amd 8150 stuck at 4.6

[P8ntballjunky06]

i can get my 8150 4.6. but if i go any faster it gets stuck in a restart when booting.

i am at work but i think my settings are
20.5 maltiplyers
250
1.4v


any help would be great i would like to get to 5.0

 

[RGone]

1.4v = n0r enough Vcore. There is a BIG voltage increase to cpu needed, just about where you are now.

 

[P8ntballjunky06]

well i uped the volts and got it to 4.95 and. 5.0 was not stable

 

[trents]

well i uped the volts and got it to 4.95 and. 5.0 was not stable

I hope you didn't really mean "4.95" volts. Not only would it not be stable it would be dead.

Have you done any temp monitoring under full load stress (like Prime95 blend for 20 minutes) at what was your last "stable" configuration? Sure, looks like you have a nice custom water loop but don't assume anything. Have HWMonitor open on the desktop and run Prime95 blend for 20 minutes. Then post back with an attached pic of HWMonitor captured immediately after the stress test so we can see what your max CPU, package and board temps were like as well as CPU vcore under load. Do you know how to attach a pic with the built in forum tool?

Also, many FX overclockers are discovering that one of the auxiliary CPU voltages (CPU VDA? CPU PLL?) needs to be increased to get a rubust overclock with those jobbers.

Are you running your ram at the frequency rating of 2133? I'm not sure the FX BD IMC will handle that well, especially a full 16gb of it. Most folks are using 1866.

 

[P8ntballjunky06]

temps at 4.95 are 40-45c prime95 1 hour

when i try 5.0 i start prime95 and after 10min it bsod

 

[P8ntballjunky06]

yes my ram is gskill sniper 2133 ram

 

[trents]

Sorry, I thought you were saying you jacked the CPU voltage up to 4.95 back in post #3 but I misread it and I see you were referring to the CPU frequency.

Doesn't matter what frequency your ram is rated for. If that frequency exceeds the ability of the CPU Integrated Memory Controller (IMC) to work with it in a stable way then your ram might as well be rated for 4266 mhz and it would do you no good.

 

[P8ntballjunky06]

ok when i get home from work i will adjust the ram speed. to 1866

 

[P8ntballjunky06]

cvolts is set to 1.5 not 4.9 lol

 

[trents]

If your system is stable while running the ram at 2133 I'm not necessarily recommending you lower it but you might try the 1866 to see if it allows you a little higher overclock. People assume the higher the frequency of the ram the better the performance they will get but it seldom makes much real world difference beyond a certain point because the CPU may not be able to take advantage of the extra bandwidth of the high frequency ram.

 

[Anonaru]

If your system is stable while running the ram at 2133 I'm not necessarily recommending you lower it but you might try the 1866 to see if it allows you a little higher overclock. People assume the higher the frequency of the ram the better the performance they will get but it seldom makes much real world difference beyond a certain point because the CPU may not be able to take advantage of the extra bandwidth of the high frequency ram.

True..

BD chips really gain nothing out of going into the 4.8-5.0GHz realm, right? Isn't this just undue stress on the chip?

 

[trents]

True..

BD chips really gain nothing out of going into the 4.8-5.0GHz realm, right? Isn't this just undue stress on the chip?

That's what somebody posted the other day (or was it PD instead of BD?) but I think we need more data before we jump to that conclusion. The guy was saying that in their rush to get it to market AMD didn't fine tune the CPU cache to work efficiently beyond that level of frequency.

 

[Anonaru]

That's what somebody posted the other day (or was it PD instead of BD?) but I think we need more data before we jump to that conclusion. The guy was saying that in their rush to get it to market AMD didn't fine tune the CPU cache to work efficiently beyond that level of frequency.

There's a lot of testing done on these forums (RG or the OP of those threads can link you) showing that performance scales very poorly per clock beyond 4.3GHz on BD. Like, nearly non-existant. Some tests showed an actual net DECREASE in performance in some areas, while others just show performance gains that are extremely tiny vs the amount of heat and voltage created. 4.3 always looks like the sweet spot

Piledriver shows very strong scaling until 4.6, and still chugs along nicely till 4.8. It still does better than BD beyond this, but I don't know the "cap" because I've only ever pushed a PD part up to 5.1GHz, still netting decent gains over previous speeds when also using the FSB

 

[trents]

Thanks. Great info.

 

[ssjwizard]

In my testing BD clock scaling starts to choke around 3.8Ghz and gets progressively worse until 4.4ish when it essentially flatlines. You can get gains but they aren't worth attaining. RGone's testing showed similar results voltage increase+heat != performance gains when CPU > 4.3Ghz in his graphs.

 

[Anonaru]

In my testing BD clock scaling starts to choke around 3.8Ghz and gets progressively worse until 4.4ish when it essentially flatlines. You can get gains but they aren't worth attaining. RGone's testing showed similar results voltage increase+heat != performance gains when CPU > 4.3Ghz in his graphs.

Here's the guy I was talking about! Thanks man, I always forget what your forum name is

 

[ssjwizard]

NP, I wont link to my results but they are easily found by searching for "AMD FX bulldozer vs Piledriver" in this section.

That said its possible that you can get 5Ghz or faster to be stable(or stable enough for a screen capture anyways) but ~4.8Ghz is when the challenge begins you will need to start playing with alot of settings to try and work the CPU up to a higher speed. My top speed with my 8120 was 5.45Ghz on watercooling. Of course my benchmark scores at that speed were only 1--2% improved over the scores I posted at 4.8Ghz. Your in the realm of overclocking that chip just to say you did because there is literally almost no actual performance gains, so have a little fun with it!

 

[RGone]

You can get gains but they aren't worth attaining. RGone's testing showed similar results voltage increase+heat != performance gains when CPU > 4.3Ghz in his graphs.

From what I posted in 'ssjwizards' thread; you all know I had been sitting on my own results for months really. I took my data and made the graph. (freeken job for a red-nake). I posted to 'add to' what ssjw was seeing. Then 'caddi daddi' weighed in with his response that was very identical to what we were seeing and saying about the Bulldozer cpu. I think 'anonaru' has registered similar. Without the stepped testing a review site would never get this type of 'trail' out to the average Joe Six Pack user.

There are LIGHT YEARS difference in what the OP is using for cooling and what the other 99% of users come in here with asking for help to do a simple overclock. I merely told "P8ntballjunky06" that he did not have enough Vcore and he upped the Vcore without whining about raising it and his FX-8150 went on up some more.

A great difference in the calibre of motherboard between what "P8ntballjunky06" is using and most others, since he went for the gold to begin with.

Now the FX-8120 I had was P95 Blend mode stable @4.8Ghz for TWO hours. But it was using really really good water cooling. The Vcore for me was close to 1.55V. That is a lot of Vcore to get Real Stability."P8ntballjunky06's" FX-8150 seems that little bit better than an FX-8120 as we are also beginning to be able to see.

My data that I added to what "ssjw" was being set on because most people cannot hear. "P8ntballjunky06" seems to be better equipped and has a very good FX-8150 that raises him into the top 10% of users or so. Far above what is generally in the AMD CPU Forum section asking for help. Of course we have not seen any results of "P8ntballjunky06" trying to P95 Blend at his latest CPU Mhz levels. He maynot be truly stable up there. I feel he will give a real go though and stands a chance of making it to stable. But those speeds are not doable by the other +90 odd percent of posters in this forum section.

What I wanted to be able to show the 90 percenters is that you really can get 95% or maybe even a little more out of Bulldozer 8 core processor before you HIT those last two BIG voltage steps that also bring on the HEAT.

Straining to get 4.5Ghz on an 8 core seems pretty senseless in the LIGHT of day. I can use my FX-8120 below 4.5Ghz with almost 0.15 LESS Vcore than what it takes to get to 4.8Ghz stable. For me that is huge chunk of lessened VCore. Big lot of lesser VCore. Plus a huge lot of LESS heat.

That is what I wanted the 90 percenters to see. The overclock they could reach on a well cooled case with a CM 212 EVO coooler was PLENTY for using about +95% of the FX-8120 and FX-8150. But what is the old addage about the hand-writing on the wall? Well the writing is here, just unseen.

By the way "P8ntballjunky06" nice system man and to the Forums. RGone...ster.

 

[P8ntballjunky06]

From what I posted in 'ssjwizards' thread; you all know I had been sitting on my own results for months really. I took my data and made the graph. (freeken job for a red-nake). I posted to 'add to' what ssjw was seeing. Then 'caddi daddi' weighed in with his response that was very identical to what we were seeing and saying about the Bulldozer cpu. I think 'anonaru' has registered similar. Without the stepped testing a review site would never get this type of 'trail' out to the average Joe Six Pack user.

There are LIGHT YEARS difference in what the OP is using for cooling and what the other 99% of users come in here with asking for help to do a simple overclock. I merely told "P8ntballjunky06" that he did not have enough Vcore and he upped the Vcore without whining about raising it and his FX-8150 went on up some more.

A great difference in the calibre of motherboard between what "P8ntballjunky06" is using and most others, since he went for the gold to begin with.

Now the FX-8120 I had was P95 Blend mode stable @4.8Ghz for TWO hours. But it was using really really good water cooling. The Vcore for me was close to 1.55V. That is a lot of Vcore to get Real Stability."P8ntballjunky06's" FX-8150 seems that little bit better than an FX-8120 as we are also beginning to be able to see.

My data that I added to what "ssjw" was being set on because most people cannot hear. "P8ntballjunky06" seems to be better equipped and has a very good FX-8150 that raises him into the top 10% of users or so. Far above what is generally in the AMD CPU Forum section asking for help. Of course we have not seen any results of "P8ntballjunky06" trying to P95 Blend at his latest CPU Mhz levels. He maynot be truly stable up there. I feel he will give a real go though and stands a chance of making it to stable. But those speeds are not doable by the other +90 odd percent of posters in this forum section.

What I wanted to be able to show the 90 percenters is that you really can get 95% or maybe even a little more out of Bulldozer 8 core processor before you HIT those last two BIG voltage steps that also bring on the HEAT.

Straining to get 4.5Ghz on an 8 core seems pretty senseless in the LIGHT of day. I can use my FX-8120 below 4.5Ghz with almost 0.15 LESS Vcore than what it takes to get to 4.8Ghz stable. For me that is huge chunk of lessened VCore. Big lot of lesser VCore. Plus a huge lot of LESS heat.

That is what I wanted the 90 percenters to see. The overclock they could reach on a well cooled case with a CM 212 EVO coooler was PLENTY for using about +95% of the FX-8120 and FX-8150. But what is the old addage about the hand-writing on the wall? Well the writing is here, just unseen.

By the way "P8ntballjunky06" nice system man and to the Forums. RGone...ster.

thank you for the complement on my setup, i will report back with my resolts after work i did run p95 at 4.95 for 1-2 hours last night and it was stable temps were 40-45c