There’s an interesting thing happening in phones, a quiet trend of product differentiation. Likely a factor influenced by component shortages, and at the same time, a more competitive market than many enthusiasts might assume.
In the gadget lab, I have two phones from Vivo. They’re both called “X70 Pro”, but one is a “Pro” and the other is a “Pro+”. There are numerous small differences between the two phones, but one significant difference, SOCs from two completely different chip suppliers. The Pro+ is using the best Qualcomm has to offer, a Snapdragon 888+. The Pro is equipped with the best chip available from MediaTek, the Dimensity 1200.
Vivo has accomplished something impressive, that these two phones “FEEL” like members of the same family. That said, we do get to take a look at how these kinds of hardware differences can impact performance.
Here are the Vivo X70 Pro and X70 Pro+ by the benchmarks!
How are each of these tests performed? Here’s my process for benchmarking phones!
SOC Differences?
Before jumping into the test results, it is worth quickly breaking down the differences between each chip.
X70 Pro+ (Qualcomm)
The Snapdragon 888+ is a minor refresh of the 888 launched earlier in 2021. Likely a result of better binning chips, and selecting the best performers as the fabrication process has matured. Qualcomm’s claims of improved performance over the “regular” 888 are fairly modest.
The 888+ uses three different CPU cores: an X1 high performance core, three A78 “big” cores, and four A55 “little” cores. Most of the claimed performance improvements from Qualcomm are by way of AI hardware, which at present I don’t have consistent ways to test.
X70 Pro (MediaTek)
The Dimensity 1200 is an interesting competitor from MediaTek. Rather than join Samsung, Google, and Qualcomm with big “PERFORMANCE” cores like the X1, the D1200 looks a lot more like the CPU arrangement we saw in 2020 from chips like the Snapdragon 865. We get four “big” A78 cores. One of the “big” cores can operate at a higher clock speed for single threaded tasks. We also get four “little” A55 cores.
If you’d like more of a breakdown on the two phones, here’s my first video comparing the two!
Looking at the specs on paper, we should expect a generational difference in performance.
Synthetic Bench – Geek Bench 5
We can use these synthetic scores as a starting point. Increasingly this year, better numbers in GeekBench have been poor predictors of performance in other real-world apps.
Looking at CPU scores, the X70 Pro+ barely edges out the top spot in single core CPU performance, with a very competitive multi core score.
Moving down the list, the X70 lines up about where we would expect in single core against phones from last year. The multi core score gets us closer to the Snapdragon 870 in the BlackShark gaming phone (the 870 is a Plus PLUS version of the 865 from last year).
Checking out GPU scores, something interesting happens. The X70 Pro posts the biggest scores I’ve seen yet from an Android in OCL and Vulkan benches. The X70 Pro+ lines up not far behind, and with a slight edge over the rest of the 2021 phones.
This is very surprising. If real-world performance follows these results, the X70 Pro should not only demolish phones from 2020, but should maintain a healthy lead over current premium phones.
Now that we’ve seen the synthetic scores, keep those numbers in mind as we time some real-world apps.
Video Render – PowerDirector
Starting off with a complex video project, we see the effects of optimization over brute force power. Xiaomi and LG had some kind of secret sauce that Powerdirector REALLY likes. An 870 and an 865 lead these charts still.
The X70 Pro+ still delivers a great render time, but it’s the slowest of the 888’s in my testing.
The X70 Pro here can’t find the same “magic” optimization the V60 and BlackShark have. The final render time is still good, but the Pro is closer to the MS Duo (with an 855) than the OnePlus 8 Pro (with an 865).
Video Render – KineMaster
KineMaster’s render engine is faster than PowerDirector’s, but we haven’t seen much change in scores over the last two processor generations.
The X70 Pro+ lines up at the top of the list for 888 powered phones. The X70 Pro slots right in at the top of the list for last gen phones.
Finishing videos in KineMaster is shockingly fast on phones, but phone hardware upgrades aren’t particularly dramatic on this app. We’d never sneeze at a 10% improvement to rendering, but the real life improvements are measured in 3-5 second faster renders for a one minute final video. We have to drop down to the MS Duo (SD855) to see a demonstrably slower score, and even then, it’s finishing the video faster than the video takes to play in real time.
Video Transcode – PowerDirector
A simpler test, with a longer video file, PowerDirector really likes the Duo and older phones here. Whatever power improvements are discussed on newer phones, this is a long enough render test that we might be witnessing some thermal throttling by the test’s completion.
Being able to spike high performance doesn’t help as much when you want more consistent performance over a longer period of time.
Unfortunately, BOTH X70’s drop to the bottom of the pack here. The X70 Pro is roughly a 1:1 render time against the length of the video file, which is honestly still pretty good for a phone, but we can find faster results.
Video Transcode – KineMaster
Similar to PowerDirector, KineMaster simple project render scores might be showing some mild evidence of throttling.
The X70 Pro takes the top spot, and is immediately followed by two 865 powered phones. The Pro+ falls farther behind, landing with the XPERIA 1iii.
Podcast MixDown – Audio Evolution
One of the tests I run which really likes the 888, Audio Evolution mixdown times have been fantastic on this newer SOC. The X70 Pro+ delivers a great time, but it’s the slowest of the 888s, while technically having the most powerful version of the 888. By contrast, the X70 Pro delivers the fastest score of the “older” SOC design.
As a brief tangent, it’s worth highlighting the BlackShark here.
The 870 SHOULD be more powerful than the other 865 powered phones, but really struggles in this test. Performance testing is complicated, and the myriad differences from phone to phone matter. The sum total of all the hardware and software differences contribute to performance. A phone is not just “a fast processor”. Buying with only an eye on the SOC might mean you get poorer performance.
If reviewers aren’t testing for the apps, games, and services YOU really use, there’s no good way to predict which phones will perform better BEFORE you buy.
Stabilization – Google Photos
Google Photos is still my voodoo test. I’m basically just tracking data here until I can better sort out what makes one phone faster than another.
I haven’t been able to suss out what influences performance here. The hunt for answers continues!
Compression Test – RAR Lab
Getting into the harder tests now. RAR Lab has a built-in synthetic benchmark of its own. The X70 scores here are “good”, but both are bringing up the rear for each of their SOC generations.
However, when we compress actual files through the app, the scores differ significantly from the synthetic test. The X70 Pro+ falls behind the 865 and 870 powered phones. The X70 Pro falls off a cliff.
I ran this test on three different days with similar results (you’re seeing the fastest score, but the delta in finished times was roughly ten seconds). My best hypothesis, the X70 Pro saw SOMETHING in the RAR Lab app which moved it to the low power A55 cores.
This is a test the Duo can finish in under 14 minutes. The X70 Pro took almost two and a half times longer. Contributing to that, while most phones finish this test with elevated temperatures (noticeably warm to the touch on 888s), the X70 Pro was only a couple degrees above its standby temperature.
Photo Processing – PhotoMate R3
PhotoMate also includes a built-in synthetic test, but it’s one we should likely devalue. Where the app cautions it might take two minutes to complete, most modern phones can finish the test in under 20 seconds.
This test is my torture test for throttling.
200 RAW photos, timing per 100 batch processed. We’re not just looking to see which phone finishes the fastest, but HOW MUCH each phone slows down over the test.
If we care about optimization, there’s something impressive about the Duo, as an example. It’s slower than most of my 865 powered phones, but it throttles less than 1% from first to second batch. The OnePlus 9 Pro is the overall fastest performer in this test, but slows around 9% between batches.
Unfortunately, both X70’s fall flat for BOTH metrics. They are overall the slowest at completing both batches, and they each suffer noticeable throttling per batch. It shouldn’t be a surprise that the 888+ powered X70+ also hit around 9% throttling. The Dimensity fared better with around 5%.
I’m at a loss to properly describe these results. If again, this app is being moved to the A55 cores, why would the synthetic scores show so high? It can’t just be an issue with the thermals on the BIG X1 core in the 888+. The X70 Pro doesn’t have that core, and takes almost twice as long as the V60 to complete the same tests.
I believe vivo might be employing some kind of thermal management, to balance performance against temperatures, but these results are SIGNIFICANTLY different than our video rendering tests. I wouldn’t normally harp on this as much, this test is an extreme example of a sustained workload for a phone. Yet, vivo makes bold claims about camera performance.
The person looking to shop an X70 is probably more likely to dabble with shooting in manual modes and in RAW. Editing a single RAW file, or a small batch of RAW files, likely won’t be noticeably slower than many other phones, but it is philosophically disappointing that these phones don’t handle batches better.
Gaming – BrightRidge
Looking at some gaming, Brightridge continues to be an interesting way to compare performance. As phones get more powerful, we can turn up more and more performance options on these devices. I might start testing QHD phones in QHD, as those frame rates are now more playable with all the eye candy.
The X70 Pro and Pro+ land about where we should realistically expect given the different SOCs. The Pro+ sustains mid 50’s in most areas, small peaks into the low 60’s, this is very smooth game play.
The X70 Pro is also capable, floating into the low 50’s and sustaining mid 40’s for frame rate.
This is nicely playable, and owners shouldn’t need to fuss much with performance options to find smooth performance.
Gaming – Undead Horde
Undead Horde is a more challenging title, not for extreme graphics, but for an incredible amount of unit management.
The X70 Pro+ handles this workload really well. Getting into nearly 30 minutes of game play, the average frame rate is still managing to stay in the high 50’s, with only occasional higher spikes, and the dips in this graph are often from dialog or transitions.
The X70 Pro also hangs with solid performance in the high 40’s, though it is more noticeable while playing, that transitions are not as smooth. With lots of units on screen, the game doesn’t “chug”, but you can see where it doesn’t stay as fluid as on some more expensive phones.
There are two “flavors” of game management on the 888.
Samsung and Xiaomi don’t put much of a limit on the GPU, and this game is largely unplayable on a Galaxy S21 or Mi 11. The GPU CAN peak a higher frame rate, but crashes much lower. Phones like the OnePlus 9 put a cap on frame rate, and that seems to help sustain smoother game play over longer play sessions.
You might remember the synthetic GPU scores from GeekBench?
The Pro scored higher than the Pro+. There might be specific games where those synthetic scores resemble some real-world game play, but testing on a handful of games like Riptide Renegade, Pascal’s Wager, Battlechasers, and Tesla vs Lovecraft, I can’t find any where I could spot an advantage on the Pro over the Pro+.
Conclusion
I feel it’s worth pointing out again, synthetic benchmarks are poor predictors of real-world performance.
If you saw the GPU scores from GeekBench on the X70 Pro, and expected that vivo to deliver a consistent advantage for your favorite games over premium phones from 2020, you would probably be disappointed. That doesn’t mean the X70 Pro is “bad”. It is a solidly competitive performer, and considering the SOC architecture, is right in line, often ahead of, similarly produced chips from Qualcomm.
Looking at how these chips were produced, we wouldn’t have expected the MediaTek to significantly outperform Qualcomm, but those GeekBench scores set an unrealistic expectation.
As a second data point, this whole year we’ve struggled to realize consistent performance advantages for phones running the 888.
From the 835 to the 865, Qualcomm demonstrated across the board improvements for almost all of the apps I can measure in my tests. The 888 is more powerful, but it’s finnicky where I can really put that power down.
We’re likely seeing a shift in focus from traditional CPU and GPU iteration to improving AI or machine learning capabilities. Also, more companies are utilizing co-processors to improve specific applications like photo and video capture. The Pro+ rarely outperforms “regular” 888’s, but when it loses to “regular” 888’s it often loses while feeling cooler to the touch.
The flipside of that, the phone performs surprisingly well in ways I can’t as reliably test. When vivo claims the V1 co-processor improves low light photo and video performance, it does capture night photos shockingly fast, and can almost deliver night-vision like results for video.
The OnePlus 9 Pro caps 4K/60 video at 5 minutes per clip. The X70 Pro+ ran for 32 minutes in 75 degree weather (24 degrees C) without giving me any thermal warning.
It’s difficult to interpret these results.
In years past, we might have been more cynical. The manufacturer might have put in an “overclocked” performance mode that only activated when it detected a benchmark.
I don’t think that’s what’s happening in 2021.
While a subtle difference, I think Qualcomm and Samsung have delivered BIG PERFORMANCE chips that are VERY good at brief spurts of processing. I don’t believe Xiaomi, Samung, OnePlus, vivo, Sony, etc are gaming the benchmarks with sneaky software. Quite the opposite in fact.
A synthetic benchmark rapidly runs a series of small tests, and that’s where these new chips excel. A short burst of activity, a quick reset, and another short burst of activity. This in no way resembles how “average consumers” might use their phones. Qualcomm and Samsung have built chips that run those mini-tests very well and get BIG SCORES out of benchmarking apps.
These chips burst performance SO well, that the phone manufacturers are all trying to find ways to reign in that performance. Thermals and battery draw are precious considerations for a mobile pocket computer. Without some limits, these gadgets become uncomfortable to hold, won’t last as long away from a charger, and might be concerning for longevity. Heat has a way of breaking down batteries and components faster over time.
I’m of two minds on the vivos.
On the one hand, someone spending a lot of money on a phone might want to know they have a certain tier of performance. We like to think our phones are “winning” some kind of implied race. It’s very likely that, depending on the apps and services YOU might use, there is a phone out on the market which can outperform the X70s in a straight up sprint.
On the other hand, I feel vivo might be aiming for more of a “performance where it counts” flavor of a premium phone.
Minus one outlier score for each X70, they’re never slow. It’s less sexy, and more difficult to describe, but maybe some folks out there might be willing to sacrifice the absolute fastest and hottest performance.
Maybe “really good” performance is better than “always the bestest” performance, if “really good” arrives cooler and a little more consistent?
Regardless, the vivo X70 Pros are monster flexible workhorse phones, especially for content creators.
Even if the GeekBench scores set some unrealistic expectations…