Sign in with
Sign up | Sign in
Your question

2.8GHZ with AMD x2 3800+

Last response: in CPUs
Share
September 1, 2006 9:50:07 AM

Hi,i recently went into the overclocking thing bought a new case with Gigabyte watercooling and must say it works awesomely..

I got my CPU from standart 2.0GHZ to 2.8GHZ stable.

cant get it further wanted to get to 3.0GHZ but after that it doesnt boot.

Is that considered a good overclock?

More about : 8ghz amd 3800

September 1, 2006 10:46:18 AM

an 800mhz??
very good. thats fx-62 speed.
try for 2.9 or 2.85 if u really want but for the sake of benchmarks id leave it where it is.

its pretty quick.
September 1, 2006 10:46:52 AM

considering that its an AMD cpu thats a really good overclock, most of the time ppl can only get it up to 2.2-2.4 range. can u post ur idle and load temps
Related resources
September 1, 2006 11:09:24 AM

it runs at an idle temp of 22 degrees and could get it passed 34 degress with super pi 3dmark or pcmark and after an hour of far cry..

busy with a webiste to post screenshots
September 1, 2006 11:14:14 AM

Quote:
considering that its an AMD cpu thats a really good overclock, most of the time ppl can only get it up to 2.2-2.4 range. can u post ur idle and load temps


:?: In most cases 90nm Athlons OC up to 2.7-2.9G, especially single cores can achieve these speeds pretty easily on air cooling if you look around. Only conroes beat them.
September 1, 2006 11:18:39 AM

I think he was referring to the X2 3800+, not the 90nm Athlons in general.
September 1, 2006 11:22:19 AM

OK but 2.6-2.8 is still the norm for them. Many people OC the X2 3800+ to 2.6
for prolonged periods.
September 1, 2006 12:19:35 PM

yes i was referring to the X2 3800 not the Anthlon single core and i based what i said on personal experience and what i've seen from ppl that i know personally. srry if the informations slightly off.
a b à CPUs
September 1, 2006 12:31:20 PM

Quote:
OK but 2.6-2.8 is still the norm for them. Many people OC the X2 3800+ to 2.6
for prolonged periods.


It's definitely not the norm. The avg is a 400-500MHz overclock.

This website took in averages and compilled charts.
September 1, 2006 12:48:38 PM

Yes but at what voltage??? I keep my x2 3800 at 2.45@1.4v.
At 2.55 it needs 1.475v at 2.6 it needs probably around 1.55v and thats WAY too much. I dont want to screw my mobo or PSU.

Chips OC to some margin then they need more and more voltage I would like to know how many off those ppl who have been able to hit 2.7-2.8 have hit that with only minor voltage boost (to 1.45) ??

This could be also mobo related I think there is a difference of even 200Mhz that a CPU will do on a different mobo.
September 1, 2006 12:49:37 PM

But Vcore is 1.35 and they send it to 1.45-1.50 for more than that. An we're talking about water cooling while those are the figures you get on air.
September 1, 2006 12:54:27 PM

Quote:
Yes but at what voltage??? I keep my x2 3800 at 2.45@1.4v.
At 2.55 it needs 1.475v at 2.6 it needs probably around 1.55v and thats WAY too much. I dont want to screw my mobo or PSU.

Chips OC to some margin then they need more and more voltage I would like to know how many off those ppl who have been able to hit 2.7-2.8 have hit that with only minor voltage boost (to 1.45) ??

This could be also mobo related I think there is a difference of even 200Mhz that a CPU will do on a different mobo.


OK, not all CPUs are the same. I can keep my 3000+ @ 2.1G with 1.20V Vcore and it is ultra stable but I presume it's just my luck.
BTW; most 939 mombos don't have problems going to 1.55V because 1.50V is the nomial voltage of the onder 939 single cores.
a b à CPUs
September 1, 2006 1:07:14 PM

Quote:
But Vcore is 1.35 and they send it to 1.45-1.50 for more than that. An we're talking about water cooling while those are the figures you get on air.


He's not using a DIY kit.. he's using a gigabyte watercooling kit. Which means it probably performs about on par with todays high end water coolers.

And the Voltages quoted on the charts are not those used to overclock. They're the stock voltages quoted for each model number because some models differenciate in core voltages from one stepping to the other. So the CPUVID is only there so people can figure out what to expect from there particular stepping.

400-500MHz seems to be the norm for an X2 3800+ (the average) Watercooling doesn't help all that much when overclocking X2 Processors. I've found that you need to move towards moe extravagant cooling solutions to get the most out of X2 processors. I mean I thermo-electrically cool my X2 4800+ so that it can hit 3.2GHz. So yeah, they're not really good overclockers (compared to the competition that is).
September 1, 2006 1:10:02 PM

My mobo can go to 1.8V. That is what bios says puting something like 1.7v would probably make your cpu consume 200-300W. who knows. Can the voltage regulators keep that current or will they break??

Its not about the options in bios I say you have a gigabyte mobo and can run the cpu at 2.45 with 1.4v and then you get an ASUS premium mobo and that same cpu now runs 2.6 at 1.4v.
Its about the quality of mobos.
September 1, 2006 1:11:08 PM

Yes, I have to say it; they're not as good and even the A64 single cores are much better.
September 1, 2006 1:12:51 PM

Right. (My mombo is an ASUS BTW :D  )
a b à CPUs
September 1, 2006 1:12:56 PM

Quote:
My mobo can go to 1.8V. That is what bios says puting something like 1.7v would probably make your cpu consume 200-300W. who knows. Can the voltage regulators keep that current or will they break??

Its not about the options in bios I say you have a gigabyte mobo and can run the cpu at 2.45 with 1.4v and then you get an ASUS premium mobo and that same cpu now runs 2.6 at 1.4v.
Its about the quality of mobos.


Anything above 1.45v is not recommended if you're not using some sort of extravagant cooling setup. You can do it, try it... but don't be surprised if you end up shortening the life of your processor.

1.50-1.55v is alot of heat. You're looking at around 133-140W of heat.
a b à CPUs
September 1, 2006 1:14:16 PM

Quote:
Yes, I have to say it; they're not as good and even the A64 single cores are much better.

MUCH better... You see far more Single Cores past 3Ghz then you do Dual Cores.
September 1, 2006 3:35:17 PM

Wow, if 400MHz is the norm, then I am very happy with my overclock. I'm running an AM2 Athlon X2 3800+ @ 2.8GHz and 1.42V. My idle temp is 27C to 30C, with one core maxed it's 37C to 40C, and the highest temp I have seen is 43C (I can monitor the temp during benchmarks from mobo LED readout). My CPU cooler is the Arctic Cooling Freezer 64, and I have the Antec P180b case.

September 1, 2006 3:51:18 PM

Better cooling doesnt make higher voltage any safer.
Actualy the below zero cooling should kill the proc even better than air or water cooling.

Its not about the absolute temperature but the difference between the highest temp and the ambient one the processor cools to when its off.
September 1, 2006 3:52:51 PM

Try and run two instances of prime95 and set affinity for every core.

Impresive clock mate :) 
a b à CPUs
September 1, 2006 5:36:09 PM

Quote:
Better cooling doesnt make higher voltage any safer.
Actualy the below zero cooling should kill the proc even better than air or water cooling.

Its not about the absolute temperature but the difference between the highest temp and the ambient one the processor cools to when its off.

In order to theorise you need to understand the fundamentals of what causes a processor to die slowly when voltages are raised. It's heat related. It's called electrical migration. The transistors (small gates) slowly cease functioning.

Super cooling your machine will ensure it lasts longer. Electricity flows more easilly on cooler circuits then it does on hot circuits. So the cooler the CPU the more easilly power runs through it. Also the more voltage you can implement seeing as electrical migrations is caused by heat & voltage combined.
OC Forums Northwood Sudden Death Syndrom
September 1, 2006 5:40:12 PM

So, in theory, sub-zero cooling won't do any harm to the CPU.
Like Phase-Change Cooling.
a b à CPUs
September 1, 2006 5:46:43 PM

Quote:
So, in theory, sub-zero cooling won't do any harm to the CPU.
Like Phase-Change Cooling.


So long as it's sealed.. In use Dielectric greese.

See the one thing that can kill it is Condensation.
September 1, 2006 5:55:07 PM

Not the condensation per se, the result of the condensation: Water. :wink:
September 1, 2006 6:02:44 PM

Quote:
Better cooling doesnt make higher voltage any safer.
Actualy the below zero cooling should kill the proc even better than air or water cooling.

Its not about the absolute temperature but the difference between the highest temp and the ambient one the processor cools to when its off.

In order to theorise you need to understand the fundamentals of what causes a processor to die slowly when voltages are raised. It's heat related. It's called electrical migration. The transistors (small gates) slowly cease functioning.

Super cooling your machine will ensure it lasts longer. Electricity flows more easilly on cooler circuits then it does on hot circuits. So the cooler the CPU the more easilly power runs through it. Also the more voltage you can implement seeing as electrical migrations is caused by heat & voltage combined.
OC Forums Northwood Sudden Death Syndrom

He's very right because it all has to do with motion at the atomic level. A hot processor's atoms and electrons will be boucing around like crazy and will either not conduct electricty as well or will start to deform and cease to work. A super cooled cpu will have very sturdy atomic structure and will conduct electricty very effieciently. The other guy was right though that the difference between peak temp and low temp also affects it because too much shift and it can "warp" somewhat.
September 1, 2006 6:09:59 PM

I have my X2 3800+ at 2.56ghz on air cooling with the Arctic Freezer 64 Pro CPU cooler @ 1.3825v and is 12 hours prime 95 stable (running 2 instances) with a max temp of 50 degrees celsius, which I relise is not low, but it only got above 48 around the 10 hour mark and average is probably around 46 during the test and during gaming im 44 - 46 depending on how long ive been on and what im playing and I have the BH-E4 stepping version of it, so im glad to see I got above the average of that chart.

Best,

3Ball
September 1, 2006 6:15:53 PM

thats a great over clock
September 1, 2006 6:22:28 PM

Yes I know all this and I have been reading a lot on the subject. And what I read somewhere is that is all bulls***. Sub zero will kill it cos its a big temp differential from ambient. And when you turn your comp off the cpu warms up to ambient. I have read that this temp difference is what actualy bothers chips in general and the old rull that for every 10C you lover the temp is so old rule it isnt used for years.

Actualy in theory a subzero cooled cpu that runs 24/7 and is never turned off would run indefinatly.

But I think this is realy all stupid cos how many pople have killed their cpu from overclocking ??? did it just die over a year or two ??
I hardly saw any data on the net and I was looking really hard.

From what I have seen its way much more probable you would damage your mobo from ocing then your cpu. Pople neglect the other components on the mobos like capacitors and mosfets.

I have never damaged a cpu and I have used PCs from the 286 era. But mobos I have killed a few. And in my experience if you burn a CPU then so what ??? I'll buy another one im not in habbit of buying 400-1000$ cpus I buy cheap and oc.
But when I kill a mobo Im in deep shit cos im running RAID 0 on my mobo and you need the exact chipset to run the hdds on the other mobo and after running a mobo for a year it might just become hard to get that mobo again or that chipset cos it becomes obsolete.
September 1, 2006 6:40:05 PM

that is a very nice overclock from what i seen either 939 or am2 i could only get around 2.5ghz out of my 3800x2 (939) i was a bit disapointed untill i saw that it was an average but would of like to keep up with my d805 setup.
what stepping is that for your am2?
September 1, 2006 6:54:40 PM

Your temps are suspiciously low, see if Core Temp comes up with the same numbers...

Considering your rather low Vcore, that CPU probably have some more headroom to it, don't be afraid to increase it, the long term effects of electromigration are overrated.

The HT link is most likely holding you back, try to use a 3x multiplier instead of 4x, it'll keep the it below 1GHz.

Quote:
He's not using a DIY kit.. he's using a gigabyte watercooling kit. Which means it probably performs about on par with todays high end water coolers.


Try "on par with high-end, heat pipe based HSFs" instead, that thing is not even in the same league as DIY systems based on car heatercores and powerfull pumps.
September 1, 2006 7:22:07 PM

You keep that thing at 3Ghz! wow man bravo!

Please tell me how long you have keept that system so oced? do you use it 24/7 ?? did you make any aditional cooling to the mobo northbridge? mosfets?
Did you measure the electricity it draws from the outlet ??
September 1, 2006 7:46:06 PM

Heh, not quite 3GHz but Close Enough, I'm using an older Manchester core fabbed in the 43rd week of '05, those need a lot more Vcore than their newer siblings that were fabbed in early '06.

I've been running that CPU 24/7 under full load (two instances of F@H) at 2.8 to 2.9GHz ever since I got it back in December, I'm still using the stock NB fan and the MOSFETs seem to do fine with the 80mm rear intake fan blowing on them.

I never bothered measuring how much juice my 'rig uses but my EG365P-VE 365W Enermax PSU isn't causing me any problems...

EDIT: I just found one of my older validated CPU-Z link from December, still 2.9GHz but my memory was running slower than it is now.
September 1, 2006 8:00:23 PM

Nice OC; please post a CPU-Z screenshot or verification of some sort. 2.8GHz on air would be amazing; with water it's a nice overclock.
September 1, 2006 8:32:25 PM

Quote:

In order to theorise you need to understand the fundamentals of what causes a processor to die slowly when voltages are raised. It's heat related. It's called electrical migration. The transistors (small gates) slowly cease functioning.

Super cooling your machine will ensure it lasts longer. Electricity flows more easilly on cooler circuits then it does on hot circuits. So the cooler the CPU the more easilly power runs through it. Also the more voltage you can implement seeing as electrical migrations is caused by heat & voltage combined.
OC Forums Northwood Sudden Death Syndrom


8O Interesting link; I have put my ex 2G Celeron @ 2.67 (FSB from 100 to 133, that's all it's board allows), it's been 2 weeks, stock Vcore. I have also reinforced it with the latest 478 P4 cooler type with copper core.
Is it in danger to die so prematurely???
September 1, 2006 9:10:22 PM

This seam to be only the Northwood A problem. There is no proof of this happening to presccots or to Athlons. None that I know of. If anyone does please submit a link.
If you stay within stock voltage then you are in all probability ok.

This problem might occur more to newer processors. What if those overclocked Conroes suffer from this sindrom?? Mass panick ??
September 1, 2006 10:34:36 PM

I hope so. However, the person who now has that PC really needed more power but had no budget. If I am not wrong, the 2G and up celerons came later, after the P4A Northwood cores and scaled up to 2.8Ghz, so I am virtually within frequency range.
All this said: That Celeron really deserves to die if it can't work @ 2.67 because even so, it performs worse than a 2.2Ghz P4 :lol: 
September 2, 2006 12:47:57 PM

Quote:
Quote:
I mean I thermo-electrically cool my X2 4800+ so that it can hit 3.2GHz. (compared to the competition that is).


With what a peltier chip?
September 2, 2006 4:15:47 PM

Quote:
It's amazing how much I've forgotten how to overclock A64/X2 since I started researching on C2D overclocking.
Now I have to dig through all my past posts to get my memory back.

BTW, the characteristics of Orlean/Windsor is no different, but has been increased by overall of 100Mhz at the same voltage compared to S939.


I think there's more than meets the eye, AMD revised the TDP of Orlean & Windsor downward compared to their S939 counterparts, there also is the marked difference between late '05 and early '06 Manchesters, the later having more OC headroom despite needing less Vcore to get there. This leads me to believe that AMD did something good to their 90nm process, it seems like the quality of their cores suddenly increased in the span of a couple months.
September 2, 2006 4:44:32 PM

Quote:
2.6Ghz (260x10 or 289x9) +0.075, $33(NewEgg) Zalman 7700Cu


I did overclock my A64 3200+ Venice E6 to 2.6 GHz +0.1V on stock cooling. No need for a Zalman HSF. :wink:
September 2, 2006 6:51:22 PM

The funny thing about it is, I bought this system without the intention of doing crazy overclocking. I just got this CPU to hold me for a year until AMD quad cores come out, so I would have been totally happy with just a 400MHz overclock. Needless to say that this makes me even happier.

For those who wanted a picture with two instances of Prime95 on separate cores, here's a shot for you. Also, I included a shot of the SPD tab in CPU-Z since I messed that up in the other screenshot. My new highest temp is 45C (although the screenshot shows 44C), and that should be correct since one main feature of the BIOS update was a corrected temperature readout.




Also, here are some pictures of my whole system:






Side note...everyone who sees these pictures comments on the can of Cheese Whiz, so I might as well state ahead of time that I built the system on our kitchen table :lol: 


EDIT:
I forgot the benchmarks. Here are my 3DMark scores for various speeds. Sorry, no screenshot verification because I didn't think my overclock was anything special at the time, and I just wanted to show my friends, so I wrote down the scores and didn't export to Excel or take screenshots. I might do some verification runs later today after I finish my first wave of homework -- blah to assignments first week back to college :x

3DMark01 @2.0GHz - 26362
3DMark01 @2.4GHz - 30747
3DMark01 @2.8GHz - 34427

3DMark03 @2.0GHz - 17460
3DMark03 @2.4GHz - 18351
3DMark03 @2.8GHz - 19051

3DMark05 @2.0GHz - 10155
3DMark05 @2.4GHz - 11387
3DMark05 @2.8GHz - 12060

3DMark06 @2.0GHz - 5409
3DMark06 @2.4GHz - 5679
3DMark06 @2.8GHz - 5866

My vid card clocks are given below in my sig.
September 2, 2006 9:49:56 PM

I purchased the CPU on August 13, so yes it was quite recent. Is there any way you can find out the manufacture date from the numbers etched on the heat spreader? Here's a better shot of the CPU just in case:


Now that you bring this up I'm very interested in learning about it.
September 2, 2006 10:44:05 PM

I'd hold it right there for now. That is a very good overclock you have there. Let it burn in for awhile, tweak some other parts of the system. Then have at it again, Sounds like you can hit 3.0.
September 2, 2006 10:59:54 PM

Cool stuff. Thanks for looking into that. Now we just need more people with the same CPU and systems that are not limited by the RAM to post up their results.

@ data-freak
Could you please post a shot of your CPU or just the serial numbers themselves?
September 2, 2006 11:34:29 PM

Quote:
I purchased the CPU on August 13, so yes it was quite recent. Is there any way you can find out the manufacture date from the numbers etched on the heat spreader? Here's a better shot of the CPU just in case:


Now that you bring this up I'm very interested in learning about it.


ADA3800IAA5CU
LDB4F 0626XPMW
Q963032f60915

The first line is very straightfoward, An Athlon64 model 3800+ (duh!) for socket AM2, variable voltage, variable temp with 1MB (2x 512KB) of L2 based on a Windsor core, nothing new here.

I highlighted the important parts, here is the rundown :

LDB4F Indicates the minor revisions that happened within a stepping, not much is known about the exact meaning of every single digit but that string is often used to differentiate between "good" and "not so good" cores.

06 Is the year of manufacture, 2006 in that case.

26 Is the week of manufacture, 26th week of the year, putting it in early July.

XPMW Almost nothing is known about that string, probably some kind of production batch number.

The last line is even more cryptic, most likely a serial number that identify that particular CPU.
September 3, 2006 12:00:21 AM

My 3200+ says ADA3200DAA4BW, and, in the chart you posted it says the part definition BW indicates this: "Stepping E6, CPUID 2F, single processor, 0.09 micron, socket 939"

What is CPUID 2F? :?
September 3, 2006 6:28:45 PM

Quote:
Thanks dude.


ADA3800IAA5CU
LDB4F 0626XPMW
Q963032F60915

Anytime, I did some more reading and found out that the last four digit of the 3rd line (0915) are thought to be the core's number in that batch (915th core fabbed that week ?), still not very usefull but good to know.
September 3, 2006 7:17:21 PM

Outstanding clock! If you really want to take it to the max, and you think you've hit your highest mark, try removing the IHS from the CPU. With that out of the way, maybe but the multiplier down for the ram, up the CPU volts and away you go!
September 3, 2006 7:51:43 PM

very good overclock sir.you should remain happy with that kind of boost from your cpu.800mhz is a significant overclock whether Intel or AMD.enjoy.

Dahak

EVGA VF4 SLI MB
X2 4400+@2.4
2 7800GT'S IN SLI MODE
2X1GIG DDR400 MEMORY IN DC MODE
WD300GIG HD
520WATT PSU
EXTREME 19IN.MONITOR
3DMARK05:11,582
!