A Day with the Huawei Mate 30 Pro: Kirin 990 and 7680 FPS Slow Motion Tests
by Dr. Ian Cutress on September 23, 2019 4:00 AM EST- Posted in
- Smartphones
- Huawei
- Kirin 990
- Mate 30 Pro
- 7680fps
A Benchmark or Two
Having had the smartphone for only a short while, I have not spent too much time on benchmarks – we will have that for the full review, when Andrei gets a chance to do a full teardown of the Kirin 990 SoC. For now, we have some PCMark data for you. The key test we like here is Writing, because this test is typically representative of a standard user experience and responsiveness.
With the extra CPU frequency and wider GPU, the Mate 30 Pro gets some gains over the previous generation Mate 20 Pro and P30 Pro, but ultimately nothing groundbreaking.
We attempted to run our AI benchmarks, however it would appear that Huawei is not yet ready with the NNAPI libraries for use with its Da Vinci NPU architecture. We expect those to roll out in due course.
Facebook
Twitter
RSS
47 Comments
View All Comments
pberger - Monday, September 23, 2019 - link
I'm more impressed with the 76'206 AI-Benchmark score of this phone thanks to the new NPU more than twice more performant than the 980 one.Now how this can be compared to desktop GPU scores? Any link?
designerfx - Monday, September 23, 2019 - link
", the bottom left is a 40MP f/1.6 camera with OIS, and the bottom left is an 8MP f/2.4 telephoto camera and lens, also with OIS."might want to specify which bottom left, apparently - or fix your typo
eastcoast_pete - Monday, September 23, 2019 - link
Thanks Ian! Question: how is the reception and call quality of the phone? While I use my smartphones for all kinds of other things, my smartphone is also my main phone, and I had some otherwise good smartphones fail at being good at that - making phone calls. Any impressions?saru44 - Tuesday, September 24, 2019 - link
I am really upset with the direction the smartphone reviews are going! They have so many advanced tools and benchmarks to test the display, processor etc, but they outright skip the most important aspect of a phone to many people - Telephony. There is nothing in the reviews about call quality, ear piece volume levels/ clarity, network reception quality inside buildings/ poor network coverage areas, microphone quality (how good the person sounds to the caller on the other side), etc.. Just nothing!peevee - Tuesday, September 24, 2019 - link
"This does one of two things. Firstly, it affects how we change the volume, as there is no volume button control. In order to adjust the volume, the user double taps on the edge of the display, and the volume pop-up allows the user to swipe up and down, using a thumb on the edge of the display, to adjust the volume"Meaning that:
1) There is no rim, so placing the phone display down will scratch the glass with a grain of sand etc
2) a case cannot protect the sides of the phone as that would disable volume adjustment function.
3) You cannot adjust volume by feel in your pocket/phone holder without taking it out and turning on the screen first.
Stupid designs are stupid.
s.yu - Tuesday, September 24, 2019 - link
I noted the volume issue at GSMA, I said that if the rumors are true (which they turned out to be), then the volume implementation would be a total gimmick. NEX3 has a highly similar screen yet their volume is pressure based, which is compatible with a case.peevee - Tuesday, September 24, 2019 - link
"At 256x however, the video is also 32 seconds, meaning that the sensor can only record 0.125 seconds of video."Useless. Typical reaction time is over 0.3s. Way more when a person is tired. Nobody can catch the event lasting 0.125s.
Somebody needs to start prosecution for false advertisement.
peevee - Tuesday, September 24, 2019 - link
The resolution in the high-fps video is not 720p either, it looks like 160p. Looks like they "interpolate" not just frames but pixels in the frame. All while advertising high fps and high res. Fraudsters.s.yu - Tuesday, September 24, 2019 - link
Wow I was thinking 360p, 160p is.........I think they read out more than that, but it could be a little harder to test, you'd need something flying by at high speed on basically the same plane as an MTF chart.Or have the MTF chart flying at high speed, which is even harder.
Or have the phone fly by the MTF chart at high speed while keeping a constant plane of focus, which sounds equally as hard as the flying MTF chart.
s.yu - Tuesday, September 24, 2019 - link
...However, one could shoot the same scene with the slo-mo 720p, and with a regular 60fps 720p, and see if there's a notable degradation in resolution. If the degradation approaches what the video looks like downscaled to 360p then upscaled back, then it suggests fraud.