Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

J the Ninja

macrumors 68000
Original poster
Jul 14, 2008
1,824
0
Just a few I keep seeing that I wanted to clear up. If you know of any more, or see an error I made, please post. We're all here to learn.


80C is a safe operating temp since the max temp is 105C

Truth: Not quite. While its true the CPU won't outright melt/burn/whatever at 90C, it still isn't good for it. The hotter a chip runs, the sooner it dies. Don't be afraid to punch those fans up.

http://www.pcpower.com/technology/optemps/


Increasing my fan speed will kill them within months

Truth:
Well, there is a bit of truth to this one. You are definitely cutting the lifespan down. Except....you are cutting from a decade+ to only about half a dozen years. These fans are designed to be run at full power. Your machine uses something called PWM to slow them down since 6200rpm is REALLY FRAKING LOUD



Getting my machine with the 512MB 9600M GT is worthless since you can't address more than 256MB with a 128bit bus. The extra VRAM is marketing junk

Truth:
Honestly, I have no idea where this came from or what the logic behind it is, but it gets repeated to death when certain topics come up. Odd, considering these threads are often a few posts away from topics describing how the CPU can address 6GB of memory on a 128bit bus, but..........well, for starters:

http://en.wikipedia.org/wiki/128bit

128bit refers to the bandwidth. Someone perpetrating the myth once described it saying: "Imagine you have a parking lot that holds 50 cars. The road out can carry 50 cars. But, imagine you increased the lot to 100 cars. Doesn't do much good, does it?" If that were true, you'd need a 2,147,483,648bit bus for 256MB (that's how many bits are in 256MB) :O Maybe that's what the video card in the computer controlling the Matrix uses for its VRAM, but we don't really have that here. 128bit memory bus means it pushes half as much stuff per cycle as a 256bit bus. Nothing more.


DDR3 has a higher clock than DDR2, but it's actually slower, because of the latency.

Truth:
Well, yes, DDR3 does have higher latency numbers than DDR2. What is never mentioned (I never even realized this until I saw it on a BIOS screen) is that memory latency is measured in # of clocks. The higher latency is due to the clock itself being higher. If you double the clock rate, but don't change the actual latency time, the number that comes out for latency is twice as big. Do the math. 7-7-7-20@1066Mhz means LESS time waiting on the memory than 5-5-5-15@667Mhz, not the other way around. All with less voltage. Isn't technology great?
 
80C is a safe operating temp since the max temp is 105C

Truth: Not quite. While its true the CPU won't outright melt/burn/whatever at 90C, it still isn't good for it. The hotter a chip runs, the sooner it dies. Don't be afraid to punch those fans up.

http://www.pcpower.com/technology/optemps/

We do have a thermal specification of 105 deg C at 35 watt draw on these CPUs according to intel (http://processorfinder.intel.com/List.aspx?ParentRadio=All&ProcFam=2643&SearchKey=). Until you hit above 90 deg C for idle, I wouldn't worry much since you are in spec of what intel offers. These chips were designed to run "hotter" then Desktop variants due to the fact we enclose them in a sub 1 inch casing.
 
We do have a thermal specification of 105 deg C at 35 watt draw on these CPUs according to intel (http://processorfinder.intel.com/List.aspx?ParentRadio=All&ProcFam=2643&SearchKey=). Until you hit above 90 deg C for idle, I wouldn't worry much since you are in spec of what intel offers. These chips were designed to run "hotter" then Desktop variants due to the fact we enclose them in a sub 1 inch casing.

Did you bother to click that link? Because you completely missed my point, and I don't see how you could've if you'd read the link.
 
Getting my machine with the 512MB 9600M GT is worthless since you can't address more than 256MB with a 128bit bus. The extra VRAM is marketing junk

Truth:
Honestly, I have no idea where this came from or what the logic behind it is, but it gets repeated to death when certain topics come up. Odd, considering these threads are often a few posts away from topics describing how the CPU can address 6GB of memory on a 128bit bus, but..........well, for starters:

http://en.wikipedia.org/wiki/128bit

128bit refers to the bandwidth. Someone perpetrating the myth once described it saying: "Imagine you have a parking lot that holds 50 cars. The road out can carry 50 cars. But, imagine you increased the lot to 100 cars. Doesn't do much good, does it?" If that were true, you'd need a 2,147,483,648bit bus for 256MB (that's how many bits are in 256MB) :O Maybe that's what the video card in the computer controlling the Matrix uses for its VRAM, but we don't really have that here. 128bit memory bus means it pushes half as much stuff per cycle as a 256bit bus. Nothing more.

Hmm? I think the 50-car parking lot increased to 100 cars makes sense. Sure, you can fit more cars, but they still can't come and go any quicker than when there were only 50.

However, at a resolution of 1440x900 and below, yes, the memory bus is a bottleneck and the difference between 512mb and 256mb VRAM is negligible. One result here: http://enthusiast.hardocp.com/article.html?art=MTM4OSwxLCxoZW50aHVzaWFzdA== Just compare the FPS numbers for the 512 and 256MB versions. They all use a 128bit bus. Even with tons of AA enabled, the difference is 5% at most.
 
Hmm? I think the 50-car parking lot increased to 100 cars makes sense. Sure, you can fit more cars, but they still can't come and go any quicker than when there were only 50.

That was exactly the OPs point with that analogy -- that it's a silly statement to make.

Specific example: Extra VRAM can be stuffed full with higher resolution textures, so when you're gaming you can see more details when you're close up to things instead of lower-res textures zooming and blurring under your feet.
 
Just a few I keep seeing that I wanted to clear up. If you know of any more, or see an error I made, please post. We're all here to learn.


80C is a safe operating temp since the max temp is 105C

Truth: Not quite. While its true the CPU won't outright melt/burn/whatever at 90C, it still isn't good for it. The hotter a chip runs, the sooner it dies. Don't be afraid to punch those fans up.

http://www.pcpower.com/technology/optemps/


Increasing my fan speed will kill them within months

Truth:
Well, there is a bit of truth to this one. You are definitely cutting the lifespan down. Except....you are cutting from a decade+ to only about half a dozen years. These fans are designed to be run at full power. Your machine uses something called PWM to slow them down since 6200rpm is REALLY FRAKING LOUD

-----------------------------------------------------------------------------------------------------------------------------------
pcpower is run by ocz, of course they are going to try and scare non technical people into thinking that.

I maintain my own renderfarm and have for about 10 years, it has varied from 10-22 machines all running *full blow* 24/7.. in all the time I've been maintaining and building my own machines (far longer than 10 years) I've seen *one* cpu fail. I have motherboards fail, drives, ram.. never processors. I've had 2 bunk CPUS DOA. Next year I plan on replacing the whole lot with two of the new Crays.

I have fans fail so often I have a box of them and the right fan in my mbp has already failed.

I trust CPU manufacturers more than OCZ, more than you and more than almost any other component in any machine. They are built to run hot, it's their job.
 
I trust CPU manufacturers more than OCZ, more than you and more than almost any other component in any machine. They are built to run hot, it's their job.

You're missing the point. Rather than try to start a fruitless argument, you could pull your head out of your ass and realize that running a processor under a higher temperature load WILL shorten the lifetime of the CPU.

Mobile CPUs are especially vulnerable, because the heatsink systems are not as efficient as their desktop counterparts.
 
Not sure if this fits here as a myth but it gets really old seeing all the posts that imply that anything slower than a c2d 2.0 is unusable beyond word processing and checking email.
 
Did you bother to click that link? Because you completely missed my point, and I don't see how you could've if you'd read the link.

Your point (and your link also) was that the CPU running hot isn't good for the CPU. My point is that, intel designed these chips to be capable of running at maximum of 105 deg C running at a 35 watt draw (For example, T8300).

I'm pretty sure intel followed this rule, in which, Intel says the CPU life is 10 years, I'm pretty sure they'd followed the saying by "10 years of running it at 105 deg C w/ 35 watt draw". Because if theres a optimal temperature, pretty sure they'll give us a range.

Sure you can use your CPU for 15 years rather then 10 years if you run it 10 deg C cooler, but who really cares. I personally don't think this myth is really such a big deal for many users these days.

Safe operating temperature, is actually within specification limits designated by the creator of these chips, in which our limits is 105 deg C. Thats your safe operating temperature. If you have a failed CPU because you were using it within the limits of what Intel says, go complain to them about their obvious defect.

Furthermore, when was the last time a CPU failed within 3 years without any manufacturing defects, CPU overclock, bad install (incapable of conducting heat properly)?
 
You're missing the point. Rather than try to start a fruitless argument, you could pull your head out of your ass and realize that running a processor under a higher temperature load WILL shorten the lifetime of the CPU.

Mobile CPUs are especially vulnerable, because the heatsink systems are not as efficient as their desktop counterparts.
I would say that 999/1000, folks upgrade before that is even an issue. So as long as your machine is stable and under the max temp, you're fine IMHO.

My machine will only last 8 years instead of 11 because I ran too hot? Big deal. What you worry about more than the life-span of a CPU is the life-span of the fan attached to the heat-sink mechanism. Even still, I've only once had a fan on a HSF die, and that was just recently on an old Tyan Tiger MPX motherboard. One of the stock fans on CPU 2's HSF unit died. It's an old dual Athlon MP machine. I've been running both CPUs @ 100% (Folding@Home) for almost 7 years I suppose (I think I built this machine is late 01 or 02).

Time to find a replacement fan!
 
You're missing the point. Rather than try to start a fruitless argument, you could pull your head out of your ass and realize that running a processor under a higher temperature load WILL shorten the lifetime of the CPU.
But it doesn't really matter. And that's exactly the problem I have with the original posting: While the statements itself are probably all true, I don't think it gets the emphasis right. In the end, it's almost rather fostering popular myths instead of "debunking" them.

1. Hotter temperatures will shorten CPU lifespan. Yes - but most probably you're not going see it fail.
2. Yes, the extra 256 MB VRAM are addressable and usable - but you're not likely to notice it, either. The shader units and the bus system are simply not capable enough to let the VRAM make a difference. So it's still pretty much "marketing junk".
3. Yes, cranking up fans should decrease their lifetime - but this is the most probable thing you are going to notice within a "reasonable" time frame of using your machine. For the simple reason that the fans are mechanical parts.
 
But it doesn't really matter. And that's exactly the problem I have with the original posting: While the statements itself are probably all true, I don't think it gets the emphasis right.

1. Hotter temperatures will shorten CPU lifespan. Yes - but most probably you're not going see it fail.
2. Yes, the extra 256 MB VRAM are addressable and usable - but you're not likely to notice it, either. The shader units and the bus system are simply not capable enough to let the VRAM make a difference. So it's still pretty much "marketing junk".
3. Yes, cranking up fans should decrease their lifetime - but this is the most probable thing you are going to notice within a "reasonable" time frame of using your machine.

While VRAM may not make a difference when running at 1440x900, it definitely will when I start to play TF2 at 1920x1200 on my 24" LCD. It isn't useless.
 
While VRAM may not make a difference when running at 1440x900, it definitely will when I start to play TF2 at 1920x1200 on my 24" LCD.
Have you tried that? I have seen many GPU benchmarks but even in games VRAM hardly made a difference. Only in some specific cases. But yes, as you say these were at higher resolutions. And even then it is not often a difference between being playable or not.
It isn't useless.
Yep - but in mostly not worth the money, I'd reckon.
 
While VRAM may not make a difference when running at 1440x900, it definitely will when I start to play TF2 at 1920x1200 on my 24" LCD. It isn't useless.

I doubt you'll stress a 256MB card too much with TF2 anyway :p I was playing CS:S (which uses the same game engine) maxed out at 1680x1050 with a 128mb card from a couple generations ago just fine, haha.
 
Not sure if this fits here as a myth but it gets really old seeing all the posts that imply that anything slower than a c2d 2.0 is unusable beyond word processing and checking email.

yeah, that one i agree with. considering we have been using machines to do email, writing documents and even rendering way back when cpu were only 50mhz or lower.

even on newer os that demand more resources, you can do just about anything with a slower processor, you just have to wait longer. that and unless you really are using the cpu, you will not really see much difference between an ibook and a macbookpro. well, the 4200rpm hd will make a huge diffrerence.
 
You're missing the point. Rather than try to start a fruitless argument, you could pull your head out of your ass and realize that running a processor under a higher temperature load WILL shorten the lifetime of the CPU.

Mobile CPUs are especially vulnerable, because the heatsink systems are not as efficient as their desktop counterparts.

yes, because the magic froo froo fairies will get their wings burned.

learn to read specs.
 
I like this post, very interesting topics unlike the "I just returned my macbook pro for the millionth time and want to bitch about apple" posts.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.