CPU temp vs core temp

January 13, 2007 at 18:49:41
Specs: win xp pro, 512

I am currently running an MSI K8N Neo4 Ultra with a 3700+ AMD64 SD. I am running at 240x11 with RAM at 166 and cpu v@ 1.4. When I run Prime95 torture test, "Core Center", the MSI mobo oc monitoring program says the CPU temp is 42C and system temp. at 29C. Whne I look at Speedfan, it shows a core temp of 52C (with a little flame next to it)
From what I have read, 55 or less is ok for this PCU. Whish number should I go by?
Also, I can run the torture test for 1.5 hrs with 230X11 with stk voltage. I fail @ 10 minutes in when I try to go to 240X11. I would like to raise the voltage to more than 1.4v, but I am concerned with the stated core temp as listed on speedfan.
ATM I shoe 36Ccpu/ 28C sys. at idle with 240X11. I tried to do the same prime95 test with the ram speed set to 133 just to rule it out with no luck.

"So I have that going for me....Which is nice"

See More: CPU temp vs core temp

Report •

January 13, 2007 at 19:00:15

also, I can't seem to set spread spectrum to disable. I am not sure what that changes.

"So I have that going for me....Which is nice"

Report •

January 13, 2007 at 21:08:29

Here's a couple of programs you should try. The 1st is "Core Temp":


The 2nd is CPUZ:


Core Temp is pretty self explanitory. Download, unzip, & double-click on Core Temp.exe to find your Core Temp. Hopefully it will confirm that one or the other program you're using is correct. I'm betting on the MSI software.

CPUZ will give you all sorts of info about your CPU, board, & RAM. Download it, unzip, then double-click on CPUZ.exe...it may take several seconds to popup. Click on the memory tab to find out what speed your RAM is actually running at. If you click on the SPD tab, you'll find the programmed memory timings for the standard bus speeds (133, 166, 200MHz). With the CPU Freq set to 240MHz & the RAM setting at 166MHz, the memory *should* be running at 200MHz.

Report •

January 13, 2007 at 21:44:02

Spread Spectrum is a setting that is supposed to cutdown on EMI (ElectroMagnetic Interference) to conform to FCC regulations. But what it actually does is spread the EMR (ElectroMagnetic Radiation) over a wider bandwidth that's not within the range that the FCC tests for. Enabling or disabling Spread Spectrum doesn't change the amount of EMR being given off, all it does is lower the amount that's detected within the specific frequency range that's tested by the FCC.

Anyhow, enabling Spread Spectrum on a PC can have negative effects on stability & overclockability, so it's always recommended to disable it whenever possible.


Report •

Related Solutions

January 14, 2007 at 00:40:25

20 years from now someone was is going to blame a pc for giving them cancer. :)

AMD Athlon 64 X2 4600+ OC 2.7ghz
2GB Dual Channel DDR 3200
Nvidia 7900GT
SATA II 2x 300gig 7200rpm 16mb cache RAID-0+1
Gigabyte Nforce 4 SLI

Report •

January 14, 2007 at 07:00:44

I ran prime95 for 12 hrs and it was stable running at 237x11 with ram at 185. When I tried to run 240x11, I would need to raise voltage to 1.43 and I was getting temps in the low 60's (too high?)
I uninstalled the MSI Core Center and am running prime95 at 240x11 at 52C. I think the MSI program kept dropping the cpu fan speed.
What I need to find out is the max temp. my cpu core can go to safly. I downloaded coretemp (which reads 2-3 degreres higher than speedfan) and it says I have a Toledo (new SD stepping, 1 core disabled) It has a window labeled TcaseMAx, 61C. I am assumi8ng that is the rated limit for the core? I am going to see if I can get it stable at 240 with temps under 60C.

"So I have that going for me....Which is nice"

Report •

January 14, 2007 at 08:30:44

You never said, but did you finally get the BIOS updated to version 1D?

"it says I have a Toledo (new SD stepping, 1 core disabled)"

There are two different versions of the A64 3700+... the true single core San Diego, or the dual core Toldeo w/1 core disabled. I don't know if one overclocks differently than the other or not. AFAIK, newegg no longer sells the true single core version.

The TcaseMax is just what is sounds like...the max temp for the case, not the CPU. AMD isn't exactly clear on the max safe CPU temp. For your CPU, they list it as 49-71C. IMO, 49C is ridiculously low.


I would say your load temps are too high though...not dangerously so, but the A64 generally runs cooler. Did you check your temp readings in the BIOS?

I believe we have the same case...the blue Ultra Wizard mid tower from Outpost? The only case cooling fan I have is the 120mm in the rear (blowing outward), plus the PSU. With my Asus board, all temps check...the BIOS, PC Probe, & Core Temp all report the same readings. Currently, my overclocked 3000+ (2.0GHz @ 2.5GHz) is running at 29C. However, I have a 51W CPU while yours is an 89W, so it's not a fair temp comparison. 36C at idle for your CPU isn't excessive, but your load temp *may* be on the high side.

I'm guessing that you used the stock HSF w/thermal pad? You made sure it was seated properly & the securing lever was fully locked in placed?

"it was stable running at 237x11 with ram at 185"

185MHz for the RAM seems wrong. I would think with the CPU at 237MHz & the memory setting at 166MHz, the RAM would run at (or close to) 198MHz. You do have the memory setting at 166MHz?

Have you tried temporarily lowering the CPU ratio from 11x to 10x or even 9x to see if that will allow you to raise the CPU freq above 240MHz? At 10 x 250MHz w/memory at 166MHz, the CPU should run at 2.5GHz w/RAM at 208Mhz.

Report •

January 14, 2007 at 10:11:11

I just ran prime95 for 3 hours at 240x11, 1.42v. My max temp was 56C. I am thinking all the high temps I was getting were due to Core Center messing up my cpu fans rpm.
I used live update and got the 1.D version of bios and it unlocked everything except spread spectrum for the CPU. (although, I can't find the 1T/2T command setting for memory, its at 2T)
I am using the stock HSF and I am assuming it is on good judging by my new temps. The bios temp is the same as the "temp2" in speedfan. That temp barely went over 42C. I am going to try to run 245x11 to see if it will run stable at this voltage. I think I can go a bit higher on the temp. If I can keep it below 60C under full load, I think thats safe.
From what I've read about the Toledo core SD, 250X11 X4 is doable with stock heating. I am running 120mm in back going out, 120mm from PSU, 80mm in front going in and I added an extra 60mm cpu fan in the side grate (removed the baffle that sticks over the cpu) I am not sure thats going to make a big difference, but I had it laying around.
CPUZ shows my FSB:DRAM at CPU/14. I am assuming that this means my memory runs at cpu speed divided by 14 when set to 166. I don't think there are any other adjustments. Not too big a deal as I should be very close to 200 if I go to 250.
Right now at idle, I have a core temp of 35C. Speedfan shows temp2 at 34C and core temp of 36C.

"So I have that going for me....Which is nice"

Report •

January 14, 2007 at 10:30:16

I am also assuming that the only limit to voltage is heat? Is there a way I can damage the CPU with too much voltage BEFORE it overheats? I am about to try 245x11 at 1.46v. Thjis doesn't seem too high, as long as my core temps are normal.
Also, is it noremal for a BIOS to have CPU v core options that are 1.30, 1.325, 1.35, 1.375, 1.40 etc AND another option below that to adjust by percentages, +3.3%, 6.6%, 9.9% etc? I am using both to "fine tune" my voltage. The problem is, when I look into core temp etc, it will chow what I set the voltage too, but not the added percentage. If I look in cpuz, it is accurate.

"So I have that going for me....Which is nice"

Report •

January 14, 2007 at 11:10:33

Removing the side air duct & replacing it with a fan probably wasn't the greatest idea. The duct acts as a fresh air source for the CPU fan & side panel fans are usually disruptive to good front-to-rear airflow. The front case fan probably isn't necessary either. A good rear 120mm fan & a PSU with a 120mm fan on the underside should be plenty. People tend to "overkill" their case cooling setup...more is not always better.

That being said, it seems like your temps are more reasonable now that you've removed the MSI software, so it was probably just a reporting error.

You can fry a CPU by increasing the voltage too much & it has nothing to do with overheating. The A64 has built-in overheat protection similar to the P4, but it's not protected from overvolting. Keep the voltage as low as possible while still maintaining a stable system.

According to your manaul, the 1T/2T setting is in DRAM Configuration Menu. It may be hidden until you set the "Timing Mode" to Manual. What memory timings are you using?

Report •

January 14, 2007 at 11:36:01

I am using 2.5-3-3-11 as you suggested. There are more settings that I left to auto.
I just went 40 minutes at 245x11 and peak temp was 60C at 1.46v. I am going to try to bump the voltage to 1.48, but am nervous about going over 1.5.
I will try the side panel without the fan. The only reason I removed the baffle is that it is 2" higher than my cpu. I would rather have nothing protruding into the open area if it isn't drawing air directly over the cpu.
The stk voltage for this chip is 1.35. What the highest I can go?

"So I have that going for me....Which is nice"

Report •

January 14, 2007 at 11:39:19

"I am also assuming that the only limit to voltage is heat? Is there a way I can damage the CPU with too much voltage BEFORE it overheats?"

The thermal-protect mechanism in newer CPU has pretty nullified the probability of your CPU going up in flames - if you apply too much voltages, so have no fear in that regard. But despite this, the CPU might fail to boot or it boots but becomes very unstable in Windows, and this may not have a specific pattern or it may.

In other words, you could be fine browsing, doing office type stuff, but turn around and burn a CD or attempt something else that is aggresive CPU-wise and the system reboots or you get the familiar BSOD, hence the reason why the adjustments is in 0.025 increments, but you'd be pretty amazed though, at how high a temperature CPUs can still go on crunching numbers without skipping a beat. Usualy though as is the case, the numbers reported would have scared you into doing something about it, so rarely do you see this happening.

However, it has been alleged that running the CPU at temperatures that is too high over a seemingly *extensive* period of time *may* affect the CPU's longevity, since the CPU (even though stable) could become susceptible to to electromigration - but I won't so much worry about this, because almost no one holds on to a particular CPU so long for that to be a substantial consideration.

Now, despite what AMD might have disclosed as your CPUs max operating temperature, a program like CPU Burn-in will help you get fairly close to the real deal, as relative to your specific CPU since you are actually running on it.


How To Ask Questions The Smart Way

Report •

January 14, 2007 at 12:56:48

Is there another name for Command Rate? Some of the other settings I have are: Row to Row delay (trrd), Read-to-Write time (trwt), Write recovery time (twr), Refresh Period (tref). All settings that can be set to manual, are. The ones I just listed are all set to auto.

"So I have that going for me....Which is nice"

Report •

January 14, 2007 at 17:06:20

"People tend to "overkill" their case cooling setup...more is not always better."

Too many fans could disrupt the airflow current.

AMD Athlon 64 X2 4600+ OC 2.7ghz
2GB Dual Channel DDR 3200
Nvidia 7900GT
SATA II 2x 300gig 7200rpm 16mb cache RAID-0+1
Gigabyte Nforce 4 SLI

Report •

January 14, 2007 at 18:35:51

well, the best I can do is 250x10.5 with ram at 166 (201.9), htt at 4x and cpu v at 1.375 +6.6% (1.42-47)
My core temp stays below 60C (55-57) and I have been stable for over 2.5 hours. I am going to run it over night.
I may try for a bit more with an after-market cooler, but this is good enough for now. Besides, I get my BFG 7600GT tomorrow, need to get this safely dialed in.

I tried to up the cpu past 2650 with a lower multiplier, but I would get errors at 1.52v and I was running at 65C-67C. I don't want to go higher until I can get a better cooler. Good start so far. Thanks for the info :)

"So I have that going for me....Which is nice"

Report •

Ask Question