What?
I have no idea on what that has to do with anything. But that's ok. Thanks anyway for sharing your... hmm... thoughts.
What?
I have no idea on what that has to do with anything. But that's ok. Thanks anyway for sharing your... hmm... thoughts.
Last edited by Mario F.; 03-24-2011 at 12:56 PM.
Originally Posted by brewbuck:
Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.
You said you might want to play with it... I was telling you that it's part of the SMBuss and there are windows API calls to access it via an analog to digital converter. The remainder of the discussion touched on the difficulties you would encounter with non-linearity in the thermal diodes...
I suspect this was exactly what my board or CPU was doing due to heat thus causing the stutters in various games. The throttling most likely saved my CPU and possibly my board.it triggers the thermal throttling...
Believe it or not my case does not have any front ventilation in it. I have several quick tool-less quick install hard drive bays there. AFAIK I cannot put a fan in there as I have in my other cases. I will check on it though b/c in the past I've always opted for a fan in the bottom front of the case b/c it does cool the case much better.
Undoubtedly... I'd still be curious to know if the old heat sink was poorly seated or not.Originally Posted by VirtualAce
A case without front ventilation is a heat problem waiting to happen... might even explain some of the troubles you had. Cases are cheap... might be a good idea to re-package...Believe it or not my case does not have any front ventilation in it. I have several quick tool-less quick install hard drive bays there. AFAIK I cannot put a fan in there as I have in my other cases. I will check on it though b/c in the past I've always opted for a fan in the bottom front of the case b/c it does cool the case much better.
The bottom front fan is sometimes helpful but unless it's got a real clear path to the exhaust fans on the back, all it really does is make noise... Without a clear path, they don't actually move any more air than would come through from natural pressure equalization. Don't forget that things behave quite differently with the case open than when it's closed... you might feel a strong airflow with the side off but when you button it back up that airflow might (probably will) reduce sharply.
Warm (pun not intended). The throttling starts to happen close to the thermal maximum in an attempt to avoid the processor reaching it. Somewhere within 10 degrees of TJMax for pretty much all processors, I believe. But what will actually save your processor is #THERMTRIP. If the throttling doesn't work and your processor keeps heating up, it will reach TJMax. Then, on AMD processors (Intel works differently) it will immediately activate the #THERMTRIP and your system will shutdown. No warning, no nothing. These were almost certainly the shutdowns you were experiencing before.
And absolutely right about those stutters. Knowing what we know now, no doubt it was AMD's Cool'n'Quiet doing its thing.
EDIT: Unfortunately since you already experienced some thermtrips, depending on how long this was going on and how frequent they have become, I suggest you start thinking on replacing your processor. You are in no rush probably, but chances are its life has been shortened already.
That's how I have mine too. It's a good option. On my case, it blows directly to the HD (which I placed on a slot that takes the bottom half of the fan) and into the case. Keeping the HD cool is just another way to help keeping the whole temperature down inside the case.I will check on it though b/c in the past I've always opted for a fan in the bottom front of the case b/c it does cool the case much better.
Last edited by Mario F.; 03-24-2011 at 08:55 PM.
Originally Posted by brewbuck:
Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.
Ummm... doesn't this rather contradict your previous claim that this is a process we can trust to safeguard our hardware?EDIT: Unfortunately since you already experienced some thermtrips, depending on how long this was going on and how frequent they have become, I suggest you start thinking on replacing your processor. You are in no rush probably, but chances are its life has been shortened already.
No.
I seem to remember a VirtualAce post maybe 1 or 2 months ago where he's discussing exactly the issue he was having with his computer shutting down unexpectedly. And that it had been going for a while already. Assuming maybe 3 months of this behavior and these shutdowns having been frequent enough, no one in their right mind can't claim his processor life may have not shortened. It's only sensible to assume that under extreme conditions, as it seems to have been the case, all bets are off.
This in no way contradicts my arguments on that thread you speak of.
Originally Posted by brewbuck:
Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.
It's actually a well-known fact among overclockers that cpu heat-spreaders are next to useless as-is. There are very significant improvements to be had by sanding them down to be perfectly flat, some even go to the length of sanding them to a perfect mirror finish using fine grain sanding paper to incrase contact between the heatsink and the heatspreader.
How I need a drink, alcoholic in nature, after the heavy lectures involving quantum mechanics.
I've always preferred draw filing... the heat sink, not the top of a chip... NEVER the top of a chip! The technique gives a very flat surface (provided you know how to use a file) with a medium-gloss finish, patterned in fine diamond patterns. Then I'll use a space filling, phase change compound such as Thermstrate TC 175-125 for the best surface mating.
The thermstrate is especially impressive stuff. When first applied it's abuot the consistency of bar soap (actually it's champhor and silicon) then as the chip heats it softens and flows out into all the tiny crevaces producing a very uniform fill throughout the heat spreader area. After a few heat-cool cycles you'll begin to see the chip temps rolling off until it stabilizes at some lower temperature, where it will stay until disturbed.
It comes in 50 gram bars, enough for about 200 CPUs and at $16 each, that's quite the bargain.
Last edited by CommonTater; 03-25-2011 at 12:31 PM.
I've heard of these wax compounds before. But I also hear the thing that has been stopping their widespread use in computer heatsinks is their high phase change temperatures. It seems they only start to expand after a certain temperature, somewhere around 60 C. For many current cooling solutions out there (including air cooling) the processor will rarely reach that temperature, which makes this wax solution somewhat wasteful.
What's the phase change temperature of this particular model you mention?
Originally Posted by brewbuck:
Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.
It softens at 125f ( 50c ) and reaches full liquidity at 175f (79c). Hense, the catalog number.
I've used Thermstrate on about 150 CPUs by now. It's never been a problem and usually, after a settling in time, I get a 3 to 5c temperature reduction using it, as compared to the regular white pastes.
The white pastes will come in better *at first* because of the oils in them. As the oil finds it's way out of the thermal junction leaving only the powder, the paste's performance degrades over time... Most typically you will see your CPU temperatures start off very low, then they will rise over time (in days or weeks) and eventually settle at some higher temperature.
Phase change compounds do the exact opposite. They start off a bit hotter and as things settle in (over a period of days) the temperatures come down and they stay down.
Very well. I'm sure it is a good solution. But I hope you don't mind me saying though that a different of 3-5 degrees between this and other options seems a bit excessive. I've never ever heard of anything like that. If you are experiencing such a big difference, then something is definitely wrong in the way the paste is being applied. Besides everyone would be flocking to your solution like bears to a honeypot if the difference was that big. But people aren't. So...
Also note that many pastes today have lifetimes that go well beyond the processor own lifetime, although admittedly nothing lasts forever. Instead what you are seeing with grease solutions is known as break-in period. This is a period of time that is required for the compound to gain maximum thermal conductivity. In fact, temperatures tend to drop over time after applying the paste, not climb. This period is different depending on the type of material the paste is made up. The well known Artic Silver 5, for instance, has a break-in period of 200 hours. Only after this time, and after several thermal cycles, will the paste gain maximum thermal conductivity.
As for lifetime... Depends on who you are. Enthusiasts will not hesitate to tell you they will replace their thermal paste once an year or so. While most common users will have it for the entire life of their system which is usually measure in 3-5 years without any noticeable increase in temperature. Besides, a thermal paste is the least important thing in the arsenal of cooling solutions. Some people don't even use it(!), while even a dry paste retains all conductivity if it is one of those metallic or ceramic ones.
Last edited by Mario F.; 03-25-2011 at 06:41 PM.
Originally Posted by brewbuck:
Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.
Of course. You *always* know more and better than everyone else, Mario.
I've only been at this for 30+ years professionally. What could I possibly know?
Last edited by CommonTater; 03-26-2011 at 05:53 AM.
The top of the chip needs work aswell if you want it as cool as possible. Its called CPU Lapping, here's an example: Intel IHS Lapping and Removal For LGA775 CPU's - Giving Intel Core 2 Duo A New Appearance - Legit Reviews
How I need a drink, alcoholic in nature, after the heavy lectures involving quantum mechanics.
Y'know what... the nickel plating is on there for a reason... It is to avoid the galvanic reaction between the copper spreader and the aluminim heat sink. It may look real spiffy and it may even work better for a short term, but when the reaction sets in corrosion byproducts will pollute the thermal junction and it will eventually overheat wildly. The short term gain utterly defeats the long term stability of the cooling solution.
Galvanic corrosion - Wikipedia, the free encyclopedia
Also, you don't want an absolutely flat smooth surface there. The slightly dished center of the heat spreader actually helps prevent the thermal interface material from creeping out of the junction and leaving it dry, with heavily degraded thermal performance.
No matter how good the surface match of the chip and sink, even terrible interface materials like peanut butter or toothpaste will perform better than a dry thermal junction.
Thermal transfer compound comparison
The article reveals something very surprising... Vegimite sandwich spread beats arctic silver... for about 4 hours, until the vegimite dries up.
In my experience overclockers are like audiophiles... they're really not much beyond spending $120 on a power cord because they think it sounds better... The only real difference is the fads they run with and a whole lot of this thermal grease stuff they pay ridiculous prices for is no better than good old fashioned zinc oxide cream.
To give you an example... How many of these guys realize that the zinc oxide cream they buy in the computer stores at $10.00 or more for a 5g tube is actually the same stuff --chemically identical-- as the lifeguards use on their noses and buy in 150g toothpaste sized tubes for $3.99. As P.T. Barnum says: "There's a sucker born every minute."
For Marios comment about people flocking to a new product over 2 or 4 degrees. Wow, talk about a fools errand. This difference is so small as to be impossible to measure reliably. On one day you use zinc oxide and record 48c, the next day you use arctic silver and record 44c and you get all excited... except that the ambient temperature is actualy 5 degrees cooler today than it was yesterday. Have you gained 4 degrees or lost 1?
Why don't people flock to Thermstrate TC if it's so good? Well, that's largely because it's a production compound, not generally available to the public and not well known outside professional circles. Overclockers aren't using it because a) they don't know about it and b) they couldn't buy it even if they did.
http://www.dansdata.com/powerdev.htm
There are so many misunderstandings and truly bad ideas floating around out there that I really do get a good chuckle over some of these guy's antics....
Last edited by CommonTater; 03-26-2011 at 08:35 AM.
