Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Intel Downplays Apple's M1 Chip With 'Carefully Crafted' Benchmarks
- Thread starter MacRumors
- Start date
- Sort by reaction score
That's the absolute bottom line. Intel is comparing a chip with something like twice the TDP, and in any case it's definitely the best they have to offer right now even, proportionately speaking, compared to their own desktop chips. And they're up against a first-outing that will be sitting at the lowest-end and lowest-power side of Apple's lineup.
Even if it outperformed the M1 across the board--which it most certainly doesn't, and you can cherry-pick benchmarks that make the i7 look worse than these make the M1 look--that would be such a meager victory it looks like a loss. "Look at us! We outperform our competitor's first ever, low-end laptop chip while using substantially more power!"
The gaming benchmark is even more desperate--they're openly admitting that they're barely even keeping up--again, with their competitor's very first laptop chip--and their entire argument is OS based. Sure, there's a lot of games you can't run on an M1--because it doesn't run Windows. Which for over 99% of people who buy a Mac was equally true when they had Intel chips in them. That's a victory for the OS that runs on your chips, not your chips.
Intel has products that can do decent things, but this just reeks of desperation.
Intel just haven't had the competition to justify any spend on packing more cores / more compute units on consumer mobile silicon. AMD were the only competitors with their G series APU. The AMD Threadripper desktop platform and non-G products have cancelled Intels dominance on desktop, but now Apple came from left field, with their non-x86 architecture, Intel are going to significantly up their game in the laptop and desktop segment. We've already seen this in the server segment where the high core count AMD EPYX series forced Intel to release higher core count (up to 28 physical cores on their Xeon Platinum series vs 64 cores on the AMD EPYX 2 7xx2 series - albeit with differing clock speed; the more cores, the lower the clock).
Intel are running behind and their lack of innovation has left gaps for others to fill. I suspect Intel will fight back but it may take a couple of years.
Intel are running behind and their lack of innovation has left gaps for others to fill. I suspect Intel will fight back but it may take a couple of years.
It's highly unlikely that the benchmarks are inaccurate. Intel isn't usually accused of providing incorrect benchmarks. Benchmarks are easy to test.
Here are a couple references. Notice that in such comparisons, AMD generally came out just a bit slower. They aren't stupid. They don't make their hardware look twice as fast, as that would draw too much attention to how the test cases were compiled and run.
In this case (involving Apple), I think even if the benchmarks are representative of real workloads, a lot of people will not care, particularly when running long computationally expensive tasks resulted in a lot of heat and noise on the intel based models, regardless of how much of that was attributable to cpu choice.
As a tldr, I'm suggesting that they're more likely to be uninteresting than falsified.
Intel's "cripple AMD" function - Agner's CPU blog
www.agner.org
Crippled No Longer: Matlab Now Runs on AMD CPUs at Full Speed - ExtremeTech
MathWorks has issued a major update for Matlab. AMD CPUs now run at full speed and use AVX2 by default if you install the 2020a update.
www.extremetech.com
Almost every review so far has shown that the M1 Macs beat comparable intel-powered laptop, yet intel thinks these reviewers are all wrong.
I couldn't find the link to intel's original article, but I suspect the 1185G7 uses several times the power of the M1 during similar workloads, and it must have a lot more cooling capacity for longer tasks. That or intel mostly used short tasks to avoid throttling.
As for the choice of tests. I suspect that using Chrome instead of safari was done on purpose. But at least it uses the same piece of software. Apparently, as PCWolrd notes, the developper of the benchmark had has a long relationship with intel. So I wouldn't call that a fair test.
As for the productivity test. Well, it's no surprise that Office runs better on Windows than on macOS. We've known that for some time. It's certainly no coincidence that Microsoft develops both Office and Windows.
Handbrake is interesting. The intel part is a bit faster. Note that both SoCs used hardware encoding. It's possible that Intel Quick Sync, which is actually quite good, is faster than the M1 hardware encoder. A thing to note though is that they're encoding a H.264 file, and that handbrake always uses software decoding (at least on macOS), for some reason. And x264 decoding has had tons of hand-crafted optimisation of X86 over the years. Not so much for ARM.
As for "Topaz Labs", this is an app that uses some Intel specialised hardware on the one end and that is not even M1 native on the other end.
As for Premiere Pro, it may indeed run better on the intel laptop. I haven't seen many benchmarks comparing the M1 to Tiger lake. Like ppt to pdf, intel likes using Premiere in their tests as they partnered with Adobe.
But the lightroom test is curious. Even on the non-M1 version, Max Tech have show the M1 to be faster than intel's TGL. I suspect Intel found the one task that is faster on their CPU. It's a very specific taks, while Max Tech uses a export of 50 photos with various filters.
As for the gaming tests. Well, all these games run under Rosetta 2... And even then, I find the Tomb Raider results suspicious, as other tests I've seen have shown the M1 to be much faster than the Intel Xe, not just 10% faster.
Did they use the built-in benchmark tool or have they managed to find a scene that minimises the difference?
As for Shadow of Mordor, this one is an openGL game. This is one of the very few games that requires tessellation, which was not available on Metal at that time. To make things worse, it uses compute shaders, which Apple's openGL never supported. So Feral had to use openCL, which was a choir to use in conjunction with openGL. Not to mention that openGL and openCL both use a translation layer on the M1. So yeah, Intel couldn't have chosen a worst case for the M1. But, hey, It could have been worse. They could have tested games under crossover.
In general, it's no secret that games run better on Windows, because they've been coded for DirectX and then ported to the Mac. Add Rosetta 2 to the mix. But to be honest, it's the simple truth that you shouldn't buy Macs for gaming.
As for the battery life test, I also suspect that intel has managed to find the one task were it can compete, because again, this is not what other testers have found.
As the PCWolrd guy says, the results are real, they're just not very representative. And why would you trust results from one vendor (Apple or Intel)? Just go check the independent reviews.
I couldn't find the link to intel's original article, but I suspect the 1185G7 uses several times the power of the M1 during similar workloads, and it must have a lot more cooling capacity for longer tasks. That or intel mostly used short tasks to avoid throttling.
As for the choice of tests. I suspect that using Chrome instead of safari was done on purpose. But at least it uses the same piece of software. Apparently, as PCWolrd notes, the developper of the benchmark had has a long relationship with intel. So I wouldn't call that a fair test.
As for the productivity test. Well, it's no surprise that Office runs better on Windows than on macOS. We've known that for some time. It's certainly no coincidence that Microsoft develops both Office and Windows.
Handbrake is interesting. The intel part is a bit faster. Note that both SoCs used hardware encoding. It's possible that Intel Quick Sync, which is actually quite good, is faster than the M1 hardware encoder. A thing to note though is that they're encoding a H.264 file, and that handbrake always uses software decoding (at least on macOS), for some reason. And x264 decoding has had tons of hand-crafted optimisation of X86 over the years. Not so much for ARM.
As for "Topaz Labs", this is an app that uses some Intel specialised hardware on the one end and that is not even M1 native on the other end.
As for Premiere Pro, it may indeed run better on the intel laptop. I haven't seen many benchmarks comparing the M1 to Tiger lake. Like ppt to pdf, intel likes using Premiere in their tests as they partnered with Adobe.
But the lightroom test is curious. Even on the non-M1 version, Max Tech have show the M1 to be faster than intel's TGL. I suspect Intel found the one task that is faster on their CPU. It's a very specific taks, while Max Tech uses a export of 50 photos with various filters.
As for the gaming tests. Well, all these games run under Rosetta 2... And even then, I find the Tomb Raider results suspicious, as other tests I've seen have shown the M1 to be much faster than the Intel Xe, not just 10% faster.
As for Shadow of Mordor, this one is an openGL game. This is one of the very few games that requires tessellation, which was not available on Metal at that time. To make things worse, it uses compute shaders, which Apple's openGL never supported. So Feral had to use openCL, which was a choir to use in conjunction with openGL. Not to mention that openGL and openCL both use a translation layer on the M1. So yeah, Intel couldn't have chosen a worst case for the M1. But, hey, It could have been worse. They could have tested games under crossover.
In general, it's no secret that games run better on Windows, because they've been coded for DirectX and then ported to the Mac. Add Rosetta 2 to the mix. But to be honest, it's the simple truth that you shouldn't buy Macs for gaming.
As for the battery life test, I also suspect that intel has managed to find the one task were it can compete, because again, this is not what other testers have found.
As the PCWolrd guy says, the results are real, they're just not very representative. And why would you trust results from one vendor (Apple or Intel)? Just go check the independent reviews.
Last edited:
Let's see. Intel pushed their various Pentium processors and the PPC604/604e was faster.
The Pentium 4 figuratively hit a brick wall, and they discontinued it, going back a step back to re-design.
What they did not tell is, that their foolish CPU design is the reason why Apple ditched them. No real performance upgrades since 2015. At second it is a shame that they had to switch the CPU for the Battery test. Noteworthy to mention, that the Acer Swift 5 has a 7Wh bigger battery capacity than the MacBook Air. So all in all it is true, the M1 is the best around consumer CPU on the market.
But on the other hand, let's see it from Intels perspective. There is a small Company, which was nearly death and then they designed a Chip which "Ryzes" them to the top of the Perfomance tables. Now Intels really under pressure.
But on the other hand, let's see it from Intels perspective. There is a small Company, which was nearly death and then they designed a Chip which "Ryzes" them to the top of the Perfomance tables. Now Intels really under pressure.
It’s just the case of ultimate cherry picking. They have looked wide and hard until they could find some marginal tests where Mac versions of the software is poorly optimized. I would t be surprised if the “ppt to pdf” test for example used a carefully crafted example that just happen to tank on the Mac office version.
At any rate, there is not much to talk about here. These “benchmarks” are a sad joke and Intel is already being called out on this.
Yes, nobody said that, so why did you pretend somebody had when you know better?
To be honest. Who cares. If you use a Mac. Why worry about what intel says.
the cheerleading here is ridiculous at times. Apple isn’t going anywhere and either is Intel.
The tribalism is strong in this thread. Not very evolved, is it?
They often use this test to compare their CPUs to AMD's. I'm not sure how it favours intel CPUs, but it certainly does.
On a side note, PowerPoints generally sucks at converting slides to pdf. It cannot make a different page for each animation, which is a must-have feature.
Apple won't directly make this kind of hardware but can push the industry in the direction of Arm, especially since x86 is used more for legacy reasons than for performance. And a lot of things you're naming here already use Arm or PPC, especially those microcontrollers that definitely won't be x86-based. The often cited "fastest supercomputer" uses Arm CPUs.
For now you have the choice between M1 and Intel Macs, so it matters somewhat.
Apple‘s language might have been very vague, but in retrospect, all their claims have been fairly accurate. In terms of processor performance, M1 is up there with the fastest currently available desktop chips that consume 5-8 more power.
The fastest desktop chips have like 64 cores with SMT. The M1 cannot compete.
Unless you mean for single-threaded taks, but this is not what these multicore CPUs are made for.
At those tasks, I think the M1 consumes about 2-3x less power than AMD Zen 3.
Made a mistake ... it is an internal PCI-E 4 ssd m.2 -- the user can replace it. The pci-e 4 samsung SSD 980 Pro 2TB speeds up to 7000 mb/s read and 5000 mb/s write.
All it takes is to try one out for about 10 minutes to see that M1 is very noticeably faster at almost everything compared to a similarly priced Intel option. Add to that the fact that it does not get warm, no fans run, and battery life feels like twice as much with the same sized battery, and I don't know what else there is to prove.
If this is the strategy of the new Intel CEO, then I expect really hard times to intel in the next couple of years. I think that they need to ditch core i series, time to switch to a new name and product to make people interested again. Just copy what Apple is doing, simplify the line and instead of having dozens of variants of each series, then focus in one processor that is “powerful efficient” and name it with a cool name.. it’s time to take drastic decisions in my opinion !