How can I tell if my iPhone is being throttled?

[rui no onna]

I downloaded Geekbench 4 and and ran a test on my iPhone 8 Plus but I don’t see a section that shows what CPU speed is running at? Only gives you scores for single and multi core. CPU DasherX does show CPU speed, should have downloaded that instead or does Geekbench 4 show you it and I’m missing where to find it??

Geekbench 4 just tests performance. You still need CPUDasherX to see clock rate.

 

[Paddle1]

I downloaded Geekbench 4 and and ran a test on my iPhone 8 Plus but I don’t see a section that shows what CPU speed is running at? Only gives you scores for single and multi core. CPU DasherX does show CPU speed, should have downloaded that instead or does Geekbench 4 show you it and I’m missing where to find it??

What you can do with Geekbench is compare your score with the average score for your iPhone model to see if it's performing slowly. But for now, the iPhone 8 doesn't get throttled so there's no need for Geekbench.

 

[joetavola]

Geekbench 4 just tests performance. You still need CPUDasherX to see clock rate.

Damn new I should’ve paid for CPU DasherX instead oh well!

 

[rui no onna]

Damn new I should’ve paid for CPU DasherX instead oh well!

Lol, I got both when they were running free promotions (at different times).

 

[RPB65]

Well, I have the 6S+ and according to CPU DasherX my CPU speed changes between 1848, 1512, 1200. Anyway, when I look at the battery health it is 80%.
Geekbench on mine still returns these scores and lets be honest, I cannot tell its 'slow' as it doesn't feel it, so when Geekbench is run the CPU must be at full speed to return these scores: Single-Core 2536 and Multi-Core 4458

 

[cbautis2]

Well, I have the 6S+ and according to CPU DasherX my CPU speed changes between 1848, 1512, 1200. Anyway, when I look at the battery health it is 80%.
Geekbench on mine still returns these scores and lets be honest, I cannot tell its 'slow' as it doesn't feel it, so when Geekbench is run the CPU must be at full speed to return these scores: Single-Core 2536 and Multi-Core 4458

IMO, I don't think it's just the CPU that's being throttled. The GPU should've been throttled too. I'd add a good GPU benchmark like Antutu or 3D Mark

 

[rui no onna]

IMO, I don't think it's just the CPU that's being throttled. The GPU should've been throttled too. I'd add a good GPU benchmark like Antutu or 3D Mark

I recommend GFXBench (Manhattan). I'm also really curious on the impact of throttling on GPU as that's the more likely cause for stutters.

Afaik, Low Power mode doesn't slow down the GPU.

 

[roggymac]

I'm pretty sure CPU DasherX is free, or they have a free version. I downloaded it to test my iPhone 6, and in general I loathe paying for apps. It will provide a status on your actual CPU frequency, which can be compared to the spec. My phone is almost 3 years old, and measures 1.127 GHZ, while the spec is 1.4 GHz.

So, with the CPU speed knocked down nearly 20%, how much does that decrease the overall thruput speed? 10%? 5%?

 

[RPB65]

I recommend GFXBench (Manhattan). I'm also really curious on the impact of throttling on GPU as that's the more likely cause for stutters.

Afaik, Low Power mode doesn't slow down the GPU.

This is just the single Manhattan test, no other tests ran.
First test was run with CPU showing 1512mhz and battery just 20%, the frame rate was -5.29% when compared to the results shown in the app for a 6S+
2272 frames vs 2399 frames. So down, but not annoyingly so I don't think. Although the battery was quite low at 15% down to 11% during my 4 tests.
So, plugged in phone to charge, then ran it again, CPU still only showing 1512Mhz, result 2245 frames, so -6.42% on the example of 2399 frames on the comparison.
Ran again when battery at 32% (still charging) and CPU now back at 1848Mhz, this time got 2460 frames which is +2.48% up on the comparison.
Conclusion - CPU speed definitely affects the GPU capabilities during the test, ergo it will affect any graphical changes during normal use.
Is this enough to cause stutters? i dont know but it was interesting!!!!

 

[rui no onna]

This is just the single Manhattan test, no other tests ran.
First test was run with CPU showing 1512mhz and battery just 20%, the frame rate was -5.29% when compared to the results shown in the app for a 6S+
2272 frames vs 2399 frames. So down, but not annoyingly so I don't think. Although the battery was quite low at 15% down to 11% during my 4 tests.
So, plugged in phone to charge, then ran it again, CPU still only showing 1512Mhz, result 2245 frames, so -6.42% on the example of 2399 frames on the comparison.
Ran again when battery at 32% (still charging) and CPU now back at 1848Mhz, this time got 2460 frames which is +2.48% up on the comparison.
Conclusion - CPU speed definitely affects the GPU capabilities during the test, ergo it will affect any graphical changes during normal use.
Is this enough to cause stutters? i dont know but it was interesting!!!!

Thanks. Quite honestly, those differences seem negligible. That said, hard to tell with those numbers.

What FPS did it report for the onscreen test? That's the more important metric. The results should still be in history.

As a point of comparison:

If you just want to see raw performance, you can look at the offscreen FPS. That test runs at 1920*1080 regardless of device display resolution. Comparing the offscreen with the onscreen results gives an idea how much harder the GPU needs to work at higher resolutions.

 

[RPB65]

Thanks. Quite honestly, those differences seem negligible. That said, hard to tell with those numbers.

What FPS did it report for the onscreen test? That's the more important metric. The results should still be in history.

As a point of comparison:



If you just want to see raw performance, you can look at the offscreen FPS. That test runs at 1920*1080 regardless of device display resolution. Comparing the offscreen with the onscreen results gives an idea how much harder the GPU needs to work at higher resolutions.

The test was just Manhattan, not 3.0 or otherwise. 2460.37 frames reported as 39.6834 FPS

 

[rui no onna]

The test was just Manhattan, not 3.0 or otherwise. 2460.37 frames reported as 39.6834 FPS

Thanks. The one that says just Manhattan is 3.0.

The test seems to run 62 seconds long, so average frame rates are as follows:

2272: 36.65 FPS
2245: 36.21 FPS
2399: 38.69 FPS

Real world, that 1-2 FPS difference isn't going to be noticeable so looks like no significant GPU throttling going on.

From the online results database (Apple A9 across different resolutions):

OpenGL




Metal

​

 

[cynics]

iPhone 6S purchased Nov 16. Various battery apps report 20% wear. CPU DasherX reports current frequency 911mhz. Using there Xtreme Frequency and Bandwidth it reports lows around 890mhz and highs around 914mhz (max being 1850mhz) by repeatedly running the test.

Phone doesn't "feel" slow but maybe I've just gotten used to it. However the battery is terrible. I'm actually surprised everything tells me its 20% worn because it feels like 50% at the minimum. If it stayed the speed it is now if that improved the new batteries life that I would fine with that. Once upon a time I got 11+ hours of usage and could go a normal day without charging it.

I actually have a new battery on a shelf to replace it with.

I wont be able to do it until next weekend though. So I'll do a before and after quoting this post.