X2 3800+ Steppings

[Guest LithoTech]

The temperature difference between the two on all dual core processors is common. My core 0 at idle usually read 3C - 4C below ambient.

 

Ahhh k, thanks!

 

So it's safe to assume the higher temp is more accurate? If the lower one (for you at least) is below ambient at idle, the higher one must be closer to actual temp.

 

MBM is taking an average of these? Or is there yet another sensor it is getting data from?

 

My system is folding at home on both cores ATM, overnight tonight should help set in the TIM, I'll let it rest tomorrow.

 

But the temp difference in FAH is almost 10 degrees under load, yet I'm certain it was only 5 degrees when under P95 load.

 

Perhaps P95 generates more heat, and hence closes the gap due to one core convecting the heat to the other?

 

I was also trying out Orthos with this last night, using different tests, I may have got the 5 degree diff reading from the one that stresses the CPU most and generates max heat.

[Praz]

Unfortunately there is no known accurate reading. Being able to see the core temps is just another tool for us to use. If you use the reading from MBM you should be fine. That's what we have used all along. Even if the reading isn't correct we are familiar with what the relative values should be. Read through this thread for an insight (or confusion) to the core temperature readings.

"Core Temp" - New way to monitor temps on K8s and Core/2 CPUs

[Guest LithoTech]

Unfortunately there is no known accurate reading. Being able to see the core temps is just another tool for us to use. If you use the reading from MBM you should be fine. That's what we have used all along. Even if the reading isn't correct we are familiar with what the relative values should be. Read through this thread for an insight (or confusion) to the core temperature readings.

"Core Temp" - New way to monitor temps on K8s and Core/2 CPUs

 

Excellent, thanks for the link, I was planning on spending some time digging up more info on this and now I can spend most of that just reading.

 

I agree regarding relative temps. It's always been so, inaccurate temp sensors, but we manage simply by watching the rise, and putting a limit on the values we do receive.

 

It helps to know what the actual temp is via a thermistor, although even that is squewed because it can only be mounted near or along side of the IHS. It still verifies, one way or another, a more accuratre reading and lets you know roughly how much to add or subtract.

 

Which reminds me, since I updated the BIOS I see there is a setting for plus or minus temperatures on the CPU, it's in the system health part with all the fan speed control.

 

Is this setting for calibrating the main CPU sensor for like MBM readings?

 

I assume it is, but could be completely wrong.

 

Either way, when I get to putting the extra cooling back in the system, and plug in my FPD, I'll be attaching one of the thermistors to the cpu for a reading, allowing a better guestimate. If this BIOS setting is for calibration, this is a very nice addition to the BIOS indeed, and while not exactly long overdue, it is very welcomed by me personally!

 

.

[Praz]

Which reminds me, since I updated the BIOS I see there is a setting for plus or minus temperatures on the CPU, it's in the system health part with all the fan speed control.

 

Is this setting for calibrating the main CPU sensor for like MBM readings?

That's precisely what it's for. But it is of limited value since there is no really good method to use to calibrate against.

[boo_masters]

edit: wrong post, sorry :-}

[Guest LithoTech]

That's precisely what it's for. But it is of limited value since there is no really good method to use to calibrate against.

 

It's things like this that get me kicking myself for not buying that infra red thermometer I saw for like half price a few months ago. Point and shoot. Would be very handy if you could get the right angle at the cpu with it.

[scar69]

Hello everyone, I am in need of some advice. I just bought the ram in my sig a few days ago hoping that I would get lucky and be able to hit 265 - 275 to go along with my sweet cpu which maxes out at 2850mhz. So far I am quite disappointed with the ram. No matter what I try I can't get it memtest stable at 260. The highest I can get is 257 which is not much of an overclock considering the speeds that other guys here are getting with this stuff and I was even able to get 260 with my OCZ XTC Plat 4000. I just found some very expensive Corsair 4400 DDR550 memory on a site but would like to run a few questions by you guys before I buy it.

 

1 - Since the corsair is rated to run at 275 how much of a performance increase would I see if I was to run at 275x10 with the ram at 1/1?

 

2 - Would it be better to run a higher fsb and a lower multiplier with a divider?

 

3 - In your opinion would the performance increase justify the $410 cdn price tag?

 

4 - In most of the reviews I've read about this corsair kit they used DFI mobos for testing. Has anybody had compatability issues using corsair ram with DFI mobos?

 

5 - If I do decide to go ahead and get the corsair I will be selling the g-skill and my OCZ XTC's on Ebay. In your opinion how much should I try to get for them?

 

Thanks in advance for your help. Looking forward to reading your opinions.

[Praz]

Using apps and games you are going to notice no difference. I just posted an example of this the other day at the speeds you are looking at. Look at this post.

[gord27]

i'm at 280x10 with the 180 divider it puts my ram at 254. just bump the htt from there til you're at 257 on the ram and be happy. should be pretty darn close to your max. there's certainly no need to be on the 9x multi...

[ooztuncer]

I am in the bandwagon of X2 3800, again. After the 4th boot with my new toy, I remember in pain that why I sold my X2 3800 6 months ago. Frakkin thing needed 1.55 volts to hit 2.5ghz... And it seems that this one also in need of juice and more juice I gave her 1.40 volts to hit 2.3 ghz for a successful 12hrs orthos run.

 

It is an ADBHE 0521 TPMW (stepping 1, E4 manchester) and I believe it's an engineering sample - anyone have any experience?

 

I checked the OCDB and found out two or three ealy models but all of them needed 1.5+ volts to even reach 2.6 or 2.7ish...

 

Know anybody who can fix the mach1 in east coast? basically it needs a new suction line and regassing...

[ooztuncer]

I think I will stay here... It seems that I need to give 1.6+ volts to hit even 2.6ghz.

 

Current settings:

vcore: 1.425x1.04 = 1.48

vdimm: 2.6

ht=2000 (4x250x2)

10x250 1:1

load temps are 32ish under mbm BUT 45ish under A64 info tool (which I believe to be more accurate)