Apple MacBook Pro 13 (Mid 2017, i5, Touch Bar) Review

Our test model is the entry-level SKU of the MacBook Pro 13 Touch Bar for 1999 Euros (~$2282) that we bought in Apple's online store. Except for the processor and the SSD, there are no changes compared to the 2016 model, so we will not cover all sections in this article. All the corresponding information is available in our in-depth review of the MacBook Pro 13 (Touch Bar, Late 2016). The following pictures are from the 2016 model as well.

We got a second test model (i5, 512 GB) from the online retailer Cyberport. We repeated the runtime tests and performed additional benchmarks with the 512 GB SSD.

Change log:

• 06/15: SSD performance, temperature, Wi-Fi runtime
• 06/17: fan noise, battery runtime load & video
• 06/21: input devices, color accuracy, PWM, gaming performance, power consumption
• 06/29: communication, several additions, pros/cons, verdict, rating
• 07/07: 2.4 GHz Wi-Fi problem specified

The keyboard looks identical and Apple still uses second-generation Butterfly keys, but our colleagues from iFixit noticed a slightly changed construction in their teardown. In addition to the 12-inch MacBook there were also reports about keyboard problems on the Pro models. Our MacBook Pro 15 (Late 2016) we use in the office also had some problem with the "S" key recently. The key stroke was either not triggered or triggered twice. However, the problem was gone after a while. Obviously, we cannot say whether there will be any long-term issues, but it does not leave the best impression. Subjectively, the typing experience did not change compared to the previous model in a direct comparison.

The display did not change. The 13.3-inch screen still has a resolution of 2560x1600 pixels, which results in a pixel density of 227 PPI. The Panel-ID is APPA034 and is therefore identical to the 2016 model without the Touch Bar. The maximum luminance in the center is 551 nits and the average result is still very good at 514 nits, which surpasses Apple's claimed 500 nits. That the contrast is still "just" 1148:1 is a result of the slightly increased black value (0.48). The battery is reduced on battery power, because the center value drops to 502 nits. However, this result easily surpasses the Windows competition.

We have used a more sensitive measuring device since the beginning of this year, which can detect higher frequencies. We can measure brightness fluctuations at a very high frequency of 119 kHz for the MacBook Pro 13 for brightness levels of 80% or below. We also checked the old 2016 MacBook Pro 15 with the new sensor and got a similar result. Because of the very high frequency, there should not be any problems even for sensitive users.

The MacBook Pro 13 with the Touch Bar is once again the more powerful model out of the two 13-inch systems since Apple uses dual-core processors with a TDP of 28 watts. This is an advantage for the GPU in particular, because the higher consumption limit gives more headroom for the Turbos of the two components. However, you should not expect a revolution from the Kaby Lake generation, because it is only a small update compared to Skylake.

All in all, the current MacBook Pro 13 Touch Bar is suited for many applications. If you need a lot of CPU/GPU performance though, you are still better off with a real quad-core and a dedicated graphics card.

The processor can maintain its performance under sustained workloads. A loop of the Cinebench R15 Multi Test determines a steady score after 20 runs; the only small outlier was probably affected by a background process. The CPU performance is not reduced on battery power.

More benchmarks and technical specifications for the Intel Core i5-7267U are available in our Tech section.

We performed the synthetic PCMark benchmarks running on Windows (BootCamp). You should note that the results also depend on the drivers, which are provided by Apple. All comparison devices are still pretty close in the Home Test, but the two Windows-based models are clearly ahead (+20%) in the Work Test. The new MacBook Pro 13 also falls behind the direct predecessor, despite the more powerful hardware.

So what does this mean? Subjectively, you cannot notice a difference and the MacBook Pro 13 is extremely responsive and there are hardly any delays when you launch applications.

Our test model is equipped with the 256 GB PCIe-SSD with the designation AP0256. Not only the designation is similar to the predecessor; we can also see similar performance results. AS SSD running on Windows still has some issues (also the case for the previous models), but all three drives are basically on the same – good – performance level.

The integrated GPU Intel Iris Plus Graphics 650 is the direct successor to the 550 inside the 2016 model. The basic architecture of the GT3e chip did not really change with the switch to Kaby Lake and the synthetic 3DMarks also determine almost identical scores. Apple's new video standard HEVC was already supported by the old Iris Plus Graphics 550, but the new GPU now also decodes HEVC Main10 with a color depth of 10-bit as well as Google's VP9 codec.

The GPU performance is not reduced on battery power. More technical information and benchmarks for the Iris Plus Graphics 650 is available here.

The gaming performance of the Intel Iris Plus Graphics 650 is between the two dedicated chips Nvidia GeForce 930MX and 940MX, depending on the game. Modern titles are usually limited to low details and resolutions if you want to reach smooth frame rates. Less-demanding titles like Overwatch or Battlefield 1 can also be played at medium settings. The Iris Plus Graphics 650 is generally best suited to older titles.

The fan control is one advantage of the small 13-inch MacBook Pro. The two fans don't run while the device is idling or with light workloads, so the system is silent. This is also a big advantage over the 15-inch models, where the fan is always spinning (albeit hardly audible). The fan speed is seamlessly adjusted to the workload. We can only measure 33.3 dB(A) in the first 3DMark06 run, but more in the second run than even in the extreme load scenario with FurMark and Prime95 (43.2 dB(A)). It will then level off at the result of the stress test. We even measured 48 dB(A) for a short while (only CPU load by Prime95) and we even noticed slight vibrations on the desk (see the peak in the fan noise diagram at low frequencies). The fans are generally quieter in macOS; we can only determine between 32-34 dB(A) (3000 / 3200-3600 / 3900 rpm) in the Unigine Valley test.

The surface temperatures of the Apple MacBook Pro 13 2017 are not too high and there are only small differences compared to the predecessor. It is possible to use the MacBook on your lap, even under maximum load or when you play games, since we only measure up to 39 °C at the bottom. The maximum temperature at the top is 43 °C in the upper area of the keyboard. The Dell XPS 13 QHD+ i5 gets similarly warm, but we recorded higher load temperatures for the Lenovo X1 Carbon 2017.

As per usual, Apple tolerates high CPU temperatures to provide the highest possible performance in load scenarios. We can see CPU temperatures between 90-100 °C running on macOS as well as Windows. It is not easy to see exact clocks, especially with Windows. The stress test with the two tools Prime95 and FurMark is executed with a CPU clock of 2.1 GHz (~10 watts). The consumption of the GT cores (GPU) is 18-19 watts in this case. For comparison: The GPU consumes 22-25 watts when we only stress the GPU via FurMark, so the stress test clock should be a bit lower. This behavior changes a bit on battery: The graphics card is slightly throttled (~15 watts), which gives the CPU a bit more headroom (2.3 GHz, ~12 watts).

The speakers did not change compared to the previous model, but there was no reason to change them. Despite their small size, they manage excellent results in our measurements and also convince subjectively.

Frequency comparison (checkboxes select/deselectable!)

We can measure higher idle consumption results compared to the predecessor. This also corresponds with the shorter battery runtime we determined (next section). Both models are on par under load though.

It is also very interesting to compare the power consumption when you are working with an external monitor. The display is eliminated as a consumer and we can compare the consumption of the hardware under load. A test with The Witcher 3 (Ultra) shows that the two MacBook Pro models (13 & 15-inch) are basically on par. This shows that the internal Iris Plus Graphics (4.8 fps) is a powerful iGPU, but it is not very efficient compared to the dedicated Radeon Pro 555 (13 fps).

The battery capacity is still 49.2 Wh. The Kaby Lake processor is manufactured in an optimized process (14 nm+) compared to its Skylake predecessor, so it should consume less power. We expected longer battery runtimes from the new model, but this is not the case in our Wi-Fi test. The previous model still lasted almost 9.5 hours at an adjusted luminance of 150 nits, but the current model only manages ~8:40 hours. We repeated the test three times, but the result was the same with a couple of minutes' tolerance. This means the MBP Pro 13 is on par with the ThinkPad X1 Carbon and Dell XPS 13 QHD+, but they also feature bigger batteries and different resolutions.

This trend continues in the video test, where the short movie Big Buck Bunny (H.264, FHD) runs in a loop at 150 nits. The result of almost 7 hours is not only beaten by the predecessor, but all the other comparison devices are significantly more enduring as well. The load test determines a couple of minutes more compared to the 2016 model, but you still have to recharge it after 70 minutes. We can see the impact of the 28-watt processor in this scenario and the ULV systems from Dell and Lenovo manage longer runtimes.

A second test model, which is also equipped with the Core i5 processor, managed 08:38 h in the Wi-Fi test and therefore confirms our initial result. The Wi-Fi runtime running Windows (BootCamp) is a bit shorter at 06:54 h. The charging time is around 2 hours, but 50% of the capacity is available after little more than 30 minutes.