Thread: iPhone 15 / Apple Watch 9 Event

[exoticspice1]

Really disappointed. The GPU is worse than Snapdragon gen 2 and that's a year old chip.
The CPU is more powerful but way less efficient than A16.

Not looking good for Apple

https://twitter.com/Golden_Reviewer/status/1704126470032114147?t=id_KW166UhRbIVfAaUjTTg&s=19

 

[amonduin]

Really disappointed. The GPU is worse than Snapdragon gen 2 and that's a year old chip.
The CPU is more powerful but way less efficient than A16.

Not looking good for Apple

https://twitter.com/Golden_Reviewer/status/1704126470032114147?t=id_KW166UhRbIVfAaUjTTg&s=19

From what I understand the Snapdragon GPU is very specialized compared to apple's GPUs. Where Apple has taken its GPUs more in the direction of general purpose compute Qualcomm continues to manly focus on optimizing for games. This, IIRC is reflected in the compute scores of, say the M2 vs the snapdragon. The M2 gets 26,000 on OpenCL while the snapdragon only gets 8,000. Compared to the M2 the A17 offers about 61% of the GPU performance so rough napkin math means the A17 would score around 16,000 on Open CL (if such a benchmark existed).

 

[exoticspice1]

From what I understand the Snapdragon GPU is very specialized compared to apple's GPUs. Where Apple has taken its GPUs more in the direction of general purpose compute Qualcomm continues to manly focus on optimizing for games. This, IIRC is reflected in the compute scores of, say the M2 vs the snapdragon. The M2 gets 26,000 on OpenCL while the snapdragon only gets 8,000. Compared to the M2 the A17 offers about 61% of the GPU performance so rough napkin math means the A17 would score around 16,000 on Open CL (if such a benchmark existed).

From what I understand the Snapdragon GPU is very specialized compared to apple's GPUs. Where Apple has taken its GPUs more in the direction of general purpose compute Qualcomm continues to manly focus on optimizing for games. This, IIRC is reflected in the compute scores of, say the M2 vs the snapdragon. The M2 gets 26,000 on OpenCL while the snapdragon only gets 8,000. Compared to the M2 the A17 offers about 61% of the GPU performance so rough napkin math means the A17 would score around 16,000 on Open CL (if such a benchmark existed).

The Snapdragon uses Vulkan which is more game oriented yes but Qualcomms GPU is faster while being on an old arch, on an older node and also it's more efficient than A17.

At this rate Apple is not even surpassing Qualcomm in efficiency.

 

[amonduin]

The Snapdragon uses Vulkan which is more game oriented yes but Qualcomms GPU is faster while being on an old arch, on an older node and also it's more efficient than A17.

At this rate Apple is not even surpassing Qualcomm in efficiency.

It isn't Vulkan vs Metal that I'm comparing but rather a question of general compute flexibility. If Qualcomm is only faster in game engines by virtue of specialization and tailoring their GPU pipeline to only those features that game engines require then the GPU capabilities won't translate to general purpose compute. Apple needs good general compute performance in their GPUs as they are not just focussed on the mobile game market but are also pitching their GPUs to the GPGPU market in their Mac line.

 

[exoticspice1]

CPU seems 9 wide decode. All in all TSMC 3NB is a bad node

 

[Jimmyjames]

The Snapdragon uses Vulkan which is more game oriented yes but Qualcomms GPU is faster while being on an old arch, on an older node and also it's more efficient than A17.

At this rate Apple is not even surpassing Qualcomm in efficiency.

Hmmm. I don’t think it’s accurate to say 8gen2 is faster. Every test I’ve seen says the opposite. Given the speed of this review, my guess is geekerwan are using a pre-production model, or a beta version of ios17.

Curious for sure though.

 

[Cmaier]

Lol

CPU seems 9 wide decode. All in all TSMC 3NB is a bad node

 

[Jimmyjames]

Lol

The speed with which people are willing to dump all over the A17 Pro and N3B. Let’s wait and see.

 

[Aaronage]

All this talk makes me sad we won’t get a review from Andrei Frumusanu

 

[dada_dave]

Really disappointed. The GPU is worse than Snapdragon gen 2 and that's a year old chip.
The CPU is more powerful but way less efficient than A16.

Not looking good for Apple

https://twitter.com/Golden_Reviewer/status/1704126470032114147?t=id_KW166UhRbIVfAaUjTTg&s=19

All this talk makes me sad we won’t get a review from Andrei Frumusanu

I remember Andrei fighting with such reviewers on how to properly measure power consumption on mobile devices during such tests. Still, I won’t definitively say one way or the other until more testing comes out.

 

[dada_dave]

The Snapdragon uses Vulkan which is more game oriented yes but Qualcomms GPU is faster while being on an old arch, on an older node and also it's more efficient than A17.

At this rate Apple is not even surpassing Qualcomm in efficiency.

Vulkan is not more game oriented than Metal, each API has strengths and weaknesses but using Vulkan does not intrinsically net you any more performance than Metal.

Very suspicious of these results in general.

 

[Jimmyjames]

Hmmmm. Riiiiight.

 

[Jimmyjames]

Damn…that did concern me at the time

Thread: iPhone 15 / Apple Watch 9 Event

Another thing to keep in mind is that the M3 will likely have the same cores as the A17 while the M2 had the same cores as the A15, so the M3 ST perf could effectively get a near 30% jump in performance over the M2, which I think is what @Cmaier was predicting? Though perhaps weighted more...

techboards.net

 

[Yoused]

Can’t you iMessage them?

I never thought of that because, quite honestly, I never use iMessage.

 

[dada_dave]

Damn…that did concern me at the time

Thread: iPhone 15 / Apple Watch 9 Event

Another thing to keep in mind is that the M3 will likely have the same cores as the A17 while the M2 had the same cores as the A15, so the M3 ST perf could effectively get a near 30% jump in performance over the M2, which I think is what @Cmaier was predicting? Though perhaps weighted more...

techboards.net

View attachment 26013

MaxTech can’t help but oscillate between everything is amazing and everything is awful.

It is mildly interesting that the phone can reach that high score under aberrant cooling conditions, but I’m not sure that’s particularly relevant for the kind of throttling we normally worry about. Nor was that level of performance what was promised by Apple.

 

[Jimmyjames]

MaxTech can’t help but oscillate between everything is amazing and everything is awful.

It is mildly interesting that the phone can reach that high score under aberrant cooling conditions, but I’m not sure that’s particularly relevant for the kind of throttling we normally worry about. Nor was that level of performance what was promised by Apple.

Indeed. They are wild! One minute it’s terrible, the next it’s brilliant. I do find the geekerwan results curious though.

 

[dada_dave]

Indeed. They are wild! One minute it’s terrible, the next it’s brilliant. I do find the geekerwan results curious though.

Absolutely. Definitely interesting and worthy to dig into.

 

[theorist9]

Here's the specifics from the Geekerwan video that MaxTech was referring to. When they check for thermal throttling by removing the cooling attachment, they see the A17 Pro throttling during a SPEC 2017 test. They note that, when the A17 throttles, it drops down to the A16's frequency, yet shows the same power consumption as the A16, in spite of the smaller process. But remember that the 3 nm's efficiency benefit is relatively small; plus it's probably optimized to run at the higher frequency. So it's not surprising the difference isn't evident when rounding power consumption figures to one significant figure.

They also show non-throttled comparative SPEC 2017 results (score vs. wattage) for both the efficiency and performance cores across four generations. They say this is the first time they've been able to measure the efficiency cores separately.

 

[Cmaier]

Here's the specifics from the Geekerwan video that MaxTech was referring to. When they check for thermal throttling by removing the cooling attachment, they see the A17 Pro throttling during a SPEC 2017 test. They note that, when the A17 throttles, it drops down to the A16's frequency, yet shows the same power consumption as the A16, in spite of the smaller process.

I’m confused. Why wouldn’t it do that? P=1/2 CfV^2. F is the same. V is the same (I think? Do we know?). ½ is the same. The only difference is C, which goes down for transistor fan out loads, and probably goes down a little for wire loading (but not as much because the wire height doesn’t scale the same as width, and the width doesn’t scale the same as transistors). More importantly, there are more capacitors this time, because there are many more wires and many more transistors.

 

[theorist9]

I’m confused. Why wouldn’t it do that? P=1/2 CfV^2. F is the same. V is the same (I think? Do we know?). ½ is the same. The only difference is C, which goes down for transistor fan out loads, and probably goes down a little for wire loading (but not as much because the wire height doesn’t scale the same as width, and the width doesn’t scale the same as transistors). More importantly, there are more capacitors this time, because there are many more wires and many more transistors.

I was thinking of tables showing the smaller process is expected to give reduced power consumption for the same performance, everything else being equal. I thought this meant that, if you took the A16 architecture, and were able to do an exact port from N4 to N3, it should show lower power consumption at the same frequency. That would be akin to what Intel did with a "Tick": Same architecture, newer process.

In this case, though, you have the confounding factor that the architecture isn't the same. So it seems the only way you'd get the same (rather than lower) power consumption at the same frequency on a smaller process would be if, for the same frequency, the new architecture itself caused an increase in power consumption.

And if so, is it typical that newer (i.e., more advanced) architectures draw more power for the same frequency? [For instance, when Intel did a "tock" (putting a newer architecture onto the existing process).]