Apple A17 Pro SoC Within Reach of Intel i9-13900K in Single-Core Performance

[T0@st]

An Apple "iPhone16,1" was put through the Geekbench 6.2 gauntlet earlier this week—according to database info this pre-retail sample was running a build of iOS 17.0 (currently in preview) and its logic board goes under the "D83AP" moniker. It is interesting to see a unit hitting the test phase only a day after the unveiling of Apple's iPhone 15 Pro and Max models—the freshly benched candidate seems to house an A17 Pro system-on-chip. The American tech giant has set lofty goals for said flagship SoC, since it is "the industry's first 3-nanometer chip. Continuing Apple's leadership in smartphone silicon, A17 Pro brings improvements to the entire chip, including the biggest GPU redesign in Apple's history. The new CPU is up to 10 percent faster with microarchitectural and design improvements, and the Neural Engine is now up to 2x faster."

Tech news sites have pored over the leaked unit's Geekbench 6.2 scores—its A17 Pro chipset (TSMC N3) surpasses the previous generation A16 Bionic (TSMC N4) by 10% in single-core stakes. Apple revealed this performance uplift during this week's iPhone "Wonderlust" event, so the result is not at all surprising. The multi-score improvement is a mere ~3%, suggesting that only minor tweaks have been made to the underlying microarchitecture. The A17 Pro beats Qualcomm's Snapdragon 8 Gen 2 in both categories—2914 vs. 2050 (SC) and 7199 vs. 5405 (MC) respectively. Spring time leaks indicated that the "A17 Bionic" was able to keep up with high-end Intel and AMD desktop CPUs in terms of single-core performance—the latest Geekbench 6.2 entry semi-confirms those claims. The A17 Pro's single-threaded performance is within 10% of Intel Core i9-13900K and Ryzen 9 7950X processors. Naturally, Apple's plucky mobile chip cannot put up a fight in the multi-core arena, additionally Tom's Hardware notes another catch: "A17 Pro operates at 3.75 GHz, according to the benchmark, whereas its mighty competitors work at about 5.80 GHz and 6.0 GHz, respectively."



View at TechPowerUp Main Site | Source

 

[aktpu]

Geekbench? Yawn

 

[cowcatwithahat]

It is absolutely incredible to me how powerful and efficient a phone is.
I feel so small in comparison with those brilliant engineers that work on such projects.

 

[AnotherReader]

There's a lot of variance in the scores of the 7950X and the 13900k. Just looking at the first couple of pages of results, the 7950X has validated scores ranging from 2858 to 3253 while the 13900k falls between 2993 and 3475. Nevertheless, Apple's cores are unmatched for IPC and 3.75 GHz is a pretty high speed for their extremely wide cores, but this clock speed suggests that the performance increase is entirely due to the higher clock speed.

 

[R0H1T]

Meh, let's try again with quad channel LPDDR5T @9600 MT/s & 3nm nodes!

 

[Wirko]

Meh, let's try again with quad channel LPDDR5T @9600 MT/s & 3nm nodes!

Yeah, Intel really should try that.

 

[Nanochip]

Set the 13900K to 3.77 GHz and see who comes out on top.

 

[AnotherReader]

Set the 13900K to 3.77 GHz and see who comes out on top.

In that case, the A17 would win by a landslide in single threaded performance. Still, the clock speed is a part of the design and equalizing clocks doesn't make sense for two very different designs.

 

[Aquinus]

Meh, let's try again with quad channel LPDDR5T @9600 MT/s & 3nm nodes!

It helps a lot when memory is soldered to the board and is physically a lot closer to the CPU cores.

 

[GreiverBlade]

looking at the last pic ... i am more impressed at the SD8 Gen 2 results ... which is also clocked 160mhz higher on the prime Cortex X3 core than that in my S23 (due to the "For Galaxy" edition )

knowing that it's out since some time now ... less than 900pts more single core, while clocked .57mhz higher, is a bit "yawn" but +1794pts in multi is not bad tho ... (not that it's needed in a phone ... the SD8G2 feel already overkill


edit, 3.19? the std SD8G2 is 3.2 on the X3 (3.36 in the "For Galaxy" variant) bah, not that 0.01ghz is an issue

 

[Wirko]

It helps a lot when memory is soldered to the board and is physically a lot closer to the CPU cores.

The absence of connectors helps too.

 

[alwayssts]

I understand that it appears Apple chose greater IPC xtors in some respects for this arch, but it's worth stating they also generally tend to favor max clock speed improvements for the cost savings. In that respect the improvement from N4P->N3B is less than 9%? I suppose it beats the earlier rumblings of straight 3.7ghz, which would have been only ~7%, but still not anything to write home about. I can understand why Apple may be upset by the results (as has been the rumor for some time).

Not a slight toward the chip at all, rather the process (the mobile-leaning density over speed improvement); That's pretty awful. I'm sure N3E will help eek out the required generational performance (twice the clock improvement of n3b at the cost of some density) but one has to factor in the cost/fab space and question for how many products it will truly be worth it (at least until the price comes down or fab allocation becomes less of an issue).

Look at TSMC's press release for N4X, which claims "4% improvement over N4P at 1.2v", which, to save you some time, implies N4X can hit 3600mhz at 1.2v with the capability of being driven even harder. How hard, I don't know know, but I would imagine at least 1.3v (think of chips [cpus/RV790] in the past that were ~1.3125v) isn't out of the question, maybe even 1.35v (which appears to be the Fahrenheit 451 for this type of use case; think about AMD's SOC voltage problems).
Personally, I always think back to ~1.325v being the sweet spot for many chips over several different nodes and generations. Ye ole' 4.7ghz that Intel and AMD were stuck at for some time, granted that's anecdotal.

It's also worth noting, while it has often been lost in the shuffle of the news cycles, that it has been said N4X has been able to achieve higher clockspeeds than N3B, which we now have some context. So probably between ~3.7-4.0ghz. It could potentially perform very similar to n3e just with worse density and power characteristics.

So, just think about that for a hot second. I know this is a space for talking about Apple and the A17, but I will be that guy that interjects about other uses.

Think about N31/N32, running at 1.1v and 1.15v respectively on N5, or nVIDIA's tuning/amalgamation of the process (that probably is denser but limited to ~1.08v for optimal leakage scaling on N5) and their intended power/clock constraints (which I will grant some have circumvented for N31 and the 7700xt can some-what do within it's allowable max TDP).

At some point I just have to wonder if N3E is worth it at any point in the near future. I say this because if you look at something like N32, which I would argue is bandwidth balanced at ~2900/20000, you could very realistically (if AMD can fix whatever gaming clockspeed problems exist in RDNA3) jack that up to ~3500/24000 on N4X and literally replace the 7900xt/compete with the 4080; not a bad place to be cost/performance-wise even if they lose some power efficiency. The argument I will continue to make is that once a card is greater than 225w (one 8-pin connector, easy to cool, cheap PCBs/caps/etc can be used) you may as well use the headroom up to at least 300w, if not 375w, which N32 currently does not. If would be one thing if these cards were shorter/thinner like mid-range cards of yore, but many of them are engineered to the level of a 7900-level card already, with hot spots not exceeding 80c. That gives them ~20% heat headroom assuming the same designs. Assuming N4X can scale well to those higher voltages, it appears a no-brainer IMHO.

N31 is a little more nuanced, as 24gbps memory probably would probably limit the card below 3300mhz and still play second-fiddle to 4090, even if it halves the gap. What they need in that case (regardless of memory used) is to double up the ∞$ to make it similar to nvidia's L2 bandwidth. While I have no idea if they plan to continue to use MCDs going forward into the far future, two things remain true: the cache can and needs to be doubled, and it should transition to 4nm when economically feasible. Even on 6nm, the (16MB+MC) MCD is 37.5mm2. The 64MB on X3D CPUs is ~36mm2. This implies they could add 16MB of cache while staying within the parameters of why the design makes sense. That is to say, a MCD has to be less than the smallest chip with that size bus (ex: 256-bit and <200mm2; hence a MCD should be less than 50mm2). I understand the rumors have always pointed to stacking (and maybe GDDR6W), but wonder if that's truely necessary. With that amount of cache they could even use cheaper memory, as theoretically doubling the L3 alone should allow it to scale to the limits of the process. It could be a true 4090 competitor. Sure, it'd probably use a metric ton of power by virtue of 18432/16384 at lower voltage/clocks will always be twice the proportional efficiency than 12288 at higher, but we are living in a world with a freaking 600W power cable (+75w from the slot). 3 8-pin is also 675w. If that's where we're going, we might as well freaking do it. Nobody will force you buy it, but one thing is for sure...It would be a hell of a lot cheaper up-front.

That's it. That's my spiel. I know we keep hearing about N43/44 (and perhaps being monolithic), and that's fine. Maybe they are (similar to) the things I ponder, maybe they aren't. But in a world where a 4090 is at least $1600, a 4080 $1100, and a 4070ti $800, while the competing stack (7900xtx/7800xt/7700xt) is/will be soon literally half the price, I certainly wouldn't mind a refresh on the later to compete with the former. It may sound incredibly strange, but it's absolutely possible, as it could be AMD doing what AMD does: competing one tier up the stack for their price. EX: a 4090 competitor for the price of a 4080, a 4080 competitor for the price of a 4070ti (whatever those prices are by the time those products could launch).

In a world where an ~7800xt will probably power the PS5 pro, which appears to content most people (~4k60), and that will last us another 3-5 years, and the PS6 maybe 2-2.25x that (perhaps similar/a little faster than a 4090) lasting another 4-8 years AFTER THAT, I have to ask if we truly need a $2000 3nm nvidia part, whatever it's capabilities...or do we need what I described; the smallest 256-bit/384-bit parts that can achieve their highest performance potential at the most economical price if they are in the realm of achieving that level of performance. Because of the slowing of process tech, the idea of long-term future-proofing at the high-end, just like the 'mid-range' with the 7800xt, at a relatively cheaper price isn't completely out of the question if one of the companies (which won't be nVIDIA) decides to make it so.


Source: N4X press release: https://pr.tsmc.com/english/news/2895

 

[WhitetailAni]

An Apple "iPhone16,1" was put through the Geekbench 6.2 gauntlet earlier this week—according to database info this pre-release sample was running a build of iOS 17.0 (currently in preview) and its logic board goes under the "D83AP" moniker. It is interesting to see a 16-series unit hitting the test phase only a day after the unveiling of Apple's iPhone 15 Pro and Max models—the freshly benched candidate seems to house an A17 Pro SoC as well.

@T0@st iPhone16,1 is a 15 Pro or Pro Max. The iPhonex,x numbers don't match up to the series number. The existing 14-series are iPhone15,x, 13 series was iPhone14,x, etc. The reason for the mismatch is the S models - the iPhone 6 is iPhone7,x, 6S is iPhone8,x, 7 is iPhone9,x, etc. It was partially realigned with the 8 and X being the same numbers (10,x) and the XS not increasing the series but moving to 11,x. However it means that the 11 series is 12,x and so on.

 

[mechtech]

But can it play windows??

 

[The Norwegian Drone Pilot]

And still iOS is handicapped to the maximum compared to what Android can do. Why have such power when Apple doesn't want to use the power the CPU can deliver in many things?

Yeah sure, the ProRes video recording takes A LOT of power and gaming might also use a lot of power. But still.....

 

[zatakan]

I have somehow limited knowledge of CPU design but isn't this whole Apple SOCs' uplift of single-core performance, single-core performance per watt coming from their huge cache, frequency uplift from improved process nodes, and raised frequency when they were able to attach a cooler or a fan? AMD's 7945HX gets M2 max Cinebench R15 multi-core performance per watt with a third of the transistor count (yeah I know GPU and neural engine adds to the transistor count). Which should cost Apple a lot.

source: https://www.notebookcheck.net/M2-Max-vs-R9-7945HX_14975_14936.247596.0.html

 

[Minus Infinity]

Pretty pathetic that tech sites still refer to the utterly useless Geekbench which has virtually zero relevance to real world applications.

 

[matar]

WOW No doubt A17 Pro is extremely fast,
But the i9-13900k if it is tested on iOS (I know it won't work but just to compare performance) then it will blow the A17 Pro , note the 13900k is tested on Windows AND apple make the CHIP and IOS so the A17 Pro are 100% optimized for iOS. And the same goes for A17 PRO let say it works on windows then you will see way way less of a score.

 

[Nuckles56]

The big thing that everyone seems to be forgetting is the additional width in the decode block the A17 likely has for throughput vs Intel and AMD. I remember when Anandtech did the deep dive on the A14 in 2020, it had an eight wide decode block vs intel and AMD having 4 max. Even if the x86 chips can't change, I fully expect that apple has made their CPU design have a 12 or 16 wide decode block now to keep going with their big and wide over max frequency strategy.

 

[JustBenching]

Set the 13900K to 3.77 GHz and see who comes out on top.

Set the Ferrari to 100khm/h and my fiat panda is faster.

 

[FoulOnWhite]

Yeah because you need a 13900k in your phone.

 

[Space Lynx]

Yeah because you need a 13900k in your phone.

yeah its a bit silly really, would be better if they just cut its power in half to run what needs to run with no lag or skips, then get extra battery life that way. i know they already do that to some degree, but I mean take it to the next level.

 

[R0H1T]

The big thing that everyone seems to be forgetting is the additional width in the decode block the A17 likely has for throughput vs Intel and AMD. I remember when Anandtech did the deep dive on the A14 in 2020, it had an eight wide decode block vs intel and AMD having 4 max. Even if the x86 chips can't change, I fully expect that apple has made their CPU design have a 12 or 16 wide decode block now to keep going with their big and wide over max frequency strategy.

clamchowder is doing a fine job here ~

Hot Chips 2023: Characterizing Gaming Workloads on Zen 4

AMD didn’t present a lot of new info about the Zen 4 core at their Hot Chips 2023 presentation.

chipsandcheese.com

While they're probably not as renowned as AT even 2 decades back, but from whatever I can understand their analysis is top notch!

 

[ixi]

But can it play windows??

I doubt you want one high perf cpu or at best 2 cores which can handle high boosts and temperature for short period .

 

[silentbogo]

The only question is why... I don't really give a crap about computational photography and AI/ML in my pocket. Running compute-intensive apps on the phone as-is is also stupid(from usability perspective). Having full-blown AAA games on the go, running natively, isn't that impressive these days either. All it needs is a proper dock with a proper desktop mode, at minimum like DEX. Then we'll talk.