thermal throttle on a P4A? how to test it?
- Thread starter THUGSROOK
- Start date
Thermal Throttle is when your internal temperature diole inside the cpu sense a too high of a temperature when you overclocked the cpu and it will automatic down clock the cpu back so to lower the temperature...
when your cpu gets to a certain temp - the moboard will downclock it so it doesnt overheat.
if temp still doesnt come down, it will downclock it more until the temp drops.
the wcpuid real time clock checker will make noticing downclocking very easy.
i had prime95, asus pcprobe, and wcpuid clock checker running all at the same time for an hour.
max cpu temp was only 43*C and it did not throttle at once.
if temp still doesnt come down, it will downclock it more until the temp drops.
the wcpuid real time clock checker will make noticing downclocking very easy.
i had prime95, asus pcprobe, and wcpuid clock checker running all at the same time for an hour.
max cpu temp was only 43*C and it did not throttle at once.
i've had mine running Prime95 for 18+ hours now. it keeps going between 2399.99 and 2400.00. Is that just normal?
Ok i am home now in front of my computer...i am going to try this...i think i should just keep wcpuid on 247...cause before i went to work today i had prime running...and when i came home...it was still running without an error...CPU temp went up to 52C...10 hours and 39 minutes...I think it passed...
Ran Prime95, Asus Probe 2.16.03, and Wcupid
No beeping...Max CPU temp ~50C with stock HSF
No thermal throttling
Seems to me that Asus Probe and MBM is the same and both accurate
Voltage ~1.75v
No beeping...Max CPU temp ~50C with stock HSF
No thermal throttling
Seems to me that Asus Probe and MBM is the same and both accurate
Voltage ~1.75v
No thermal throttling for me either. Ran Prime95 reached max temp. 54 as always and cpuspeed stuck @1949Mhz.
BTW has anyone ever suffered from throttling?
BTW has anyone ever suffered from throttling?
dbal eventually someone will post here that they are throttling.
this test is good piece of mind tho
this test is good piece of mind tho
I don't know that this is true. I would have to do some research to be certain, but I am pretty sure that throttling does not change the clock frequency - like Speedstep - but instead reduces the amount of activity on the core.
Having seen a couple of systems throttle here at work (someone turned on the heat over the winter in one of our temporary server rooms), I can say that the effect is definitely noticeable. Performance drops like a rock. I was just trying to "rlogin" (a Unix command that allows remote access from another computer) into a Linux box that was throttling and the lag just to type commands was substantial. I don't think that you can miss the effect when it happens... it's not subtle.
Having seen a couple of systems throttle here at work (someone turned on the heat over the winter in one of our temporary server rooms), I can say that the effect is definitely noticeable. Performance drops like a rock. I was just trying to "rlogin" (a Unix command that allows remote access from another computer) into a Linux box that was throttling and the lag just to type commands was substantial. I don't think that you can miss the effect when it happens... it's not subtle.
eventually someone will be throttling during this test, and well know for sure
we arent looking for serious throttling here - im sure wed be able to tell that just from using the machine.
its the mild throttling during gaming or benchmarking that we want to avoid.
we arent looking for serious throttling here - im sure wed be able to tell that just from using the machine.
its the mild throttling during gaming or benchmarking that we want to avoid.
<< eventually someone will be throttling during this test, and well know for sure
>>Thugs, how does that work? If I'm correct in saying that throttling doesn't affect core clock speed - something that I am pretty sure is correct - then no one will ever post that there is a reduction their core clock speed using the real time checker due to throttling. So how does that prove anything?
Patrick Mahoney
Microprocessor Design Engineer
Intel Corp.
<< but I am pretty sure that throttling does not change the clock frequency - like Speedstep - but instead reduces the amount of activity on the core. >>
I definitely think we have another serious topic to discuss here and I am very glad about it...!
Ok, well, my mistake. I called a friend who was a Northwood designer and he told me that I am wrong. The core clock does get reduced. I had thought that it was something like introducing NOPs or actually doing a HLT, but I'm wrong and the frequency is actually reduced.
But, he added, he thinks it's unlikely that anyone would be able to monitor throttling in this way. The effect is usually on the order of milliseconds, and he thinks it would be hard for a real-time clock checker to catch. His last comment was that in a well designed system throttling should never occur.
As far as better ideas. In light of my mistake, I hereby volunteer to unplug the fan on my Pentium 4 system at home tonight while running the Prime95 benchmark, a temperature monitoring program, and the real time clock program and post the results. If the effect of unplugging the fan doesn't work, then I'll hit the heatsink with a blow dryer until I see something.
All in the name of science.
But, he added, he thinks it's unlikely that anyone would be able to monitor throttling in this way. The effect is usually on the order of milliseconds, and he thinks it would be hard for a real-time clock checker to catch. His last comment was that in a well designed system throttling should never occur.
As far as better ideas. In light of my mistake, I hereby volunteer to unplug the fan on my Pentium 4 system at home tonight while running the Prime95 benchmark, a temperature monitoring program, and the real time clock program and post the results. If the effect of unplugging the fan doesn't work, then I'll hit the heatsink with a blow dryer until I see something.
All in the name of science.
me too
ive been tryin to get ppl to talk about this for 5 months!
at roughly what CPU temp does throttling activate on a Northwood?
if its 67*C like some ppl are saying - then i dont think we have anything to worry about
im topping out at 47*C max.
ive been tryin to get ppl to talk about this for 5 months!
at roughly what CPU temp does throttling activate on a Northwood?
if its 67*C like some ppl are saying - then i dont think we have anything to worry about
im topping out at 47*C max.
pm kewl - let us know what you come up with ....or any other info!
much appreciated
i would think that if running prime95 to the point of thottling you would be able to see something on WCPUid. no?
....or any other info!much appreciated
i would think that if running prime95 to the point of thottling you would be able to see something on WCPUid. no?
<< I would think that if running prime95 to the point of thottling you would be able to see something on WCPUid. no? >>
I think the answer is yes but I strongly believe that we are not able to reach the point of throttling (especially if it's over 65C) not only with Prime95 but with any other application in the world...!
Prime95 brings the cpu temp at its highest (i.e. 54C for me) fast and then the system keeps working there for ages.....i dont think WE are getting up high enough to worry about it.
however ...the throttle temp sensor is not the same sensor as the cpu temp sensor
either way - id like to see alot more info on all of this.
however ...the throttle temp sensor is not the same sensor as the cpu temp sensor
either way - id like to see alot more info on all of this.
Ok. Well, I wouldn't recommend others trying this one at home simply because it made me nervous putting my box through the fire.
But I did it.
Setup: 2GHz Pentium 4 running Windows XP. Apps running: MSI Temperature Monitor, Prime95, Seti@Home, CPUBurn, WCPUID w/ real-time clock
Experiment: I pulled off the fan plug on the CPU while running the above.
Results: At first the temperature climb was fairly gradual from my typical 48C to 52C and gradually upwards until I hit around 65C or so. Then I stopped. At this point, I was running Seti and Prime95. I downloaded CPUBurn and loaded it up and sure enough, oldfart knows what he says when he says this kicks up the temps. They climbed past the 65-66C plateau and headed up into the 70's. Things started to freeze up but I later discovered that this was an effect of CPUBurn. Around 82-84C, however, the effect was definitely noticeable. I noticed things slow down a lot - from my mouse, to network traffic, to simply closing windows. EVerything started to move like molasses. The CPU speed in WCPUID displayed 2012MHz throughout but it wasn't updating very frequently. It barely budged at all and certainly didn't drop by a substantial amount. On the other hand, I did run a few benchmarks - like the Prime95 benchmark and the results were massively lower than normal. At this point, I declared the experiment done and I put the fan power plug back in. Temps rapidly dropped down to my usual 48C typical temperature under heavy load.
I hope that at least partially answers a few questions. I'd be willing to repeat it one more time if there's something that someone wants me to do (after that you'll need to find another Guinea pig My computer will have suffered enough).
On the other hand one positive aspect of this is that I discovered that my chipset fan hadn't been plugged in for a long time now. So things are definitely cooler now than before.
But I did it.
Setup: 2GHz Pentium 4 running Windows XP. Apps running: MSI Temperature Monitor, Prime95, Seti@Home, CPUBurn, WCPUID w/ real-time clock
Experiment: I pulled off the fan plug on the CPU while running the above.
Results: At first the temperature climb was fairly gradual from my typical 48C to 52C and gradually upwards until I hit around 65C or so. Then I stopped. At this point, I was running Seti and Prime95. I downloaded CPUBurn and loaded it up and sure enough, oldfart knows what he says when he says this kicks up the temps. They climbed past the 65-66C plateau and headed up into the 70's. Things started to freeze up but I later discovered that this was an effect of CPUBurn. Around 82-84C, however, the effect was definitely noticeable. I noticed things slow down a lot - from my mouse, to network traffic, to simply closing windows. EVerything started to move like molasses. The CPU speed in WCPUID displayed 2012MHz throughout but it wasn't updating very frequently. It barely budged at all and certainly didn't drop by a substantial amount. On the other hand, I did run a few benchmarks - like the Prime95 benchmark and the results were massively lower than normal. At this point, I declared the experiment done and I put the fan power plug back in. Temps rapidly dropped down to my usual 48C typical temperature under heavy load.
I hope that at least partially answers a few questions. I'd be willing to repeat it one more time if there's something that someone wants me to do (after that you'll need to find another Guinea pig
My computer will have suffered enough).On the other hand one positive aspect of this is that I discovered that my chipset fan hadn't been plugged in for a long time now. So things are definitely cooler now than before.
awww - wcpuid didnt work
could you tell where exactly it started to throttle?
no need to retest - its been thru enough
thx alot too!
could you tell where exactly it started to throttle?
no need to retest - its been thru enough
thx alot too!
It seemed to be in the mid 80's. But everything was so slow that my temperature monitor wasn't updating often either. It would show 82C then 85C then back down to 82C again.
is it safe to say we can run 60*C and not throttle? (assuming our temps readings are accurate)
thx again - your cpu took one for the cause
thx again - your cpu took one for the cause
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 24K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 19K
-
-
-
Facebook
Twitter