![]() |
![]() |
#1 |
∂2ω=0
Sep 2002
República de California
265008 Posts |
![]()
Like many laptops, my vintage-2009 MacBook is undercooled. It`s fine for normal daily tasks, but I also run my Mlucas code (similar thermal properties as Prime95, big-FFT code which hammers the SIMD unit) on it. Even single-threaded is sufficient to get the fan maxed out and the CPU hot enough that it frequently goes into throttle-down mode. I have no complaints about the MacBook fan - it sounds as quietly powerful as the day I first used it, and is the best fan I've ever had in a laptop. These things are just not intended to run heavy-duty code 24/7.
Based on timings and 'top'-based profiling, the only time the code really runs at 100% is for a few minutes after startup, and if I set the lid-open unit next to an open window on a cool night. With the nights now being warmer as summer nears, even the latter method is not enough to keep the CPU cool enough to prevent auto-throttling, so I've been trying a few things, and finally hit upon what appears to be an optimum configuration for cooling when the unit is running DC-style code but not being actively used by me. I call it the "A-frame": simply tip the lid-open notebook onto the 2 long edges created by its opened lid, with the fan vents thus pointing up toward the ceiling. This aids cooling in 2 major ways: 1. The plastic-shielded underside of the unit, which normally gets quite hot since it is sealed off from airflow and has the plastic housing interfering with heat dissipation, is now fully exposed to the air. 2. One creates a "chimney" effect which maximizes natural convection in aiding the fan: air freely enters the fan intakes (mainly via the porously-designed keyboard) through the open ends of the A-frame, and blows out the top of the thus-created chimney, in the natural direction heated air wants to go. It looks a bit odd, but works beautifully. My results-file timings show full speed ahead for the entire night, meaning 10-20% more throughput than sitting on my desk in "orthodox" clam-style mode. I will try running stuff on both CPUs tonight and see if any throttling occurs. Here is a high-tech schematic showing a side view of the A-frame configuration, complete with a Schlieren photograph of the lovely convective whorls of hot air exiting the top of the chimney: Code:
? ? ? /\ / \ / \ / \ Last fiddled with by Batalov on 2012-07-09 at 22:16 Reason: /\ |
![]() |
![]() |
![]() |
#2 |
Jun 2003
2×32×269 Posts |
![]()
When was the last time you opened it up and blew out the dust?
|
![]() |
![]() |
![]() |
#3 |
"Mark"
Apr 2003
Between here and the
52×13×19 Posts |
![]()
I have a stand (mStand), although my MacBook Pro is even more vintage than yours, circa 2006.
|
![]() |
![]() |
![]() |
#4 | |
Sep 2009
11110100012 Posts |
![]() Quote:
On the '2006 MacBook Pro I was given in my previous job, it took only one year of 24/7 usage to damage the fan system, despite blowing it multiple times (including once case full open, when I had to change the HDD, the second HDD killed by the computer in merely four years...): it became very noisy. And yet, it was hard to keep bare hands on the metal casing, which means that its temperature was circa 50°C... Last fiddled with by debrouxl on 2012-04-30 at 07:38 |
|
![]() |
![]() |
![]() |
#5 | |
"Mark"
Apr 2003
Between here and the
52×13×19 Posts |
![]() Quote:
Last fiddled with by rogue on 2012-04-30 at 12:42 |
|
![]() |
![]() |
![]() |
#6 | |
∂2ω=0
Sep 2002
República de California
26×181 Posts |
![]() Quote:
Last week I installed a piece of freeware called smcFanControl, which allows me to monitor internal temp and fan speed. When nothing significant is running, e.g. as I write this, internal temp shows 50-60C depending on configuration (bottom elevated or not) and ambient temp/airflow. If I fire up some heavy-duty crunching program the temp very quickly gets up toward 80C, though (on my system) auto-throttling is a serious issue only for highly-optimized FPU-based code. Note the 80C is accompanied by the fan revving to 6-7krpm, and there are no odd noises from it, and a robust warm-airflow out the back vent which tells me the fan is working fine. I notice a significant difference depending on the nature of the code: When running my Mlucas TF code which is highly optimized but 64-integer (as opposed to FPU) based, it gets warm but stays below the auto-throttling temperature threshold. When running Mlucas in LL-test mode it hammers the FPU by way of SSE2, and the temperature according to the smcFan app rises from 50C to 80C in under a minute. Whether auto-throttling kicks in depends on ambient conditions ... but at ambient temps over 70 it starts auto-throttling in no more than 5-10 minutes. I sometimes will elevate the back of the unit slightly (by e.g. putting it on the edge of a book) which allows me to type and gives better airflow, but again that helps less that it should because it only keeps the plastic shell down there a little cooler - there are no air intakes on the bottom which could really benefit from it. |
|
![]() |
![]() |
![]() |
#7 |
∂2ω=0
Sep 2002
República de California
26×181 Posts |
![]()
The cooling fan on the above MacBook died last night - I have been running an app called smcFanControl to help monitor temperature and fan RPM, noticed last night that 'top' showed kernel_task running at 100%, which indicates big-time thermal throttling (I know the "100% CPU" is counterintuitive, but based on now-long experience with this system, kernel_task "CPU usage" really correlates with "throttle-idled cycles"), which normally only happens in a closed warm space, never with the system in a cool-breeze spot as it was last night. Sure enough, smcFanControl showed RMPs stuck at 0, and when I put my ear close to the vent I heard faint ticking noises which I'd never heard before.
The system is over 3 years old and out of warranty, so I'd like to do a do-it-yourself fan replacement. Have found plenty of online guides to this, but oddly, have been unable to find a replacement fan unit for this particular Macbook model (5,2) ... for example when I google macbook replacement "cooling fan" amazon.com I get lots of hits, but all are for MacBook Pro models. Adding "13-inch" "5,2" to the query does not help. (I *hate* that apple did not give their non-pro, non-air plain-vanilla MacBook a distinguishing 2nd name, to help differentiate it from the Pro and Air in searches, I ran into similar time-wastage a couple months ago when I needed to find a replacement battery for my system. How about "MacBook Classic", Apple? But, I digress). Any advice for finding a replacement fan unit for non-pro MacBook model family 5,* would be greatly welcome. I am also planning on picking up a micro-toolkit like this one. |
![]() |
![]() |
![]() |
#9 | |
∂2ω=0
Sep 2002
República de California
26×181 Posts |
![]() Quote:
I wonder why that was in none of my first-few-pages of search results ... did you use a search engine or search on Amazon? ----------------------------- Edit: Now that I know the part #, I also found it on ebay for ~1/3 the price. (eBay was $12.50 + free ship, amazon is $30 + $5 shipping, not eligible for free super saver shipping). Sweet! I will provide update once the part arrives. |
|
![]() |
![]() |
![]() |
#10 |
P90 years forever!
Aug 2002
Yeehaw, FL
67×109 Posts |
![]()
I replaced one of my fans 9 months ago and it died again last week :(
They are a pain to replace, I'd rather upgrade my laptop than replace fans every year. BTW, you can use CoolBookController to look at the actual CPU speed. My 2.5 GHz Pro is now running at 1.6 - 1.8 GHz on one fan. |
![]() |
![]() |
![]() |
#11 | ||
∂2ω=0
Sep 2002
República de California
26·181 Posts |
![]() Quote:
What tools did you need for your fan-swap? (Not being a PC repair geek, I'm still trying to figure out what a "spudger" is ... sounds like something one uses to dig up potatoes :). Quote:
BTW, and with a "do not try this at home" legal disclaimer, using the above "A frame" trick and placing the macbook in my open screened bedroom window last night allowed it run some pure-integer factoring code all night with very little throttling (inferred from kernel_task CPU % and the smcFan temperature display) even with no fan ... but my SSE2-floating-point-heavy code runs much too hot. |
||
![]() |
![]() |
![]() |
Thread Tools | |
![]() |
||||
Thread | Thread Starter | Forum | Replies | Last Post |
Submission not Properly Registered | a1call | FactorDB | 5 | 2016-04-17 16:59 |
How to properly delete Linux on a multibot PC? | Rodrigo | Linux | 8 | 2015-12-13 23:22 |
need help - assignments did not transfer properly | ixfd64 | Software | 4 | 2011-01-01 18:58 |
new MacBook Pro! | ixfd64 | Hardware | 15 | 2009-12-16 20:05 |
4 not so easy pieces? | Uncwilly | Puzzles | 35 | 2006-11-15 01:07 |