And there’s another riddle. The Core M chips are supposed to draw little power, which in turn should mean excellent battery life. Yet all our best endurance scores in 2015 thus far belong to laptops with standard Core i5 or i7 chip. The victories have not been small, either; the best standard Core system lasts three hours longer than the best with Core M inside.
Why is this happening? Does it have something to do with our benchmark, which is more demanding than those used by most reviewers? Is it battery size? Or the processor itself? To find out, I ran three systems through a gauntlet of four endurance tests and analyzed the results. What I found suggests the Core M’s benefit depends on how you use your notebook.
More about Core M
Core M is, from an architecture standpoint, essentially the same as any other Core processor. It’s based on Broadwell, Intel’s newest design, which is a 14nm die-shrink of its predecessor, Haswell. In plain English, this means Core M boasts the latest, most efficient, most compact technology Intel has to offer.
What really sets Core M apart is not the architecture itself, but how it’s used. Compared to the standard Core brand, Core M can run at a much wider range of clock speed. The entry-level chip in the brand, the M-5Y10, has a base clock speed of only 800MHz and a Turbo Boost maximum of 2GHz. At the top of the range sits the Core M-5Y71, which has a base clock of 1.2GHz and a Turbo Boost maximum of 2.9GHz. That’s a range of 1.7GHz. A Core i5-5200U, by comparison, has a base clock of 2.2GHz and a maximum Turbo Boost of 2.7GHz, for a range of only 500MHz. There’s one chip in the standard Core line, the Core i7-5550U, which comes close to the Core M’s Turbo Boost extremes, and its range is only 1GHz.
The conservative base clock of the Core M series has made it possible for Intel to target a low thermal design power. This specification is a metric that helps system builders decide how robust the cooling solution in a notebook needs to be. Core M chips target a TDP between 4.5 and 6 watts, while a Core i5-5200U has a quoted target of 15 watts. On paper, that suggests the latter users up to three times more power than the former (in reality, the situation is more complex).
Yet Core M systems have consistently disappointed in battery life tests. The Samsung ATIV Book 9, which has a 35 watt-hour battery, and the Asus Zenbook UX305, which has a 45 watt-hour battery, both delivered mediocre scores in our standard battery life benchmark. To look at this matter more closely, I’ve thrown them into additional tests. To provide perspective from a more typical notebook I’ve added the Lenovo ThinkPad T450s with Intel’s Core i5-5300U and a 46 watt-hour battery. Examining how this trio performs in the same tests should be enlightening.
Testing notes
All of these tests were performed with display brightness calibrated to 100 lux using a Dr. Meter LX1010B Light Meter. All system were running Windows 8.1, with default software installed, but with power plans set to disable any feature that might alter display brightness or turn the display off entirely. Battery life measurements were taken with Windows 8.1’s built-in battery report, and represent the time required to drain from a full charge to automatic shut off, which typically occurs when the battery hits five percent charge remaining.
Idle
I started with a simple idle test. Most people don’t leave a laptop sitting at idle for long periods of time, of course, since there’s no reason to keep a system on when it’s not being used. In general, power plans will enforce sleep mode after less than 30 minutes of idle on battery. Idle testing does not provide a realistic indication of real-world runtime. Even so, it’s a useful data point.
Idle tests show how long a system can last under ideal conditions, and they can tease out the difference in power draw between the idle and active processor state.
The results of this test are striking. Samsung’s laptop delivered almost five hours of additional battery life relative to the T450s and lasted roughly twice as long per watt-hour of battery provided. In other words, our results indicate the Book 9’s power consumption on battery is almost half the T450s. Asus’ larger Zenbook UX305 isn’t that efficient, but it’s still about 50 percent better than the Lenovo.
In short, these numbers make it extremely clear the low base clock speed of a Core M processor has benefits. When the chip is allowed to slip into its lowest active power state it can last an extremely long time. But what happens when the processor isn’t allowed to rest?
Peacekeeper
Futuremark Peacekeeper is a browser benchmark. It performs a variety of tasks on loop with the intent of judging browser speed, but it also serves as a nice reproduction of heavy, constant web browsing. Because it loops, it’s a more demanding workload than you’d typically expect from actual web browsing, and it can suck down battery with surprising speed.
This benchmark really tightened the field, bringing all three notebooks within 40 minutes of each other. It also narrowed the range in minutes per watt-hour of battery life considerably. While the T450s and its Core i5-5300U processor remained slightly less efficient than Core M, the difference was no more than 25 percent, a drastic reduction from idle.
As a result, the ATIV Book 9 ends up with the worst endurance in this benchmark. Its greater efficiency is offset by its small battery, leaving it poorly equipped to handle this workload. That notable result is amplified by the incredible reduction in endurance compared to the idle test. Running Peacekeeper on the ATIV Book 9 cut minutes per watt-hour and overall run time almost exactly in half.
Video loop
There was once a time when 1080p video could challenge a notebook. That has long passed. Today’s notebooks have the processing power to easily handle high-resolution video and have hardware decode support that drastically improves system efficiency during video playback. It was no surprise, then, that our video loop (I used the second trailer for The Avengers 2: Age of Ultron) proved the least demanding load test.
Video loop
Video minutes per watt-hour
While all three systems lasted longer in this test than in any other besides the idle benchmark, the video loop didn’t change our general impression of the Core M. The Zenbook UX305 actually wins the minutes per watt-hour metric here, followed closely by the ATIV Book 9. Compared to the T450s, the UX305 is about 20 percent more efficient.
That leads the UX305 to last about 20 percent longer as well. Such a result is expected because the Asus and the Lenovo have almost identical battery capacity. That leaves the tiny Samsung ATIV Book 9 to fall behind again. While still efficient, its smaller battery means there just isn’t enough juice to keep the system powered.
Unigine Valley
Our final test is Unigine’s Valley benchmark, which is built to test 3D graphics performance. It is the most demanding benchmark overall, as games tend to place load on the processor and the graphics hardware simultaneously.
Valley loop
Valley loop minutes per watt-hour
These numbers make it obvious that Valley is the most rigorous test by far. It forced the ThinkPad T450s to dip below 5 minutes of life per watt-hour, which cuts its overall run-time almost in half compared to Peacekeeper, and no system managed to offer more than four hours of life.
Otherwise, though, the results fit snugly into the narrative so far. The Core M systems are indeed more efficient, and in the case of the Zenbook UX305, that translates to a slight lead over the Lenovo T450s. But the gain in efficiency is not enough to let the Samsung excel, and it trails the pack because of its smaller battery.
Relative efficiency
The numbers extracted here are useful, but they’re also a lot to take in. Perhaps the best way to summarize how Core M performs is to compare the minutes per watt-hour performance of the Core M systems to the Lenovo ThinkPad T450s in each benchmark.
Relative efficiency percentage
From this perspective the reason for the Core M’s lackluster performance in our benchmark becomes clear: At idle, Core M is massively more efficient than a standard Core i5 chip, and it delivers incredible battery life as a result. The 11.5 hours obtained from the ATIV Book 9 is in fact more than Samsung’s official figures claim.
Place the chip under load, though, and the benefit drops. Of course, an efficiency improvement that typically ranges from 7 to 24 percent is still impressive, and the larger Zenbook UX305 can reap rewards. Yet even that system, which boasts a battery just one watt-hour smaller than the T450s, often lasts just a hair longer. Its advantage was 17 minutes in Peacekeeper and 10 minutes in Unigine Valley. Core M makes a difference, but it’s sometimes so small you’d have difficulty noticing it without checking a benchmark.
Where does this leave Core M?
The release of Core M was exciting. It promised something that manufacturers have wanted to offer for years, but never could: super-thin, featherweight, fanless computers with exceptional battery life. Intel has delivered on the first three, but our results prove that Core M often doesn’t always deliver in endurance. While more efficient, its gain relative to a standard Core processor is at times slim, and the small notebooks it enables often have tiny batteries.
Performance makes the results more troublesome. In Geekbench, a Core M system often scores somewhere between 20 and 25 percent below a standard Core processor. The Lenovo ThinkPad T450s, for example, hit a multi-core result of 5,918. The Asus Zenbook UX305 only hit 4,445 in the same test. Why sacrifice performance if doing so doesn’t always mean better battery life?
To Intel’s credit, Core M does achieve three out of four of its goals. Sitting the T450s next to the UX305 makes the design gap between them clear; Lenovo’s system, though barely more than eight-tenths of an inch thick, looks like a brick by comparison.
Whether you’ll enjoy the benefits of the Core M processor is entirely dependent on how you use your notebook. If you generally stick to moderate web browsing, watching media, and editing simple documents, it could be the way to go. Its performance is sufficient for those tasks, and its efficiency is best in those situations. Throw a heavier workload at the chip, though, and its advantage is drastically reduced.
People who need a light system for simple work will enjoy Intel’s power-sipping chip. For others, though, its benefit will prove elusive.
