OMG!!! You guys got to see this!!!!
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
-
- *Lifetime Patron*
- Posts: 1465
- Joined: Sun Mar 09, 2003 12:27 pm
- Location: Reading.England.EU
First (?) mention was in this thread. More recent discussion was here. Think I have seen it at another UK supplier as well - perhaps watercoolingshop.com?
i'm surprised zalman isn't a sponsor of SPCR yet.
yeh this was announced in the watercooling forum before, problem is is that recent posts in the watercooling forum aren't shown in that "recent posts" bar on the right of the main SPCR page. so it's hard to see new stuff without checking manually. i emailed mikec about this, not sure if he's fixed it yet ..
yeh this was announced in the watercooling forum before, problem is is that recent posts in the watercooling forum aren't shown in that "recent posts" bar on the right of the main SPCR page. so it's hard to see new stuff without checking manually. i emailed mikec about this, not sure if he's fixed it yet ..
Okay enough lurking, time to join the fun
Nice post even though it's not news.
If this thing is capable of cooling my Prescott and a 6800 Ultra I'm getting it! Anyone knows how many watts it's capable of moving?
No more struggling with fan's
Also how quiet is that pump? Cause the terms, inaudible, silent and quiet are being thrown around here loosely sometimes.
Nice post even though it's not news.
If this thing is capable of cooling my Prescott and a 6800 Ultra I'm getting it! Anyone knows how many watts it's capable of moving?
No more struggling with fan's
Also how quiet is that pump? Cause the terms, inaudible, silent and quiet are being thrown around here loosely sometimes.
welcome to SPCR, Leto!
i would guess that it would not be sufficient to cool both a prescott and a 6800 ultra. it might just barely make it but you have 2 VERY hot things on a single watercooling loop so your performance wouldn't be much better than aircooling. the reserator is designed for silence and ease of use, not performance.
i would guess that it would not be sufficient to cool both a prescott and a 6800 ultra. it might just barely make it but you have 2 VERY hot things on a single watercooling loop so your performance wouldn't be much better than aircooling. the reserator is designed for silence and ease of use, not performance.
-
- Posts: 121
- Joined: Sun Mar 14, 2004 6:59 am
-
- SPCR Reviewer
- Posts: 2696
- Joined: Thu Dec 11, 2003 9:53 pm
- Location: Scarsdale, NY
- Contact:
Well, what I mean is you're the only willing to/able to run it at 80C!
Btw I would think that by that temperature, the system is downthrottling your chip just to save itself, so you may be losing performance just by letting it go that high. 80C is literally scorching; please don't allow your hand to come in contact with that heatsink after running it without allowing it to cool down for a good few minutes! We don't want to see you get burned. Blowing on it with compressed air once it's shut down might help.
-Ed
Btw I would think that by that temperature, the system is downthrottling your chip just to save itself, so you may be losing performance just by letting it go that high. 80C is literally scorching; please don't allow your hand to come in contact with that heatsink after running it without allowing it to cool down for a good few minutes! We don't want to see you get burned. Blowing on it with compressed air once it's shut down might help.
-Ed
I'd not worry OVERLY about it. I'm pretty familiar with the mechanics of thermal throttling.idealcrash wrote:Is that a safe temperature for a Prescott?Leto wrote:As long as it's better than my current 80 C load temperatures.
Thermal throttling would kick in, to cut down on heat generated, yes. I am wondering if it doesn't already. I'd - in fact - suggest maybe powering up your cooling on the CPU. 80 degrees is NOT good.
While Intel recommends 70 degrees as a max, and it's "limits" are usually quite conservative, I'd still be very loathe to have the CPU running that high.
Why?
Well, think about a car engine, rated at XXX <some arbitrary number> power. If you run the engine at 100% all the time, you're going through wear & tear a lot faster. That's why the "fat" Jag's, BMW's and so on have got such massive engines.
You run the engine at a lot LESS than 100%, and therefor wear & tear on the engine is reduced.
One can use the similee in regards to CPU's and temperature.
BTW, if your system powers down for "no reason", that could also be thermal throtlling. Thermal throtlling will continue to a certain point. If that point is passed, the system SHUTS DOWN. Period.
This is all decided on circuitry on the CPU, so the motherboard doesn't get a say. All in the interest of keeping the CPU from overheating.
However, OVERHEATING doesn't mean that the damage doesn't get damaged from running for extended periods at pretty high temperatures.
I'd personally prefer to run no higher than maybe 55 degres C. Certainly, going above 60 degress, you should look into your cooling, in my opinion. My opinion stands, by the way, for both - AMD and Intel CPU's.
Keep the CPU's cool, keep your system cool.
As an FYI, I am building a 3.4 GHz Prescott system (200 Mhz more, ca. 20W more heat to deal with). I'll let you know how cooling THAT goes, once it's assembled.
But 80 degres C is "not good". Not good at all.
Hope this helps.
-
- Patron of SPCR
- Posts: 946
- Joined: Sat Sep 06, 2003 9:57 pm
- Location: Berkeley, CA, USA
- Contact:
Testing Thermal Throttling in Pentium 4 CPUs with Northwood and Prescott cores
According to this article, throttling does appear to begin at 80c, so I would define that as too hot.
You guys baby your CPUs too much! That said, I agree that 80c load is the limit of what I would personally be comfortable with (and can be the start of throttle per the linked article).
There's a difference, though, between machines that are mostly idle all the time and may occasionally peak at a high temp, and those that are under load all the time (eg, folding).
For machines uner load most of the time, I wouldn't want to see much higher than 70c. That's just too much heat radiating inside the case on a 24/7 basis.
For machines idle most of the time, peaks of 75c or 80c are fine, as long as A) these really are the true highest peaks in hot summertime conditions and B) the temp stabilizes to a lower value relatively quickly when the machine becomes idle again (eg, case airflow). This may sound too hot, but it tends to happen when you use really low airflow over the heatsink. You can get away with this in a "mostly idle" machine without any problems.
According to this article, throttling does appear to begin at 80c, so I would define that as too hot.
OUCH. You people buying Prescotts are friggin' insane. No performance benefit, massive increase in heat!Now let's compare (ed note: 3.2 ghz chips): Northwood has 34°C at idle against 50°C at a 100 percent loading, while Prescott has 50°C against 76°C. The difference is not in favour of Prescott.
I disagree. There is no evidence to support heat danger for CPUs until you get to ~80c. Anything under that, as long as it is stable and you are really talking about 100% load (eg max) temps., is perfectly safe.I'd personally prefer to run no higher than maybe 55 degres C. Certainly, going above 60 degress, you should look into your cooling, in my opinion. My opinion stands, by the way, for both - AMD and Intel CPU's.
You guys baby your CPUs too much! That said, I agree that 80c load is the limit of what I would personally be comfortable with (and can be the start of throttle per the linked article).
There's a difference, though, between machines that are mostly idle all the time and may occasionally peak at a high temp, and those that are under load all the time (eg, folding).
For machines uner load most of the time, I wouldn't want to see much higher than 70c. That's just too much heat radiating inside the case on a 24/7 basis.
For machines idle most of the time, peaks of 75c or 80c are fine, as long as A) these really are the true highest peaks in hot summertime conditions and B) the temp stabilizes to a lower value relatively quickly when the machine becomes idle again (eg, case airflow). This may sound too hot, but it tends to happen when you use really low airflow over the heatsink. You can get away with this in a "mostly idle" machine without any problems.
Ooohhh - that's a "dangerous" thing to say (thinking of flames here).wumpus wrote:OUCH. You people buying Prescotts are friggin' insane. No performance benefit, massive increase in heat!
I'll not comment on it specifically so as not to start an unhealthy argument. Suffice to say that there's enough reason to get these CPU's. And - in the end - these are personal ones, no?
I somewhat agree, and disagree.wumpus wrote: You guys baby your CPUs too much! That said, I agree that 80c load is the limit of what I would personally be comfortable with (and can be the start of throttle per the linked article).
There's a difference, though, between machines that are mostly idle all the time and may occasionally peak at a high temp, and those that are under load all the time (eg, folding).
For machines uner load most of the time, I wouldn't want to see much higher than 70c. That's just too much heat radiating inside the case on a 24/7 basis.
For machines idle most of the time, peaks of 75c or 80c are fine, as long as A) these really are the true highest peaks in hot summertime conditions and B) the temp stabilizes to a lower value relatively quickly when the machine becomes idle again (eg, case airflow). This may sound too hot, but it tends to happen when you use really low airflow over the heatsink. You can get away with this in a "mostly idle" machine without any problems.
While "spikes" of temperature going up are certainly normal, I'd rather have them that much lower. It simply has to do with survivability. Treat your CPU good, and it'll treat you good.
While my approach and whoever else you believe is "you" in "babying your CPU's" too much, is admiitedly conservative, it's also safer. I don't overclock.
That's a personal choice, and much can be discussed about it.
There's nothing wrong with wanting to keep your CPU cool. Even if it goes to such extents that you call it "babying". It's - after all - in the best interest of any owner. No harm has come from having a CPU run cold, has it?
Whereas the amount of dead CPU's caused by overheating ...?
I personally say that there's more duff CPU's due to over-volting, rather than straight overheating, I've never bothered to count / keep numbers. So it's a subjective view - flawed, and I admit that.
If you don't mind your systems running that hot (and in my opinion, 70 degrees *IS* hot), then that's your call. When it comes to hardware, and - thus - my own spent moolah, I'd rather make sure that I don't fry it .
In THIS regard, I see no fault in being a little conservative . A personal choice - which may not be yours - but valid allthesame.
- Shathal.
-
- Posts: 90
- Joined: Mon Feb 09, 2004 4:16 pm
- Location: Greece
I'll sound crazy at first but here goes , I NEVER want to see a cpu temperature over 60 (or 55 for that matter).
Why? Cause motherboard temperature readings are full of ****.
Lets take the prescott for example.
60°C in an Abit motherboard = 50°C in an ASUS motherboard....
So while 80°C in an Abit = very bad , 80°C in an ASUS = disastrous.
If you look around , most of the users that have no problems with Prescott , are those with ASUS motherboards , which severely UNDER-REPORT cpu temperatures...
Things aren't any better in the AMD side of the fence , you might as well ask a fortune teller about your cpu temperature...
With all that in mind , I'd rather have a pretty large "safety" threshhold.
Why? Cause motherboard temperature readings are full of ****.
Lets take the prescott for example.
60°C in an Abit motherboard = 50°C in an ASUS motherboard....
So while 80°C in an Abit = very bad , 80°C in an ASUS = disastrous.
If you look around , most of the users that have no problems with Prescott , are those with ASUS motherboards , which severely UNDER-REPORT cpu temperatures...
Things aren't any better in the AMD side of the fence , you might as well ask a fortune teller about your cpu temperature...
With all that in mind , I'd rather have a pretty large "safety" threshhold.
Not crazy at all.
I am also assuming we're talking CPU temperature, rather than AMBIENT. Ambient is more of a "guideline", as the actualy temp on the CPU is *LOTS* higher.
It's yet another reason why "cool is good", as a CPU will spike pretty often , even just during POST or somesuch. The cooler the entire place runs, the better these things can be tolerated .
CPU temperature spikes are pretty normal & commonplace during operation.
I am also assuming we're talking CPU temperature, rather than AMBIENT. Ambient is more of a "guideline", as the actualy temp on the CPU is *LOTS* higher.
It's yet another reason why "cool is good", as a CPU will spike pretty often , even just during POST or somesuch. The cooler the entire place runs, the better these things can be tolerated .
CPU temperature spikes are pretty normal & commonplace during operation.
-
- Patron of SPCR
- Posts: 946
- Joined: Sat Sep 06, 2003 9:57 pm
- Location: Berkeley, CA, USA
- Contact:
I guess that depends if you consider unnecessary fan/air noise as "harm" or not. It's your CPU, do what you like, but by saying..No harm has come from having a CPU run cold, has it
You're choosing to subject yourself to unnecessary noise levels by keeping the CPU far cooler than is necessary-- as specified by the real-world thermal throttling tests done by Digit-Life, as well as the manufacturer's specs. There is no evidence supporting this "must run <55c" rule, other than "because that's how I want it".NEVER want to see a cpu temperature over 60 (or 55 for that matter)
Which is fine, but don't ever try to sell this as a safety measure. There's no basis in fact for that.
-
- SPCR Reviewer
- Posts: 2696
- Joined: Thu Dec 11, 2003 9:53 pm
- Location: Scarsdale, NY
- Contact:
However, it's pretty concrete if your overclocked CPU runs Prime95 for 3 days straight when kept under 60C but won't go for 10 minutes without a lockup at 65C.
I do, however, agree with you that if it runs fine, it runs fine. I mean running your CPU at 65C instead of 45C, the life of your CPU may not be as short, but let's think about this for a minute...
Are you really going to still be using that CPU in 8 years? Would it really kill you if it only lasted 4 years (half the life span of keeping it cooler)?
-Ed
I do, however, agree with you that if it runs fine, it runs fine. I mean running your CPU at 65C instead of 45C, the life of your CPU may not be as short, but let's think about this for a minute...
Are you really going to still be using that CPU in 8 years? Would it really kill you if it only lasted 4 years (half the life span of keeping it cooler)?
-Ed
-
- Patron of SPCR
- Posts: 946
- Joined: Sat Sep 06, 2003 9:57 pm
- Location: Berkeley, CA, USA
- Contact:
Just as a point of reference, the non-overclocked Barton 2500+* in my HTPC can run stably up to around 88c.My overclocked CPUs aren't stable past 60C, for the most part, and definitely can't run them properly past 70C.
Not that I WANT to run it at those temps, mind you, but I had a severe case airflow problem (see this thread) and in the process of testing solutions, it was trivially easy to get it up to that level with a closed case and Prime95-- I always shut the system down once I got over 85c. The highest temp I saw reported was 88c, and Prime95 never once complained even running at 80c for over an hour.
The Mk I finger definitely verified that 80c is VERY VERY hot, just by touching the heatsink.
Anyway, point being, non overclocked chips should run fine up to at least 80c as a peak-- as delivered and tested from the factory.
* It does run fine as a 3200+ using default voltage, but my HTPC doesn't have 200mhz memory.
-
- SPCR Reviewer
- Posts: 2696
- Joined: Thu Dec 11, 2003 9:53 pm
- Location: Scarsdale, NY
- Contact:
Speaking of toasty Bartons...
I once mentioned elsewhere in the forums my first overclocked Barton, whose Aeroflow fan stopped one night when the fan controller blew...
I found it the next morning and the Radio Shack wire loom I so handily wrapped the TMD fan's power wires in had partially melted to the sink, and when I took the sink off, there was scorch marks around the CPU core on the packaging, and also on the underside of the packaging...
...best part of it all? That chip ran just fine for several months more!
-Ed
I once mentioned elsewhere in the forums my first overclocked Barton, whose Aeroflow fan stopped one night when the fan controller blew...
I found it the next morning and the Radio Shack wire loom I so handily wrapped the TMD fan's power wires in had partially melted to the sink, and when I took the sink off, there was scorch marks around the CPU core on the packaging, and also on the underside of the packaging...
...best part of it all? That chip ran just fine for several months more!
-Ed
-
- Posts: 90
- Joined: Mon Feb 09, 2004 4:16 pm
- Location: Greece
I don't question the fact that cpus can run at well over 80°C , it's true.
What I question , is the ability of motherboards to give us proper information.
When the REAL cpu temperature is 80°C , some motherboards will show 80 , some will show 90 and some will show 60.....
You can't really know how wrong your motherboard is (external sensors have questionable accuracy).
Also heat isn't evenly distributed over the cpu , sensors will measure either temperature in a given point of a cpu or average temperature (externals mostly do that).
So even if your motherboard measures correctly , more than likely , there are other parts of the cpu (usually named "hot spots") which will be extremely hotter (easily 20-30 degrees hotter).
What I question , is the ability of motherboards to give us proper information.
When the REAL cpu temperature is 80°C , some motherboards will show 80 , some will show 90 and some will show 60.....
You can't really know how wrong your motherboard is (external sensors have questionable accuracy).
Also heat isn't evenly distributed over the cpu , sensors will measure either temperature in a given point of a cpu or average temperature (externals mostly do that).
So even if your motherboard measures correctly , more than likely , there are other parts of the cpu (usually named "hot spots") which will be extremely hotter (easily 20-30 degrees hotter).
Very few modern Athlon mobo's read from the diode.
In theory the diode temp is as accurate as you're likely to get. the diode itself is accurate to 0.5° The trouble is getting the reading from the diode, through the mobo's sensor, and out to a piece of software to report the temp. Some motherboards that do report the diode (like a few of the Asus') have their sensor chip deliberately skewed to report inaccurate readings.
The only way to really know what the diode is reporting is to read directly from it with a sensor that is known to be accurate.
In theory the diode temp is as accurate as you're likely to get. the diode itself is accurate to 0.5° The trouble is getting the reading from the diode, through the mobo's sensor, and out to a piece of software to report the temp. Some motherboards that do report the diode (like a few of the Asus') have their sensor chip deliberately skewed to report inaccurate readings.
The only way to really know what the diode is reporting is to read directly from it with a sensor that is known to be accurate.