The exhaust vent is a thin indented strip that runs across the top of the back
panel. To a casual observer, it appears to be part of the system’s cosmetic
design. Only the slight trickle of hot air that we could feel drifting from
it betrayed its true function. Even so, the amount of air passing through the
system is probably not much more than would be achieved with convection alone.
panel. To a casual observer, it appears to be part of the system’s cosmetic
design. Only the slight trickle of hot air that we could feel drifting from
it betrayed its true function. Even so, the amount of air passing through the
system is probably not much more than would be achieved with convection alone.
In addition to the small amount of airflow, some cooling is also provided by
direct conduction. As a heatsink material, the plastic exterior has nothing
on copper or aluminum, but there is enough surface area that it probably helps
a little. In this respect it is like a laptop, which also dissipate heat through
plastic. The back of the screen got quite warm to the touch, confirming that
some heat was being transferred in this way.
direct conduction. As a heatsink material, the plastic exterior has nothing
on copper or aluminum, but there is enough surface area that it probably helps
a little. In this respect it is like a laptop, which also dissipate heat through
plastic. The back of the screen got quite warm to the touch, confirming that
some heat was being transferred in this way.
- For the technically minded, a Japanese web site called Kodawarisan has photos of a stripped and disassembled Core Duo iMac. The photos show that there are three tiny blower-style fans in the system, which suggests very intelligent fan management given the acoustic characteristics of this iMac. It’s clear that the tight component integration borrows much from mobile notebook technology.
TESTING
Testing on the iMac was quite simple in some ways, and very complex in others.
Because it is a complete system, there were no drives to suspend or fans to
tweak, so measuring noise was as simple as placing the sound meter one meter
away and taking a reading.
Because it is a complete system, there were no drives to suspend or fans to
tweak, so measuring noise was as simple as placing the sound meter one meter
away and taking a reading.
Power and thermal testing, on the other hand, was complicated immensely because
we are unfamiliar with any of the hardware monitoring and benchmarking tools
available for OS X. And, for some reason none of the OS X-based utilities that
we downloaded could detect any temperatures other than the SMART data from the
hard drive. Because of these limitations, our stress testing consisted of running
a custom-compiled version of CPUBurn
for a long period of time and hoping that nothing crashed.
we are unfamiliar with any of the hardware monitoring and benchmarking tools
available for OS X. And, for some reason none of the OS X-based utilities that
we downloaded could detect any temperatures other than the SMART data from the
hard drive. Because of these limitations, our stress testing consisted of running
a custom-compiled version of CPUBurn
for a long period of time and hoping that nothing crashed.
The ambient noise level at the time of testing was 17 dBA@1m.
Activity State | AC Power Draw |
– | |
Low Power Idle | 33W |
20 dBA@1m | |
HDD Seek @ Idle | 52W |
22 dBA@1m |
With a maximum power draw of 63W, the iMac certainly qualifies as a low power
system. At idle, the system drew 46W, which will qualify for approval from EnergyStar
if their current draft computer spec makes it to the planned 2007 release. Even better, the system falls
back into a low power mode after being left alone for a few minutes, dropping
the power even more to just 33W. By way of comparison, the lowest idle power consumption
we’ve ever seen from a custom built system is 36W — and that doesn’t include
an LCD monitor.
system. At idle, the system drew 46W, which will qualify for approval from EnergyStar
if their current draft computer spec makes it to the planned 2007 release. Even better, the system falls
back into a low power mode after being left alone for a few minutes, dropping
the power even more to just 33W. By way of comparison, the lowest idle power consumption
we’ve ever seen from a custom built system is 36W — and that doesn’t include
an LCD monitor.
Much of the power savings in the low power mode seems to come from a reduction
in the brightness of the LCD screen, which drops to the lowest possible brightness.
This sometimes got quite irritating, since the low power mode sometimes kicked
in while reading a large block of text on a web page and the mouse would have to be moved around to boost the brightness back up. One solution might
have been to turn the brightness down all the time,
but this might not be practical in a brightly lit room.
in the brightness of the LCD screen, which drops to the lowest possible brightness.
This sometimes got quite irritating, since the low power mode sometimes kicked
in while reading a large block of text on a web page and the mouse would have to be moved around to boost the brightness back up. One solution might
have been to turn the brightness down all the time,
but this might not be practical in a brightly lit room.
If the system was left alone for much longer — about 15 minutes --
the system would shut itself down into standby mode, where the total power consumption
dropped to just two watts. Like the low power mode, this also had its drawbacks.
In fact, it had a habit of shutting itself down while we were running CPUBurn,
making it difficult to sustain the test for a long period of time.
the system would shut itself down into standby mode, where the total power consumption
dropped to just two watts. Like the low power mode, this also had its drawbacks.
In fact, it had a habit of shutting itself down while we were running CPUBurn,
making it difficult to sustain the test for a long period of time.
The energy efficiency of the iMac solves the mystery of how it is able to get
away with so little cooling. At first glance, the numbers don’t look that impressive,
but keep in mind that all of these numbers include the power required by the
LCD screen. Stand-alone LCD monitors typically draw between 30~40W from the
wall, so we were quite impressed when the entire system managed
to draw this little power.
away with so little cooling. At first glance, the numbers don’t look that impressive,
but keep in mind that all of these numbers include the power required by the
LCD screen. Stand-alone LCD monitors typically draw between 30~40W from the
wall, so we were quite impressed when the entire system managed
to draw this little power.
The noise levels were no less impressive than the power efficiency. At idle
(that is, most of the time), the noise level was a paltry 20 dBA@1m, but even
then the noise was barely distinguishable from the ambient noise in the lab.
The noise character was very smooth, mostly a slight whirr masked by airflow
that sounded like the hard drive. Although the iMac contains a full size 7,200
RPM drive, there was surprisingly little low frequency hum. The secret may be
the metal stand, which seemed to absorb much of the vibration from the drive.
The drive vibration could be clearly felt by placing a hand on the back panel,
but a no vibration at all could be felt at the base of the stand.
(that is, most of the time), the noise level was a paltry 20 dBA@1m, but even
then the noise was barely distinguishable from the ambient noise in the lab.
The noise character was very smooth, mostly a slight whirr masked by airflow
that sounded like the hard drive. Although the iMac contains a full size 7,200
RPM drive, there was surprisingly little low frequency hum. The secret may be
the metal stand, which seemed to absorb much of the vibration from the drive.
The drive vibration could be clearly felt by placing a hand on the back panel,
but a no vibration at all could be felt at the base of the stand.
Seek noise from the drive was plainly audible, and was the most annoying feature
of the iMac’s sonic character. It was much more audible than the measured difference
suggests, primarily because the clicks did not fade easily into background.
We did not get a chance to test whether AAM was enabled. If it was not, the
issue of seek noise might become a non-issue.
of the iMac’s sonic character. It was much more audible than the measured difference
suggests, primarily because the clicks did not fade easily into background.
We did not get a chance to test whether AAM was enabled. If it was not, the
issue of seek noise might become a non-issue.
Although the noise levels did rise slightly with CPUBurn running, the increase
was almost imperceptible. In fact, if we hadn’t measured the difference, we
probably wouldn’t have noticed — the noise character did not seem to change
at all. Only the sudden realization that the iMac was easier to hear against
the background noise convinced us that there was any audible difference at all.
was almost imperceptible. In fact, if we hadn’t measured the difference, we
probably wouldn’t have noticed — the noise character did not seem to change
at all. Only the sudden realization that the iMac was easier to hear against
the background noise convinced us that there was any audible difference at all.
iMac: Optical Drive Noise | |
Noise Level | |
31 dBA@1m | |
24 dBA@1m | |
21 dBA@1m |
As is the case for almost any system, the optical drive was a significant source
of noise when it was in use. At full speed, it measured 31 dBA@1m — not
quiet, but better than most full-sized drives. Subjectively, it also sounded
a little bit nicer: It was smoother and sounded less mechanical. The bulk of
the noise was a clean whirr floating on a bed of airflow noise, with very little
low frequency hum. The better subjective quality can probably be attributed
to the well-sealed slot, which didn’t let much noise out, and to the thick plastic
body of the iMac, which is much less prone to resonance than aluminum or steel.
of noise when it was in use. At full speed, it measured 31 dBA@1m — not
quiet, but better than most full-sized drives. Subjectively, it also sounded
a little bit nicer: It was smoother and sounded less mechanical. The bulk of
the noise was a clean whirr floating on a bed of airflow noise, with very little
low frequency hum. The better subjective quality can probably be attributed
to the well-sealed slot, which didn’t let much noise out, and to the thick plastic
body of the iMac, which is much less prone to resonance than aluminum or steel.
Fortunately, the full blast noise level of the optical drive only lasted for
about thirty seconds, even when the drive was in heavy use. After that, the
noise level dropped to a respectable 24 dBA@1m, where it stayed while the drive
was in use. Once the drive had stopped being accessed, the noise level dropped
even further to 21 dBA@1m, where it was completely inaudible unless specifically
listened for. When left idle for another ten minutes, the drive stopped spinning
entirely.
about thirty seconds, even when the drive was in heavy use. After that, the
noise level dropped to a respectable 24 dBA@1m, where it stayed while the drive
was in use. Once the drive had stopped being accessed, the noise level dropped
even further to 21 dBA@1m, where it was completely inaudible unless specifically
listened for. When left idle for another ten minutes, the drive stopped spinning
entirely.
GOING TO BOOT CAMP
Not a day after we completed our testing, Apple announced that their Intel-based
systems would be supporting Windows XP via a firmware update known as Boot Camp.
An update to OS X would also allow the default operating system to be selected
and changed. This provided us with an excellent opportunity to run a test with
our standard tools. We jumped at the chance in hopes of being able to access
the thermal sensors.
systems would be supporting Windows XP via a firmware update known as Boot Camp.
An update to OS X would also allow the default operating system to be selected
and changed. This provided us with an excellent opportunity to run a test with
our standard tools. We jumped at the chance in hopes of being able to access
the thermal sensors.
The Mac side of the upgrade was quite painless. Step by step instructions guided
us through the procedure of updating the firmware to emulate an 8088 BIOS, burning
a Windows-compatible driver disc, updating OS X, creating a Windows partition,
and then finally rebooting the system to begin the Windows installation process.
us through the procedure of updating the firmware to emulate an 8088 BIOS, burning
a Windows-compatible driver disc, updating OS X, creating a Windows partition,
and then finally rebooting the system to begin the Windows installation process.
As usual, installing Windows was an overly long process, but it proceeded without
any more troubles than usual. Care had to be taken to select the correct partition,
but no other special procedures needed to be followed.
any more troubles than usual. Care had to be taken to select the correct partition,
but no other special procedures needed to be followed.
Once Windows was installed, the Apple-provided driver disc provided drivers
for all of the essential hardware, but even with the drivers fully installed
there were still a few unrecognized devices in the Device Manager.
for all of the essential hardware, but even with the drivers fully installed
there were still a few unrecognized devices in the Device Manager.
Apple explicitly states that iSight (the webcam) and the Apple Remote (for
HTPC functionality) are not supported under Windows XP, but a webcam did show
up in the Control Panel under Scanners and Cameras. Double clicking on this
icon immediately produced a Blue Screen of Death. Next time we’ll listen to Apple,
but it was entertaining to see an iMac with the Blue Screen of Death.
HTPC functionality) are not supported under Windows XP, but a webcam did show
up in the Control Panel under Scanners and Cameras. Double clicking on this
icon immediately produced a Blue Screen of Death. Next time we’ll listen to Apple,
but it was entertaining to see an iMac with the Blue Screen of Death.
Something we never thought we’d see: The Apple and Windows logos side by
side.
Alas, our hopes of being able to see the CPU temperature were in vain, as SpeedFan
4.28 did not detect any thermal sensors aside from SMART. Once again
we were stymied. However, we did manage to run CPUBurn again and ATI
Tool to stress the graphics subsystem. We also tried to run Throttlewatch,
but once again we were unable to detect anything of use — it reported both
versions of Intel’s Thermal Management as being disabled.
4.28 did not detect any thermal sensors aside from SMART. Once again
we were stymied. However, we did manage to run CPUBurn again and ATI
Tool to stress the graphics subsystem. We also tried to run Throttlewatch,
but once again we were unable to detect anything of use — it reported both
versions of Intel’s Thermal Management as being disabled.
Activity State | AC Power Draw |
20 dBA@1m | |
Idle (LCD Brightness @ Minimum) | 37W |
20 dBA@1m | |
2 x CPUBurn | 67W |
23 dBA@1m |
Surprisingly, power consumption was higher across the board when Windows was
running. Only at idle was the difference small enough to ignore. With CPUBurn
running, the power draw was three watts higher — quite shocking because
the code should have been identical down to the instruction. CPUBurn is coded
in x86 assembly, so there should have been no differences in the final code
that was executed. Presumably, differences in the way OS X and Windows schedule
the tasks are responsible for the differences.
running. Only at idle was the difference small enough to ignore. With CPUBurn
running, the power draw was three watts higher — quite shocking because
the code should have been identical down to the instruction. CPUBurn is coded
in x86 assembly, so there should have been no differences in the final code
that was executed. Presumably, differences in the way OS X and Windows schedule
the tasks are responsible for the differences.
We were able to push the power consumption even further when we ran ATI Tool
alongside CPUBurn. The end result was a system, LCD included, that drew a total
of 73W from the wall — an amazing result for a full CPU and GPU stress
test. Even during this intensive test, the noise level never rose above what
it was with only the CPU under stress.
alongside CPUBurn. The end result was a system, LCD included, that drew a total
of 73W from the wall — an amazing result for a full CPU and GPU stress
test. Even during this intensive test, the noise level never rose above what
it was with only the CPU under stress.
Because Windows did not support the low power idle mode that was present in OS X,
the idle power level never dropped quite as low as it did in OS X. Manually
turning the LCD brightness down to minimum (via an Apple supplied tool in the
system tray) dropped the power by ten watts, but even that was not enough to
match the 33W level reached by OS X in low power mode. Clearly, OS X disables
other features as well as dimming the LCD. In fact, even when the LCD was turned
off entirely by tinkering with the Power Management settings, the power level
was still above what OS X achieved with the LCD dimmed. Assuming that Apple’s
drivers allowed Windows full control over the motherboard, this seems to indicate
that OS X is more effective at reducing system power — notebook users take
note.
the idle power level never dropped quite as low as it did in OS X. Manually
turning the LCD brightness down to minimum (via an Apple supplied tool in the
system tray) dropped the power by ten watts, but even that was not enough to
match the 33W level reached by OS X in low power mode. Clearly, OS X disables
other features as well as dimming the LCD. In fact, even when the LCD was turned
off entirely by tinkering with the Power Management settings, the power level
was still above what OS X achieved with the LCD dimmed. Assuming that Apple’s
drivers allowed Windows full control over the motherboard, this seems to indicate
that OS X is more effective at reducing system power — notebook users take
note.
CONCLUSIONS
The 17″ iMac has almost everything a silencer could want. It’s quiet,
efficient, good value (never thought we’d say that about an Apple), and it can even run Windows. It is also the only Core Duo-based “desktop” system currently available. We want one.
efficient, good value (never thought we’d say that about an Apple), and it can even run Windows. It is also the only Core Duo-based “desktop” system currently available. We want one.
More than anything else, it was the ergonomics and design that impressed us
about the iMac. Ordinarily, we rarely go for form over functionality,
but, so long as you’re willing to leave it alone without tinkering, the iMac
is perfectly functional. It is difficult to quantify exactly how the iMac succeeds
so well, as the difference is qualitative. How to start a game on 3 on 3 freestyle. The iMac has the feel of a luxury
sedan, a Mercedes perhaps. Sure, a Lamborghini or a souped up Civic is more
thrilling, but for sheer comfort and ease of use it’s hard to beat the iMac.
about the iMac. Ordinarily, we rarely go for form over functionality,
but, so long as you’re willing to leave it alone without tinkering, the iMac
is perfectly functional. It is difficult to quantify exactly how the iMac succeeds
so well, as the difference is qualitative. How to start a game on 3 on 3 freestyle. The iMac has the feel of a luxury
sedan, a Mercedes perhaps. Sure, a Lamborghini or a souped up Civic is more
thrilling, but for sheer comfort and ease of use it’s hard to beat the iMac.
The one exception is Apple’s “Mighty Mouse”. Although it pretends
to have a scroll wheel and “two button” functionality, using the tiny, hypersensitive ball to scroll up and down was frustrating, and
it took one of us the better part of an hour to master right clicking. The arrow movement is also too slow, even at maximum speed.
to have a scroll wheel and “two button” functionality, using the tiny, hypersensitive ball to scroll up and down was frustrating, and
it took one of us the better part of an hour to master right clicking. The arrow movement is also too slow, even at maximum speed.
It’s ironic that Apple would be the first company to take an Intel processor designed originally for mobile use and create such a compelling product. The question is whether any of the Windows/PC companies will come up with a viable competitor to the Core Duo iMac. Surely, there is a market for such a product!
Long story short, the iMac is a computer for the connoisseur and for “Everyman”, just as Apple
intended. Power users, enthusiasts, anyone who wants to tweak the hell out of
everything should probably stick to building their own systems — they’ll
be happier molding the computer to their own personal quirks. However, for those
who want to use a computer, not take up computer building as a hobby,
the iMac is among the best there is.
intended. Power users, enthusiasts, anyone who wants to tweak the hell out of
everything should probably stick to building their own systems — they’ll
be happier molding the computer to their own personal quirks. However, for those
who want to use a computer, not take up computer building as a hobby,
the iMac is among the best there is.
Much thanks to Apple
Canada for supplying the 17″ iMac sample for us to review.