Phenom II Overclocking Guide - Page 3 - Overclock.net

wuttz · 06-22-09

Quote:

Originally Posted by thlnk3r

Wuttz, hey great minds think alike

thank you sir, but i am not so bold to say my mind is as great as yours!

jimibgood,

**thlnk3r** · 06-22-09

Quote:

Originally Posted by wuttz

thank you sir, but i am not so bold to say my mind is as great as yours!

Wuttz, we're all the same here

Question about jimibgood's comment, if it's only a 9% increase then how come most saw huge improvements going from a Phenom I to a Phenom II? Honestly I can't speak from experience because I have yet to upgrade to these new bad boys but really...9%? That doesn't seem very much...

logan · 06-22-09

Quote:

Originally Posted by jimibgood

Logan, you made a comment that the Phenom II's are much faster than the original Phenoms. They are only 9% faster core for core.

The statement will stand, but because I am talking about in an overall sense, rather than at identical clock speeds. I understand that it is not a massive improvement in terms of clock for clock performance, but it IS a massive gain in clock speed.

In other words, a PhII will be much faster than a PhI, but MOST of the extra performance comes from the fact that most of the will have about 500MHz clock speed advantage (stock or OC'ed).

Hopefully that clears things up... and for the record, 9% is still a decent improvement for what is basically a refinement of the same architecture.

Thanks guys!

Slappa · 06-22-09

Quote:

Originally Posted by logan

Now, to be considered fully stable, you need to run a program like Orthos, OCCT or prime95 for at least 6-12 hours, but while working with your OC, you don't need to run it near that long. You're just trying to find out if its even close to being stable at the speed it is at right now. In general, if I pass one of the above tests for more than 5 minutes, I'll reboot and continue pushing the OC. Run the longer tests when you think you're where you want to end up, and if it fails before those 6-12 hours, then lower the OC, and try running it again. Otherwise, you're risking running an unstable system for every day, 24/7 usage. Trust me, you do not want to do that, its not worth what it costs for those few additional MHz.

If you do that and the OC is still unstable, the chances are just that you need more voltage (usually marked as the Vcore in the BIOS). While what increasing the voltage does is actually quite technical, its effects are essentially to help stabilize an unstable processor, but it does this at the price of increased heat, and higher power consumption. In general, there is a point of diminishing return for increasing voltage for higher clock speed. You should be able to notice this as it will take much larger increases in voltage for the same increase in clock speed. Regardless, increase the voltage a little (I generally do one increment at a time, as you don't want to increase the voltage more than necessary) and press on. Keep in mind that you should keep an eye on the temps, and that we usually strive to stay under about 55c for the loaded temps (the temps while you're doing the stress testing).

Good guide. A few things though, as I have done extensive testing with these processors.

The only valid stress testing program is Prime 95. After about 2 hours, it starts to stress the NB/IMC. OCCT does not do this. It is essential to run P95 for at least 3 hours.

Also, voltage increasing does not always help. Especially with AM3 chips (710, 720BE, 955BE). They actually dislike more voltage a lot of the time. Read about it here and here.

logan · 06-22-09

Quote:

Originally Posted by Slappa

Good guide. A few things though, as I have done extensive testing with these processors.

The only valid stress testing program is Prime 95. After about 2 hours, it starts to stress the NB/IMC. OCCT does not do this. It is essential to run P95 for at least 3 hours.

Also, voltage increasing does not always help. Especially with AM3 chips (710, 720BE, 955BE). They actually dislike more voltage a lot of the time. Read about it here and here.

Thanks for that slappa. I actually noticed your guide a few weeks ago after I had started working on mine. You do good work

.

I am fully aware of what electron migration is, but always thought about it more in the terms of causing potential harm over time (you know, causing the CPU to require more and more energy to remain stable and the like). I thought electron migration eventually wore down the transistors, and it seems like no matter how the transistor is made, the effect should be similar (meaning intel vs AMD).

And I definitely understand the fact that there is a diminishing return on the application of more voltage to these processors (especially with higher temps). However, I thought the really high LN2 OC's were with done with crazy vcore (I seem to remember 1.6-1.9v for some reason). So obviously they are requiring more voltage for those OC's, which is what makes me wonder how that fits in with this concept of to much voltage. And it also seems that there are many OCNers with 1.5-1.55v for their high OC's

However, if you could explain those few things for me, that would be great.
Just please keep in mind that I am NOT attacking you, I just would like additional clarification to make sure that I can make the guide useful for the largest group of people that I can.

Would you say that in general that my statement is correct about voltage (particularly on air cooling)? I ask because like I said, I don't proclaim to be a guru, and would like to have the guide be correct. I am more than willing to add a qualifier in there, but the qualifier should be in reference to less common event.

Essentially, if I added something like this to the guide under that part.

*Testing has shown that at times additional voltage may actually have a negative effect on stability (particularly with more extreme cooling). If you have tried everything else listed here, you may actually try LOWERING your vcore, and see if that will help. It is for reasons like this that we recommend increasing voltage in small increments, rather than in large jumps.

Thanks for the input guys, this is great.

Slappa · 06-22-09

Quote:

Originally Posted by logan

Thanks for that slappa. I actually noticed your guide a few weeks ago after I had started working on mine. You do good work

.

I am fully aware of what electron migration is, but always thought about it more in the terms of causing potential harm over time (you know, causing the CPU to require more and more energy to remain stable and the like). I thought electron migration eventually wore down the transistors, and it seems like no matter how the transistor is made, the effect should be similar (meaning intel vs AMD).

And I definitely understand the fact that there is a diminishing return on the application of more voltage to these processors (especially with higher temps). However, I thought the really high LN2 OC's were with done with crazy vcore (I seem to remember 1.6-1.9v for some reason). So obviously they are requiring more voltage for those OC's, which is what makes me wonder how that fits in with this concept of to much voltage. And it also seems that there are many OCNers with 1.5-1.55v for their high OC's

However, if you could explain those few things for me, that would be great.
Just please keep in mind that I am NOT attacking you, I just would like additional clarification to make sure that I can make the guide useful for the largest group of people that I can.

Would you say that in general that my statement is correct about voltage (particularly on air cooling)? I ask because like I said, I don't proclaim to be a guru, and would like to have the guide be correct. I am more than willing to add a qualifier in there, but the qualifier should be in reference to less common event.

Essentially, if I added something like this to the guide under that part.

*Testing has shown that at times additional voltage may actually have a negative effect on stability (particularly with more extreme cooling). If you have tried everything else listed here, you may actually try LOWERING your vcore, and see if that will help. It is for reasons like this that we recommend increasing voltage in small increments, rather than in large jumps.

Thanks for the input guys, this is great.

So far, there has been no significant cases of degradation from electron migration in Phenom II's. I imagine it does do some damage, but to a very little extent. AMD's process is different than Intel's which is why it is not damaged as much.

Actually, with the AM3 Phenom II's under LN2, they can sustain more voltage only because of the cool temperature, which in turn gains some extra stability. So you are partially right, but those voltages only help when temperature is not a worry. For example, even on air, you couldn't boot into windows with any quads at 1.7Vcore EVEN IF temperatures are in the 20-30 range.

On air, my Phenom II 955 is not stable with any clock with more than 1.472Vcore. If it could sustain more, I may be able to get a higher overclock, but it depends on the leakage or migration of your chip.

Your air cooling statement is generally correct until users hit the voltage tolerance wall of their particular chip. For example, I am at 3.84GHz at 1.472Vcore, but increasing that to 1.5Vcore only makes my stability worse.

logan · 06-22-09

Quote:

Originally Posted by Slappa

So far, there has been no significant cases of degradation from electron migration in Phenom II's. I imagine it does do some damage, but to a very little extent. AMD's process is different than Intel's which is why it is not damaged as much.

Actually, with the AM3 Phenom II's under LN2, they can sustain more voltage only because of the cool temperature, which in turn gains some extra stability. So you are partially right, but those voltages only help when temperature is not a worry. For example, even on air, you couldn't boot into windows with any quads at 1.7Vcore EVEN IF temperatures are in the 20-30 range.

On air, my Phenom II 955 is not stable with any clock with more than 1.472Vcore. If it could sustain more, I may be able to get a higher overclock, but it depends on the leakage or migration of your chip.

Your air cooling statement is generally correct until users hit the voltage tolerance wall of their particular chip. For example, I am at 3.84GHz at 1.472Vcore, but increasing that to 1.5Vcore only makes my stability worse.

So they're able to use higher vcore simply because the temps are low enough? Thats how I read it anyway, which is generally how I've understood it in the past... and yea... I really doubt that I could boot my 720 BE at 1.7v even if I left the window open all night in the winter

.

Cool, tomorrow I'll add some more about temps in there, and throw that qualifier in for good measure.

Thanks, and +rep

Mikecdm · 06-22-09

Thanks for the guide logan. The extra info explaining things has been very useful in determining why my clocks aren't stable at times. I'd like to thank Slappa as well for clarifying the temp/volt misconception.

I wondered why earlier today I was able to reach certain clock, but later in the day when my ambient shot up, the computer would crash seconds into P95. Extra vcore wouldn't help either.

gerikoh · 06-22-09

hi guys. here's an ultimate clock for clock comparison, deneb vs agena including kentsfield, yorkfield and bloomfield, all at 3ghz.
http://translate.google.com/translat...istory_state0=

i'm thinking 10%.

hey logan, you might want to check this out.
http://techreaction.net/forums/showthread.php?t=367

hope this helps with your guide

Slappa · 06-22-09

Quote:

Originally Posted by logan

So they're able to use higher vcore simply because the temps are low enough? Thats how I read it anyway, which is generally how I've understood it in the past... and yea... I really doubt that I could boot my 720 BE at 1.7v even if I left the window open all night in the winter

.

Cool, tomorrow I'll add some more about temps in there, and throw that qualifier in for good measure.

Thanks, and +rep

No problem, glad i could clarify.

		Overclock.net - Overclocking.net > AMD > AMD - General
Phenom II Overclocking Guide

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)