reading. I have no idea how true this is, but the max wattage it ate was only 60W? That's at 1.279vcore 3.3Ghz. That seems a bit low but is these readings anything to go by in terms of TDP limits?

By all means do so, that isn't the major issues, it's more the extending trial and error i mentioned originally.

Is there any specific way to test turbocore? I have found running prime or Cinebench the cores will stick to its rated frequency. So it only really shows when I am not doing anything and have CnQ enabled. Earlier I found a lot of applications corruption and crashes, found out that Turbocore was enabled still.

Do you know in terms of voltage, what is considered "safe"? So that if I did want to play around with turbocore I don't end up reducing its life?

[Terbinator]

So it is a limit of the actual chip itself? Also ASUS AI Suite Probe gives me "apparently" accurate power consumption reading. I have no idea how true this is, but the max wattage it ate was only 60W? That's at 1.279vcore 3.3Ghz. That seems a bit low but is these readings anything to go by in terms of TDP limits?

Yeah.

As for the 60W reading, i wouldn't be too bothered by that given you've got a lower end Phenom.

Is there any specific way to test turbocore?

Running something like Prime95 on 1/2/3/4/5 threads rather than all should do it.

As for Voltage, i believe 1.5v is the max you should bother with, both in terms of chip life and having the necessary cooling to keep it stable. More importantly, the AMD chips have fairly low recommended temperature limits - something like 62c -that would be my biggest concern.

[mikeo01]

That's why I tend to go for 95W chips, a bit of headroom. Is there any software to measure its power consumption? Or is the best to go by? Not necessarily concerned about its consumption just nice to know.

And thanks, I will probably play around with it just to see what kind of performance I may get out of it.

Heat's not so much a concern, once I get a new cooler it should be fine. I have always been a bit weary of voltages because I will probably keep this chip for years, or until it gets sluggish on all new software that'll be out by then.

Thanks for the advise I will keep to that