• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Asus P5G41T-M LX throttling CPU, but CPU not overheating!

Assimilator1

Elite Member
Nov 4, 1999
23,035
71
91
Whilst running OCCT CPU test (not linpack option) to test stability I noticed that the CPU was being throttled, multiplier drops to x6, & vcore drops too, FSB doesn't change, it starts after less than a minute & then goes up & down in fairly regular intervals (throttling last about 30s, full speed mostly lasts 30s, but sometimes a bit longer).

In the bios I have disabled C1E & EIST so their should be no CPU speed drop from those, aside from which OCCT runs the CPU at 100% load.
Peak CPU temperature on the hottest core is 56C (checked with speedfan & realtemp) when running OCCT upto the point the CPU is throttled, then it drops of course. This naturally is a long way from the temperature throttle point (realtemp says '44' away) employed by Intel in their Thermal Monitoring protection. Yet if I disable that option in the bios, throttling doesn't happen!

Why is it throttling when the CPU is not getting too hot?
AFAIK Intel's specs for TM only monitor CPU temps, does Asus add monitoring of other chips? VRMs???

Btw all voltages are at default (auto) bar vRAM at 1.5v. FSB 365 MHz, PCIE 106 MHz (had to increase that to get FSB over 343 MHz!). OCCT loaded vcore is 1.144-1.152v (via CPU-Z).
I have used my IR thermometer on the NB heatsink & it showed a temp of ~60C at the hottest point when running OCCT (& when not throttled). It's rather hot but didn't seem to bad to me, besides which, AFAIK TM doesn't monitor that, right?

Does not throttle running SETI, even when Milkyway@home is running on the GPU too.
Since found it does throttle running Asteroids@Home, that is better optimised & the rig draws more power & the CPU runs 3-4C hotter than Folding@H or SETI.

Any idea what's going on? (oh & yea I know this mbrd is a POS, stuck with it :p).

Googling didn't turn up much, though 1 guy had the same problem with Prime95 - http://www.tomshardware.co.uk/forum/278526-29-thermal-monitoring - he never got a proper answer.

The rigs specs.
Q9550 @3.1 GHz, 365 MHz FSB, all voltages are on auto which are the default setting (have confirmed by measuring power draw from wall), bar vRAM 1.5v. 8 GB DDR3 1333 RAM is underclocked @1168 MHz, HD 4830, 1TB HDD, Win 7 Ult 64bit.

Btw reposted here because I only got 1 reply in the mbrd section, hoping they'll be more informed people here :).
Going by the 1 reply I did get http://forums.anandtech.com/showthread.php?t=2365659 , I've since downclocked it to 2994 MHz & I still get throttling with OCCT! Though it's not as often (I can post a graph if anyone really wants it).
I've now pointed my IR thermometer on the 2 visible VRMs* (3.1 GHz, running Asteroids@home), one's on ~51C the other 48C. This seems to be roughly bourn out by touching it ;), it's hot to touch but I can keep my finger on them for a few seconds. So not terribly hot, those are ok temps aren't they?

*I believe they're the VRMs anyway, black little boxes about 1cm across, under the HSF retention bracket.

What do you folk reckon?
 
Last edited:

schmuckley

Platinum Member
Aug 18, 2011
2,335
1
0
It's the VRMs.
Point a fan on them.
Reduce your Vcore.

You're running a Yorkfield quad on a g41 mobo.
I had to reduce Vcore to 1.28 to do that;pretty sure.
Don't run linpack too much or your VRMs will melt.
 

Assimilator1

Elite Member
Nov 4, 1999
23,035
71
91
Ok thanks guys :thumbsup:
So is VRM monitoring just something the mobo manuf. add? Because I didn't read about that in Intel's TM, unless I missed it?

I suppose I'd have to anchor any HSs to the VRMs simply using glue?
Mounting a fan nr their is going to be very difficult too, no idea what I'd attach it to.......

schmuckley
As I said, vcore is at def., ~1.15v OCCT loaded, so I can't reduce it ;).
Also the mbrd is shown as supporting quads upto 3 GHz, so I didn't expect any problems there, damn POS mbrd! :p.

Looking at your pic, I was measuring the wrong chips, doh!
Going by this pic http://www.asus.com/Motherboards/P5G41TM_LX/#gallery
I've got to measure the 6 staggered chips in a similar position to the 1s in your pic then?

[update] Ok I've just done that (& measured the 2 on the left 2), their measuring 50-60C with Ast.@H running, & their quite hot to touch, I'm able to keep my finger on them for a few seconds before it's painful. Is that too hot?? I thought they were ok until 80+C?

Oh & I didn't run the linpack test at all, I just think it creates a heat level that is totally unrealistic, even compared to Asteroids@H, so I just run the normal OCCT CPU test.
 
Last edited:

Assimilator1

Elite Member
Nov 4, 1999
23,035
71
91
Just for the record, I tried stock speeds.
Incidentally power draw from the wall @ stock is down only 6-7w vs 3.1 GHz, @160-162w running OCCT CPU. So a very small difference in power usage.

I've just seen it throttle at stock speed (well 334 MHz FSB, Asus like to bump up their stock FSB). So even at 2.84 GHz it's throttling it to 2 GHz (x6 multi).

Stupid mbrd , oh btw I've also emailed Asus tech support, see if they come back with anything useful in the next few days.......
 

Assimilator1

Elite Member
Nov 4, 1999
23,035
71
91
Oh I so nearly vented my spleen when I got this on the 2nd reply! :mad:

Dear Mark,

This related to many reasons(like the quality of the CPU).
To prevent the CPU it will throttles the frequecny, this was truly an normal phenomen.
I hope you could understand.


Wtf!!??o_O, anyway I just kept my calm & wrote this (for those who care ;)).

Hi John
Sorry but to say that the CPU throttles under normal circumstances (whilst under load) is just not true!

If you meant it throttles whilst idling, yes it normally would unless the those power saving features have been disabled, which as I mentioned, they have (for testing purposes). Also I mentioned the CPU throttled under load so those saving features wouldn't kick in then anyway.

When the CPU is under load it should never throttle unless the CPU overheats, & for Core 2 quads that's roughly 90-95C, my CPU is running at about 55-58C depending which program I run, room temp & what clock speed the CPU is at. So even when it is overclocked it is not anywhere near overheating.

I have read from a few reviews over the years (primarily Anandtech) that sometimes under extreme circumstances (very high overclocks + very high vcore) the CPU is throttled by the motherboard because its VRMs have overheated. Now obviously my CPU was not overclocked very high (3.1 GHz as mentioned, only 100 MHz over max supported) & the vcore was & is at default setting, now if the throttling only occurred whilst overclocked @ 3.1 GHz I could begrudgiling accept that it is the tiny bit of extra power draw taking it beyond the mbrds limits (but seriously if a meagre 100 MHz over Asus's max claimed supported CPU speed did that I'd be complaining about poor engineering & cutting the limits far too close!).

However as I mentioned, although it occurs less, it does still throttle at the CPUs default speed of 2.83 GHz - this is not normal! If you wish me to link you to an Intel document showing when thermal monitoring is implemented then I will do so.
Unfortunately what that doc obviously won't explain is any extra monitoring deployed by the mbrd manufacturer, e.g
throttling due to VRMs etc over heating/over loading.

Now, into the realms of speculation, whilst it is quite clear that the throttling is not occuring due to CPU temps it is not clear what is causing the throttling. I speculate that whatever is used to measure the temps/load of the VRMs is mis-calibrated & is reporting back false temperatures. (If that were true then maybe a bios update could fix that).
I base this thought on the temperature readings I recorded using my IR thermometer, 50-60C, these temperatures don't seem excessive to me, but if I'm wrong then maybe heat is the problem?
What are their safe temperature limits?
Does this mbrd also throttle the CPU based on VRM temperature &/or current load?
Also I measured the temp of the northbridge heatsink (approx. 60C under load), I didn't consider this excessive, but am I wrong?
What is the safe temp for the NB?
Does the mbrd throttle the CPU on NB temperature? (that would be a new 1 to me).

I have already asked some of these questions in my 1st email & did not recieve an answer, please could you find out.
Even if you can't give me a precise answer as to what is causing the throttling some answers as to what does trigger it could help.
Further thoughts as to what might cause VRM induced throttling are 1 or more faulty VRMs, inadequate cooling on the VRMs (btw the case has 2 120mm fans so system temp shouldn't be a problem), inadequate VRMs for the CPUs that are claimed to be supported, but if you have other ideas as to what is causing the throttling then I'm all ears! :)

I look forwards to a detailed answer, although bear in mind I am not an electronic engineer (though I wanted to be when at school!)

Sincerely
Mark D

PS I have been & am a loyal Asus customer for many years, my 1st Asus mbrd was in my 1st PC a TXP4 16 years ago! I currently run 3, 1 in my sons PC discussed above & 2 in my own 2 PCs. I've also built many PCs over the years often with Asus mbds :).
 

ASK THE COMMUNITY

TRENDING THREADS