AMD Opteron 165 CCBWE 0550 UPMW overclocking

[Vegitto]

Here we go . I'll just stress this baby for one hour per setting, as I'm confident it'll do at least 2.8 GHz. It's now 1h stable at 300x8 = 2400 MHz, stock Vcore (1.350V, 1.328 actual volts).

Will report back later.

1h stable @ 300x8, stock voltage.
1h stable @ 289x9, 1.4 Vcore
1h stable @ 289x9, 1.375 Vcore
It's now 1h Prime stable at 1.375V, (1.344 actual volts) at 2.601 GHz.

Currently, I'm attempting a process called 'burn-in', which is actually trying to decrease the voltage necessery for a specific frequency, thus lowering the temps.

UPDATE:
After a burn-in period of 24h, I'm now rock stable at 0.976V (1.000V mobo setting) at 1800 MHz. I will attempt 2.6 GHz at stock voltage in a short period of time .

 

[Doctorweir]

I'd go faster by isolating the processor an cranking up until Prime fails very fast. Then add voltage up to 1.5V and try further...What cooling?

I e.g. started with 1.4V and 2.6GHz instantly and elaborated further...well, not to much as I am hitting the heat-wall @ 2.7 with core 0 but enough for me...until I invest in high-end water

Did you already evaluate the highest HT frequency of your Lanparty board?
You are kidding about the ram size and type you use, aren't you? But it will work...just make sure to use a high divider...

 

[Vegitto]

Originally posted by: Doctorweir
I'd go faster by isolating the processor an cranking up until Prime fails very fast. Then add voltage up to 1.5V and try further...What cooling?

I e.g. started with 1.4V and 2.6GHz instantly and elaborated further...well, not to much as I am hitting the heat-wall @ 2.7 with core 0 but enough for me...until I invest in high-end water

Did you already evaluate the highest HT frequency of your Lanparty board?
You are kidding about the ram size and type you use, aren't you? But it will work...just make sure to use a high divider...

I did get the highest HT frequency, it's 344 MHz. The RAM is no joke, I'll get 2 GB of G.Skill next week. Core 0 is constantly crapping out :|.

 

[Doctorweir]

Core 0 is also my "weak" one. At what speed and volts? Not 2.4, is it?

 

[Vegitto]

Originally posted by: Doctorweir
Core 0 is also my "weak" one. At what speed and volts? Not 2.4, is it?

300x8 at stock vcore is a-ok.
333x8 at any vcore is not cool. I think it might be the RAM, though, it's never been working properly.. Even at it's rated spec of 166 MHz it gets flakey, sometimes..

 

[Munky]

Just out of curiosity, why are you using the 8x multi when 9x is the default?

 

[Vegitto]

Originally posted by: munky
Just out of curiosity, why are you using the 8x multi when 9x is the default?

Just to see if the HTT would be able to keep it that high for a longer period of time. Also, if I can get 2.4 GHz stable, I'd much rather do that at 300x8 than at 267x9, just because my RAM can take it .

 

[TigerClaw27]

I'm guessing you're running on stock air cooling? I hit the core0 barrier at 2.7 (300x9) too. I can get to 2.65 on stock Vcore, past that its tricky, my CPU load temperature goes above 50C. One thing I still have to do is short circuit the temperature sensor on the stock AMD HSF so that the fan spins at 5000 RPM all the time, and control the fan speed from BIOS instead. Even at full load my fan rarely spins faster then 2700 RPM.

Keep an eye on your CPU temperature as you do your stability tests, I've noticed that on my 165 going over 55C when running at 2.7 will cause the core to fail.

 

[Vegitto]

Originally posted by: TigerClaw27
I'm guessing you're running on stock air cooling? I hit the core0 barrier at 2.7 (300x9) too. I can get to 2.65 on stock Vcore, past that its tricky, my CPU load temperature goes above 50C. One thing I still have to do is short circuit the temperature sensor on the stock AMD HSF so that the fan spins at 5000 RPM all the time, and control the fan speed from BIOS instead. Even at full load my fan rarely spins faster then 2700 RPM.

Keep an eye on your CPU temperature as you do your stability tests, I've noticed that on my 165 going over 55C when running at 2.7 will cause the core to fail.

Yeah, that sounds about right.. However, I'm currently saving up and working my ass off to be able to purchase the following as soon as possible:
-OCZ phase-change unit, should arrive late March --> ?300-?350
-Big RAID array (>1.5 TB) --> ?600-?800
-Nice case --> ?100-?150
-DVD burner --> ?75

 

[TigerClaw27]

I dont think there are many ppl going over 2.7 on stock HSF. But i do wonder how many are going over 2.7 on other air solutions?

 

[Vegitto]

Originally posted by: TigerClaw27
I dont think there are many ppl going over 2.7 on stock HSF. But i do wonder how many are going over 2.7 on other air solutions?

Hey, I just saw you've got the same proc as me. I might try that setting sometime soon .

 

[Doctorweir]

Originally posted by: TigerClaw27
I'm guessing you're running on stock air cooling? I hit the core0 barrier at 2.7 (300x9) too. I can get to 2.65 on stock Vcore, past that its tricky, my CPU load temperature goes above 50C. One thing I still have to do is short circuit the temperature sensor on the stock AMD HSF so that the fan spins at 5000 RPM all the time, and control the fan speed from BIOS instead. Even at full load my fan rarely spins faster then 2700 RPM.

Keep an eye on your CPU temperature as you do your stability tests, I've noticed that on my 165 going over 55C when running at 2.7 will cause the core to fail.

I can fully second that...my basic watercooling keeps me running stable at 2.7 however...but the temp / core0 issue is there as soon as I go higher.

 

[robertk2012]

Originally posted by: Vegitto
Here we go . I'll just stress this baby for one hour per setting, as I'm confident it'll do at least 2.8 GHz. It's now 1h stable at 300x8 = 2400 MHz, stock Vcore (1.350V, 1.328 actual volts).

The next setting I'll try will be 333x8 = 2664 MHz. I'm using that setting, because 333x9 = 2997 MHz, AKA 'The Big 3' .

Will report back later.

1h stable @ 300x8, stock voltage.

333x8 at stock voltage is a no-go. Now attempting 333x8 at 1.375V, mobo setting.
333x8 at 1.375V is a no-go. Now attempting 333x8 at 1.400V, mobo setting.
333x8 at 1.400V is a no-go.

I'm tired of this already. I'm going to try 311x9 @ 1.5 V mobo setting.
Aaand INSTANT BSOD! Looks like I got a dud. Again :|.

I'm tired of this already . I think I'm gonna wait until my new RAM arrives.

Run it at 333x7 and see what happens.

 

[Vegitto]

Originally posted by: robertk2012

Originally posted by: Vegitto
Here we go . I'll just stress this baby for one hour per setting, as I'm confident it'll do at least 2.8 GHz. It's now 1h stable at 300x8 = 2400 MHz, stock Vcore (1.350V, 1.328 actual volts).

The next setting I'll try will be 333x8 = 2664 MHz. I'm using that setting, because 333x9 = 2997 MHz, AKA 'The Big 3' .

Will report back later.

1h stable @ 300x8, stock voltage.

333x8 at stock voltage is a no-go. Now attempting 333x8 at 1.375V, mobo setting.
333x8 at 1.375V is a no-go. Now attempting 333x8 at 1.400V, mobo setting.
333x8 at 1.400V is a no-go.

I'm tired of this already. I'm going to try 311x9 @ 1.5 V mobo setting.
Aaand INSTANT BSOD! Looks like I got a dud. Again :|.

I'm tired of this already . I think I'm gonna wait until my new RAM arrives.

Run it at 333x7 and see what happens.

Yeah, I might do that tonight.. I'm curious if I'll ever reach the big 3G. Maybe with an aftermarket cooler?

 

[Absolute0]

My 165s did 2.7 on stock Vcore. You do know that many of the 0550UPMWs suck right, there are different substrate numbers that you have to look at, i passed up one for 300$ because it was the crappy substrate. Also PC2700 ram isn't going to help at all.

 

[lopri]

What's RAM has to do with CPU clocking? I've been OC'ing all along with a pair of PC2700s. (Although my sticks seem heroic at clocking. Goes all the way up to 230MHz)

 

[Phew]

Originally posted by: TigerClaw27
I dont think there are many ppl going over 2.7 on stock HSF. But i do wonder how many are going over 2.7 on other air solutions?

Isn't the retail Opteron 165/170/175 cooler a snazzy heatpipe unit that performs almost as well as an XP-90?

These Opterons are funny. My 170 goes up to 2.5 GHz on stock volts no sweat, Prime stable all day. However, past 2.6 GHz all of a sudden requires big voltage. For me, the extra few hundred MHz aren't worth the extra heat and power consumption.

 

[robertk2012]

Originally posted by: Phew

Originally posted by: TigerClaw27
I dont think there are many ppl going over 2.7 on stock HSF. But i do wonder how many are going over 2.7 on other air solutions?

Isn't the retail Opteron 165/170/175 cooler a snazzy heatpipe unit that performs almost as well as an XP-90?

These Opterons are funny. My 170 goes up to 2.5 GHz on stock volts no sweat, Prime stable all day. However, past 2.6 GHz all of a sudden requires big voltage. For me, the extra few hundred MHz aren't worth the extra heat and power consumption.

Thats how most processors are. You reach a point where you get diminshing returns.

 

[Vegitto]

Originally posted by: lopri
What's RAM has to do with CPU clocking? I've been OC'ing all along with a pair of PC2700s. (Although my sticks seem heroic at clocking. Goes all the way up to 230MHz)

It's just that THIS RAM won't even do 166 MHz, it's not even really stable at 133 MHz. And when I tried 333x8, the memory was running at 333/(200/100)=166.5 MHz.

 

[wireeater]

Sounds like my chip(055). Past 2.4 and core 0 fails with in seconds of prime, once the temp hits about 43-45c. I am assuming better cooling will make it a little easier but I don't feel like spending the money for one. I am running stock now, with 1.2-v and it runs about 30c idle and handles my games fine so I will stick with it until I get bored again.

 

[Vegitto]

Updated.

 

[Vegitto]

Updated again.

 

[Ackmed]

Originally posted by: Absolute0
My 165s did 2.7 on stock Vcore. You do know that many of the 0550UPMWs suck right, there are different substrate numbers that you have to look at, i passed up one for 300$ because it was the crappy substrate. Also PC2700 ram isn't going to help at all.

What are good numbers?

Just got mine, and its CCBWE 0551UPMW.

 

[Vegitto]

Originally posted by: Ackmed

Originally posted by: Absolute0
My 165s did 2.7 on stock Vcore. You do know that many of the 0550UPMWs suck right, there are different substrate numbers that you have to look at, i passed up one for 300$ because it was the crappy substrate. Also PC2700 ram isn't going to help at all.

What are good numbers?

Just got mine, and its CCBWE 0551UPMW.

Those are really good, too.

 

[Vegitto]

Updated.