'iPhone 13' A15 chip performance continues dominance over Android rivals

Posted:
in iPhone edited September 2021
A benchmark on what is said to be the GPU in Apple's A15 chip, destined for the "iPhone 13," is 13.7% faster than the A14, maintaining Apple's lead in mobile performance.

A15 benchmark shows 13.7% faster GPU
A15 benchmark shows 13.7% faster GPU


A Manhattan 3.1 GPU benchmark test reportedly shows Apple's A15 chip hitting 198 FPS during the first round of testing. However, the second round isn't as impressive, with a score of about 140 FPS to 150 FPS.

The benchmark results show how the A15 GPU performed after two rounds of testing. The scores show that the A15 needs to throttle its speed after a time, but even when considering the speed drop, it outperforms competitors by a considerable margin.

According to a tweet from a leaker known as "FrontTron," benchmark tests were performed on a sample of the A15 acquired from a July batch. The results were discussed in a Korean forum called Clien.

The iPhone 12 averages 170.7 FPS peak performance on the Manhattan 3.1 test. FrontTron says the throttling isn't an issue since it is "way exceeding" the A14 peak performance, though the numbers appear to contradict that statement.

Apple A15 GPU peak benchmark test
Manhattan 3.1: 198 FPS (July unit sample)
However, after second round of test, throttling kicks in and drops to 140~150FPS.
(1/2)

Source: https://t.co/Sl1xfN5ktB

-- Tron (@FrontTron)


Apple is expected to announce the "iPhone 13" with the A15 processor during its September event. That event may take place on September 14, with the iPhone shipping on September 24.

Read on AppleInsider

Comments

  • Reply 1 of 13
    cpsrocpsro Posts: 3,192member
    These results aren't just preliminary, but we have no idea how the A15 was mounted and cooled.
    williamlondongregoriusmwatto_cobra
  • Reply 2 of 13
    why do people believe these kind of things?
    iPhone prototypes out of factory don’t usually run full iOS capable of using these benchmarks. Only hardware flashed by Apple with the correct iOS version could do this, and those prototypes wouldn’t be circulating in China or Korea, at least not in July, so many months before release - those would be in Cupertino!

    watto_cobraspock1234
  • Reply 3 of 13
    melgrossmelgross Posts: 33,510member
    ppietra said:
    why do people believe these kind of things?
    iPhone prototypes out of factory don’t usually run full iOS capable of using these benchmarks. Only hardware flashed by Apple with the correct iOS version could do this, and those prototypes wouldn’t be circulating in China or Korea, at least not in July, so many months before release - those would be in Cupertino!

    Over the years they’ve been pretty close to actual results. Since the process is just a slightly improved version of last year’s, I haven’t been expecting much improvement coming from that end. And we don’t know what Apple has been concentrating on.
    rinosaurwatto_cobra
  • Reply 4 of 13
    I’m afraid this low performance boost might really reduce the Android iPhone performance gap. 
    williamlondon
  • Reply 5 of 13
    flydogflydog Posts: 1,123member
    rinosaur said:
    I’m afraid this low performance boost might really reduce the Android iPhone performance gap. 
    Yeah because most people buy phones based on the Manhattan 3.1 GPU benchmark.

    GMAFB
    napoleon_phoneapartMisterKitwatto_cobramuthuk_vanalingammichelb76spock1234
  • Reply 6 of 13
    melgross said:
    ppietra said:
    why do people believe these kind of things?
    iPhone prototypes out of factory don’t usually run full iOS capable of using these benchmarks. Only hardware flashed by Apple with the correct iOS version could do this, and those prototypes wouldn’t be circulating in China or Korea, at least not in July, so many months before release - those would be in Cupertino!

    Over the years they’ve been pretty close to actual results. Since the process is just a slightly improved version of last year’s, I haven’t been expecting much improvement coming from that end. And we don’t know what Apple has been concentrating on.
    I don’t know of any example where people reported close values when stating they had access to prototypes months in advance.
    I am not talking about benchmarks done on iPhones 1-2 weeks before launch, I am talking benchmarks on iPhones months before launch (here they say in July)
    watto_cobraspock1234
  • Reply 7 of 13
    LOL… Who seriously believes this. First the iPhone 13 is not out. Second, no one has the iPhone except for possibly top execs within Apple & even if someone had their hands on the physical chip, there is no way to measure how it will actually perform once paired with the iPhone. . Thirdly, last year 99.9% of these writers that claimed to have test the then new iPhone 12 Pro had it all wrong. And lastly, even if these benchmark test prove to be accurate the performance of the A15 vs A14 is and will be so minimal because they’re both 5nm and for that reason there is only so much that can be tweaked to give any better performance of the A15 over the 14. Bottom line. Apple is about tapped out with squeezing anymore performance from a 5nm chip. You may see tweaks to the camera, battery and the screen, ie. 60 refresh to 120. And that’s a maybe. As for these so called reporters, writers etc, trying to make a mountain out of a molehill… just stop! 
    edited September 2021 watto_cobraspock1234
  • Reply 8 of 13
    The iPhone 13 is already being produced, so it makes sense that someone, however illegally, has procured a working iPhone and benchmarked it. What we don’t know is if this the variant of iPhone 13. 

    And even then, it’s possible the entire story is made up. 

    If true, this is good news for a niche benchmark. More telling will be industry standards benchmarks coming up soon. 

    As it is, Apple will be announcing the new iPhone in under one month’s time. 
    watto_cobra
  • Reply 9 of 13
    The iPhone 13 is already being produced, so it makes sense that someone, however illegally, has procured a working iPhone and benchmarked it. What we don’t know is if this the variant of iPhone 13. 

    And even then, it’s possible the entire story is made up. 

    If true, this is good news for a niche benchmark. More telling will be industry standards benchmarks coming up soon. 

    As it is, Apple will be announcing the new iPhone in under one month’s time. 
    But they are stating that the benchmarks were done in July, not now! In July it would have been very hard to get hands on an iPhone 13 with iOS, and even harder in Asia.
    spock1234
  • Reply 10 of 13
    mike54 said:
    Apple needs the performance of the A15 for the on-device scanning.
    They don't? It's just hashing and that will be done on low-power cores anyway.
  • Reply 11 of 13
    If Apple will come out with a 13-20% speed increase on every new release they can stretch these chips out for a long time, very nice. Pretty sure making a small incremental speed bump will cost them very little, maybe it's the same A14 but without some disabled cores. 
    williamlondon
  • Reply 12 of 13
    melgrossmelgross Posts: 33,510member
    ppietra said:
    melgross said:
    ppietra said:
    why do people believe these kind of things?
    iPhone prototypes out of factory don’t usually run full iOS capable of using these benchmarks. Only hardware flashed by Apple with the correct iOS version could do this, and those prototypes wouldn’t be circulating in China or Korea, at least not in July, so many months before release - those would be in Cupertino!

    Over the years they’ve been pretty close to actual results. Since the process is just a slightly improved version of last year’s, I haven’t been expecting much improvement coming from that end. And we don’t know what Apple has been concentrating on.
    I don’t know of any example where people reported close values when stating they had access to prototypes months in advance.
    I am not talking about benchmarks done on iPhones 1-2 weeks before launch, I am talking benchmarks on iPhones months before launch (here they say in July)
    We’re talking about two weeks, bow one week, not months. This is a new number.
Sign In or Register to comment.