$1,099 iMac review: Lose 50% of your performance to save 18% of the money

Apple's new $1,099 iMac will undoubtedly be a popular computer. People in the know who want the most computing bang for their buck would be smarter to step up to a higher-end model, but there are plenty of people—casual users, schools, businesses—who just want an iMac that's "fast enough," not one that's "as fast as it could possibly be."

For those people, we obtained one of the new entry-level iMacs so we could evaluate its performance. On paper, it sounds like a big step down—you're going from a quad-core desktop processor and GPU to a dual-core Ultrabook processor and GPU. This new iMac and the base MacBook Air models in fact use the exact same processor, even though historically there's been a big performance gap between MacBook Airs and iMacs.

In practice, the story is more complicated. Let's talk about what the new low-end iMac changes, and then we'll spend some time looking at processor performance.

What’s changed, and why?

Let's get this out of the way first: the low-cost iMac is externally identical to the model released in late 2013, which is itself identical to the redesigned model from late 2012. It has the same 21.5-inch, glare-resistant, 1080p IPS display panel and the same port layout (one audio jack, one SD card slot, four USB 3.0 ports, two Thunderbolt ports, and one gigabit Ethernet port). Its other networking features are also unchanged—it's got a three-stream 802.11ac implementation capable of up to 1.3Gbps connection speeds and Bluetooth 4.0. We're not even taking new pictures of the thing. There's nothing new to see.

Two major changes have been made to the iMac's internals. First, its 8GB of RAM is soldered to the motherboard and cannot be upgraded, either at purchase or afterward. Second, it's using the same dual-core 1.4GHz Core i5-4260U as the MacBook Air, even though every consumer iMac introduced since 2011 has included a quad-core desktop CPU.

Higher-end 21.5-inch iMacs can still be upgraded to use 16GB of RAM at purchase, but end users haven't really been able to upgrade them after the fact since the new designs were introduced in 2012. Selling a 21.5-inch iMac with soldered-in memory is just the next step down that road; larger 27-inch iMacs can still be upgraded after the fact via the RAM access door on the back, while smaller 21.5-inch iMacs become more integrated and appliance-like. If Apple can save money in a $1,099 iMac by soldering the RAM to the motherboard and limiting customization options, it can get away with it because people buying the cheapest iMac probably won't need more memory than that.

The CPU choice has less to do with the processor and more to do with the GPU. Older cut-down iMacs offered to schools and businesses have used dual-core Core i3 desktop CPUs rather than quad-core i5s or i7s. This time around, we suspect Apple is using a dual-core laptop CPU because of its Intel HD 5000 integrated GPU. Dual-core, socketed Haswell desktop processors include Intel's HD 4600 at best, and in this day and age it makes sense to sacrifice a little CPU performance in favor of a little GPU performance. Using the same processor in both the iMacs and MacBook Airs also probably nets Apple a good volume discount that it wouldn't get if it bought CPUs specifically for this low-end iMac.

Now that we know what's different about this cheap iMac and why, let's look at the performance of the processor itself. Do you get much extra performance by putting the same CPU in a roomier enclosure, or are we still looking at a desktop that acts like a MacBook Air?

Versus the MacBook Air: Playing it cool

We're at a point where model numbers and clock speed information tells you less than ever before about how a given chip will perform. This is especially true in laptops, tablets, and phones, which all throttle a chip's performance to varying degrees to save battery life and prevent heat-related problems. We covered this issue more extensively in our evaluation of the Nexus 5's SoC performance last year.

We had assumed that the low-end iMac would be able to outpace the MacBook Air, since there's more room in its case to move heat around (and therefore, more headroom for the CPU to run at its top speeds before throttling happens). Intel's handy Power Gadget will tell you detailed information about your CPU and GPU's clock speed, power draw, and temperature in real time, so we fired it up to take a look at how both computers behave.

Notice that the iMac's CPU and GPU idle at a higher clock speed than the chip in the MacBook Air, and that the power draw goes up accordingly (from about 0.54W in the MacBook Air to 0.80W in the iMac). The laptop's CPU is tuned to use less power and make the battery last longer, and the desktop's is tuned to have better performance. Also note that, despite these higher clock speeds, the CPU temperature in the iMac is around four degrees lower.

So far, it looks like our hypothesis is correct—the iMac looks like it should be able to outpace the MacBook Air. Let's run Geekbench 3, which more-or-less simulates the way people actually use their computers—a couple minutes or so of intense activity, followed by a return to idle when they're done.

These scores are identical where the CPU performance is concerned, though it looks like the MacBook Air's memory scores are consistently a little lower than the iMac's (possibly a side-effect of using low-power DDR3L rather than standard DDR3). We combed through the Geekbench results database and confirmed that our results were consistent with other MacBook Airs, and not something specific to our unit.

Now we'll kick one of the CPU's two cores into high gear using the yes > /dev/null command to send it to 100 percent usage. Since these CPUs have Hyperthreading and the OS thinks it has four logical "cores," you've got to do this four times to peg the entire CPU. We also tested one core by spawning two threads since Turbo Boost can scale clock frequency a little higher if fewer cores are enabled, but our basic findings were the same.

We examined clock speed, temperature, and power usage while the CPUs ran full-tilt for around 20 minutes, well after CPU temperature stabilized, and we were surprised to see that the iMac and MacBook Air could both run at a sustained 2.4GHz for basically as long as we'd leave them on.

The bigger difference was in heat and fan noise. The MacBook Air's fan spun up to a persistent whine and the CPU climbed to a toasty 96 degrees Celsius before leveling out. The speed never fell, but the laptop was also on a flat desk in an air-conditioned apartment; if its fan is muffled by a couch or bed, or if it is being used in a warmer climate, that CPU speed is much more likely to throttle down. The iMac, on the other hand, was cool as a cucumber and inaudible in my mostly quiet apartment. Its CPU temperature hovered around 60 degrees, 36 degrees lower than the MacBook Air. Actual performance isn't different, but less heat means less stress on the components, which should translate into a longer lifespan. With the CPU at full-tilt, the chip was consuming around 14W in both the MacBook Air and the iMac.

CPU performance between the two should be pretty similar—the iMac might be a little quicker to spring to life for short bouts of activity, but in the long haul both of them seem capable of sustaining their max CPU frequency for as long as it's necessary. If you're slamming both CPU cores without using the GPU, it's safe to say the low-cost iMac has a 2.4GHz CPU rather than a 1.4GHz one.