X4 955 Oc Help

[IGhzI]

I've been trying to hit the 4.0 mark but can't get it stable. I've tried messing with the fsb also 217 x18.5 mean while dropping everything else under stock specs ht nb etc as for the volts its 1.47-1.5 temps are around 52c. Does anyone have a clue on what else I could try

 

[Don Karnage]

Sounds about right. My 955 would start having issues over 50C

 

[bononos]

Sounds about right. My 955 would start having issues over 50C

Is 50C the real temp? AMD cpus don't report temps properly unlike Intel and some of the temps can be bizzare eg -ve.
And even if they were the real temps, AMD cpus have lower max operating temps to be able to conform to spec on reliability/power consumption.

 

[paperwastage]

Is 50C the real temp? AMD cpus don't report temps properly unlike Intel and some of the temps can be bizzare eg -ve.
And even if they were the real temps, AMD cpus have lower max operating temps to be able to conform to spec on reliability/power consumption.

AMD cpus should be able to report accurate temps (maybe you have to try different programs.... same bug with some intel cpus too)

but for unlocked AMD cpus, usually the temp sensor reports no accurate data

 

[BallaTheFeared]

Is 50C the real temp? AMD cpus don't report temps properly unlike Intel and some of the temps can be bizzare eg -ve.
And even if they were the real temps, AMD cpus have lower max operating temps to be able to conform to spec on reliability/power consumption.

That is a problem with Thuban/bulldozer, not Deneb.


52C is within reason for stability testing, you could try using more fsb and less multi. You should also be running your cpuNB as high as you can.

Don't sacrifice cpuNB speed for core speed.

Do you have a C2 or C3? You can also try up to 1.55v on air if your cooling can handle it (62C max for most Denebs).

 

[Don Karnage]

That is a problem with Thuban/bulldozer, not Deneb.


52C is within reason for stability testing, you could try using more fsb and less multi. You should also be running your cpuNB as high as you can.

Don't sacrifice cpuNB speed for core speed.

Do you have a C2 or C3?

Has to be a C3. C2s barely get over 3.6

 

[BallaTheFeared]

I thought 3.8 was upper end for C2, and he's not stable

 

[IGhzI]

I'm stable at 3.8 1.425 volts cpu-nb 2200 at 1.2 volts and ht at 2200. I guess ill have to play with the fsb more

 

[bononos]

That is a problem with Thuban/bulldozer, not Deneb.


52C is within reason for stability testing, you could try using more fsb and less multi. You should also be running your cpuNB as high as you can.

.........

A fairly recent (2011) post in AMDZone said Thuban/Deneb's still do not report accurate temps and some sort of calibration is needed.
http://www.amdzone.com/phpbb3/viewtopic.php?f=532&t=138425&start=125
http://www.xbitlabs.com/articles/cpu/display/phenom-ii-x4-965_5.html
But this article said the late Deneb's were finally calibrated correctly.

Apart from this the entire K8 family used some non-physical scale for their on-die temps which is also non-linear so a simple offset is quite inaccurate.

 

[jacktesterson]

I'm stable at 3.8 1.425 volts cpu-nb 2200 at 1.2 volts and ht at 2200. I guess ill have to play with the fsb more

If your at 3.8, stay there.

Adding a ton of juice to get to 4.0 won't be worth it.


I had a 1090T that refused to hit 4.0 without a ton of extra voltage compared to 3.9.

Some chips just dont have it to make 4.0.

 

[Vic Vega]

...

 

[Leyawiin]

Diminishing returns. With a Phenom II any increase in performance from 3.8 Ghz to 4.0 or beyond will be minimal. Its not worth the extra heat/power consumption (which often goes up dramatically those last few hundred Mhz).

 

[soccerballtux]

I've been trying to hit the 4.0 mark but can't get it stable. I've tried messing with the fsb also 217 x18.5 mean while dropping everything else under stock specs ht nb etc as for the volts its 1.47-1.5 temps are around 52c. Does anyone have a clue on what else I could try

I found it harder to get an FSB overclock, but that's because my RAM timings at the lower RAM divider were defaulting to 5-4-4-12 (DDR2-800). That kept me from overclocking past 228 on the FSB. Then I manually overrided RAM back to 5-5-5-18 and immediately went to 260mhz bus * 15.5 no problem.

I run 1.47v.

Keep the cpu-nb at 2ghz, ht at 2ghz. Give the cpu-nb another 0.1-2v just in case, then simply try 20x200. No good?

 

[soccerballtux]

I thought 3.8 was upper end for C2, and he's not stable

on 32bit yes, 64b much was ~3.5avg -3.6 if you lucky.

 

[soccerballtux]

If 20x200 and 1.47v doesn't work, better to sit happy at 3.8 or 3.9ghz and then overclock your cpu-nb. You can get ~10% better performance in games simply by getting that up to 2.4-2.6ghz. Most people can get to 2.6ghz I think, both my Phenom2s have done that.
2.6ghz especially helps with the minimums. Very noticable.

 

[TakeNoPrisoners]

I can usually run 4Ghz at 1.45V during the winter when temps are lower.

I run it all the way up to 60C beyond that I'm not comfortable.

Been running at those settings for about a year and a half now.

 

[nismotigerwvu]

Has to be a C3. C2s barely get over 3.6

Not always, I have a C2 running 3.7 under 1.4V. I haven't tried pushing it harder as it's already bottlednecked enough by my 4850, but once the midrange gets sorted out (or the 7850 drops a little more) I'll have a nice reason to try a little harder.

 

[SickBeast]

OP 4ghz is too much for a Phenom II. At 3.8ghz they hit a wall.

 

[Vic Vega]

...

 

[IGhzI]

I tried the 1.47 with no luck.. And I can't get the cpu-nb above 2200 I keep get an error as soon as I boot and thats leaving the volts at 1.2. But I'm also running 1333 so. I still think I can hit the 4 mark but do I have to increase the volts on the memory? Im missing something I know it

 

[SickBeast]

Noo...

You got lucky.

 

[soccerballtux]

I think you're doing something wrong. Any and all chips can hit >2.2ghz on the CPU-NB.

 

[tincart]

OP 4ghz is too much for a Phenom II. At 3.8ghz they hit a wall.

I have hit 4Ghz with my 955BE. Completely stable. I generally run it at 3.8 since that's where the performance/heat balance is best.

 

[Vic Vega]

...

 

[SickBeast]

No, not really. You claim 3.8GHz is the end of the world for Deneb and that's just not true. There are plenty of people north of 4.0GHz. You're just trolling though, because you personally couldn't get the clock your alternative is to bash the entire product. I don't know why you couldn't get past 3.8GHz, maybe it was your cooling, your memory, your power delivery - maybe it WAS your CPU, maybe you got a dud. Your dud doesn't speak for the entire product.

No I'm not trolling. The 3.8ghz wall is widely known, just as the 2500k hits a heat and voltage wall at 4.5ghz. Perhaps with some newer steppings or highly binned chips you can hit 4.2ghz with a Phenom II, but I would bet that the voltages and heat would not be reasonable.