Benchmark results show that the Apple M1 beats any Intel-based MacBook...

The Apple M1 launched earlier this week, and the benchmark results for the new chip are already starting to flow. Although these first tests are preliminary, they paint a very positive picture for the upcoming CPU from Apple.

In Geekbench 5, an Apple M1 can achieve a 1687 single-thread and a 7433 multi-thread score. For comparison, the A14 in an iPhone hits 1598/3995. This increase in multithreaded performance is likely due to the higher number of cores (+2 high-performance Firestorm CPU cores) and the almost certainty of consistently higher clock speeds in its larger chassis. In the past, iPhone CPUs were severely throttled due to the high power consumption. So what we’re seeing here could mean the A14 is finally stretching its legs.

These values ​​can be compared very well with existing Intel and AMD CPUs on the market. As always, I recommend caution in extrapolating synthetic tests to actual performance. Plastics are useful for low-level feature analysis and can reveal interesting differences in microarchitecture, but we don’t rely on them as our primary means of evaluating hardware.

The high-end 16-inch MacBook Pro that Apple has ever sold features a Core i9-9980HK, an 8C / 16T CPU with a base clock of 2.4 GHz and a 5 GHz boost. The measured performance for this CPU is 1149/7379 in GB5. This is an older 14 nm CPU. Intel does not currently ship an eight-core Ice Lake mobile CPU, but Apple also ships Intel’s quad-core i7-1068NG7. This chip reaches 1352/4914. Apple is therefore the leader in single-thread performance, even compared to Intel’s Sunny Cove architecture from 2019, and leads in multi-thread performance against the most powerful CPU cores that Intel supplies for mobile devices . While Apple only binds the Core i9-9980HK with multiple threads, the Geekbench scores of the M1 beat anything below this point, including the Core i9-9880H.

Since people will be curious about AMD, here’s how things compare. GeekBench 5 clearly likes Zen 3 a lot more than Zen 2; According to LegitReviews in GeekBench 5, the Ryzen 7 3700X scores 1281/8220, while the Ryzen 7 5800X scores 1673 / 11.246. This is a 1.3x and 1.36x improvement, respectively, which is well above the gains AMD expects on average from Zen 2 -> Zen 3 (1.19x IPC and up to 1.25x with IPC and Clock). Obviously, however, these CPUs are both desktop processors. We don’t know how AMD’s mobile Zen 3 cores compare to the desktop chips, but the Ryzen 4800U hits ~ 1130 in single-threaded GB5 and ~ 6800 in multi-threaded. That’s an average – we’ll be referring to this Lenovo system specifically when comparing performance.

In summary, the brand new Zen 3 8 core from AMD is the same as the M1 in terms of single-threaded performance and significantly outperforms multi-threaded performance. However, both results are achieved in terms of power consumption at the desktop level. Data from the comparison suggests that Zen 3 is significantly cheaper than Zen 2, but data also shows that AMD’s mobile performance in a 15W envelope is much lower than allowed desktop power levels, which only makes sense . Assuming that AMD can improve single-threaded performance 1.3x and 1.36x – meaning they can get the entire increase in exactly the same performance envelope – our hypothetical zen is reached 3-based mobile CPU 1495/9248. Even if we do this very Optimistic assumptions: The AMD mobile system would still use a lot more power than its Apple counterpart.

Subtest results

The Intel Core i9-9980HK loses almost every single-thread subtest on the Apple M1, but does not lose it evenly. In some cases the gap is small; In terms of text compression, the M1 scores 1292 points, compared to 1177 for the 9980HK. The Gaussian blur, face detection and horizon detection tests favor the M1 just as dramatically as the ray tracing test.

The performance of subtests versus the AMD 4800U differs from Intel in which tests Apple wins and at what margins, but the overall score doesn’t differ significantly. The Ryzen 7 4800U from AMD can achieve 1318 faces / second compared to the 750 from Intel, but the M1 reports 2209. The 9980HK from Intel achieves 1218 points in the N-body physics test compared to the 936 from Ryzen, the Apple M1 but 1769. Interestingly, Intel retains the lead on the machine learning subtest with a score of 1332 versus 1169 on the M1 and 965 on the 4800U.

Multi-Threaded vs. Single-Threaded

Multithreaded tests show some differences from the ST numbers. In ST, the M1 swept both Intel and AMD almost all along the line. With multithreading, it’s a lot more of a struggle. The M1 wins AES-XTS compression by a country mile, but Intel leads the way in text compression with 8284 to 7162 (AMD) to 5528 (Apple). AMD wins the image compression test with 10,392 compared to Intel with 9,524 and Apple with 7,213. AMD just wins the PDF rendering, while the M1 wins the text rendering over both x86 competitors. The camera benchmark performance from Intel is slightly higher than that from AMD or Apple and wins the rigid body physics test decisively. However, Intel’s scaling in the machine learning test is terrible, and Apple wins that test in multithreading.

Apple M1 multi-threading results according to GeekBench 5.

What does this tell us about M1 scaling compared to Intel and AMD chips? It is unclear. The chip’s performance may still be limited by the thermals or the split-core configuration used by Apple, or both. The scaling factors for the AMD Ryzen 7 4800U, the Intel Core i9-9980HK and the Apple M1 are 6.0x, 6.38x and 4.4x, respectively. Given that Apple operates with a thermally constrained environment and a mix of high-performance and high-performance cores, this is not too surprising. It’s also why the M1 won’t just destroy x86’s market position in the next year or two – Intel and AMD are both more competitive than ST in MT, and there are more multi-threaded workloads involved these days. These results pose a real threat to x86, but they’re not the death rattle, even if they translate well to real world applications.

Multi-threaded Intel Core i9-9980HK subtest performance, GeekBench 5.

It would be inappropriate to draw conclusions from the results of a single synthetic test, but nothing in those results implies good things for Intel or AMD. While GeekBench 5 does particularly well on Apple hardware or ARM chips in general, we’re still seeing a CPU with a nominal TDP of 5W that will pick up chips with nominal TDPs of 15W and make an early winner.

It’s entirely possible that continued testing under load shows that the M1 will have to give it a lot of throttle over time because it lacks a fan, and that Apple’s high-end MacBook pros will still outperform it in real world tests. This is actually the result I’m expecting because while Apple takes a lot of price tag for its pricing, professional customers aren’t going to buy $ 2,000 to $ 3,000 Intel Core i9 laptops if they get better performance in audio with much less – And video rendering suites can achieve expensive machine.

GeekBench 5 is not a real benchmark and we are waiting for real testing before drawing any conclusions. However, nothing in these results indicates weaknesses in the architecture, and none of the partial values ​​indicate problems that AMD or Intel can immediately exploit. Considering this is a 4 + 4 CPU that competes against full octa-core solutions, there is plenty of headroom for GeekBench 5 to outperform Apple without changing the score: the M1 seems at least to be very competitive with x86 core-for-core and clock-for-clock. Further testing will determine to what extent this is true, or, conversely, find that GB5 is too Apple-friendly to create a useful cross-platform benchmark.

With tests like Cinebench R23 now baking in support of M1, we will have an answer to that question sooner rather than later.

In addition, the multithread subtests – just to take another look at the “98 percent of PC laptops” bit – actually show why this is so is not true. “Faster” has a broad contextual meaning, and the Core i9-9980HK is way above the M1 in terms of text compression, image compression, SQLite, PDF rendering, clang, N-body physics, rigid body physics, and HDR subtest. Even within a benchmark that shows that Apple’s M1 is exceptionally well lit, there are different workloads where the Core i9-9980HK is not only competitive, but wins with a lead that is sufficient for lower-priced Intel CPUs to be good Have chances – also against the M1.

For the same reason, Apple’s marketing overshoot of this stupid claim should not lead anyone to ignore this chip. It is the most powerful threat to x86 dominance that I have seen in my entire career.

Read now::