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
Just wait till the M1X SoC with (rumored) eight high-performance CPU cores, four high-efficiency CPU cores, a revised Neural Engine core, and possibly 16 GPU cores, all closely linked to 32 or even 64 GB of RAM. It may be so fast that game publishers may actually write "Apple Silicon" versions of AAA games to take advantage of the processing efficiency offered by the M1X.
The bear case is strong, true. But the company has new leadership and the involvement of a famous activist investor (who is pressuring for "strategic alternatives.") The company is at a juncture ... much like Apple was 25 years ago except Intel is in a much stronger financial position than Apple was at that time. It is extremely unlikely that the company goes bankrupt any time soon.
There have been cases of benchmarking shenanigans in these circles. I am not saying that is the case in this instance, but it happens.
When ATI (pre AMD buyout) released the Radeon 2, they had special code in the drivers to tweak the graphics settings (Specifically antialiasing and mipmapping among others) to get higher frame rates when running Quake 3. If you recompiled Quak3 3 to Quack 3, you would get substantially reduced performance in what was a commonly used benchmark at the time.
Intel convinced 3DMARK (I think, I am going from memory), to disable MMX support on non-Intel processors, even when AMD had full MMX support.
The also benchmarked things like how fast Photoshop could blur 4 pixels, because at values higher than that, AMD was quite comparable.
So it certainly has happened in the past.
The last time probably was in 2002 during the late G4 era when Apple solely relied on photoshop benchmarks. And even then, they did so during Apple events. I'm not sure they sent results to websites like intel just did.
Before that (G3, early G4), they were using synthetic tests (something called basesmark I think). These were probably favourable to the PPC architecture, but at least at that time the PPCs were genuinely faster than the pentiums.
When they introduced the G5, they showed results from SPEC tests, which are industry standards and not cherry-picked tasks like what intel has shown here.
But the most desinguenous thing from intel are:
- showing results from tests relying on very different pieces of code, i.e. the macOS version versus windows version on an app that most likely relies on system routines. That's why standard benchmark like SPEC or Geekbench exists, they do execute the same code regardless on the OS. And that's certainly no the case here. It's even more problematic when one app runs under Rosetta, or is in Beta (Premiere) or is developed by the OS vendor (Office), or is a port from a Windows app (games).
- showing results of apps using dedicated hardware on the intel SoC and the regular cores on the M1 (Topaz, which also runs under Rosetta, AFAIK).
- Probably selecting the specific tasks within those apps that make them look the best. Apparently, they haven't shown all the WebXPRT results and I find the Tomb Raider results quit suspsicious. Independent testers have found the M1 to be 50% faster than the intel Xe at this game, but somehow it's only 10% faster when it's intel who does the testing.
- changing the hardware configurations between tests to mislead customers. Like using an unspecified laptop with their most power-hungry 1185G7 CPU (which almost no laptop vendor uses and which probably consumes 3-4X more power than the M1) for performance results and switching to their lower-power CPU when comparing battery life (and also switching to the MacBook Air, which has a smaller battery).
- resorting to meaningless metrics, like that "Evo platform", which no one except intel really cares about and which relies such weird criteria as "Select picture menu" in powerpoint. Sure, opening the "font" panel on MS Word takes like 5 seconds on macOS (which is silly, BTW), and is instant on bootcamp on the same machine. Does it make my intel CPU 20X slower than my... huh... intel CPU? That's moronic, it just shows that Office does not use the same piece of code under macOS and Windows.
-...
While Apple has been guilty of certain offences on some occasions, I don't remember them doing such a misleading job.
The specific results they're showing on Apple.com compare an M1 Mac to the intel Mac it's replacing. So they tested the same apps on the same OS, and the results they show are quite representative of the performance difference. They were confirmed by independent reviewers, whereas intel appears to be the only one to find intel's CPUs better. How strange...
Last edited:
But that's not reality if the performance disparity gets too great. Really, we went through this in the Mac world once already. I'm not saying it's going to happen again, I'm just pointing out that it's not an impossibility. Regardless of how much you want an Apple computer, if it gets to the point where something else can get your workflow done in half the time, you switch. My guess is that Apple knows what they are doing here and that they are going to be able to stay competitive for the foreseeable future. I expect they have a roadmap and have a very good idea how much performance they can expect out of their designs over the next decade or so.
The issue Intel faces is that of performance per watt. And of course the heat generated that needs to be removed. Anyone can increase performance by upping the wattage, but then they’re stuck with a chip that isn’t battery friendly and requires high capacity fans to dissipate heat, which adds to both noise and power usage. So many of the PC’s I’ve seen reviewed against the M1’s require them to be plugged in, or their performance drops precipitously, and even at the reduced performance level on battery, they only get a couple hours of usage out of them. They’re effectively desktop computers in laptop form.
if I had to guess, I’d say Apple has at least a two year lead with their SoC’s and if their next round actually does address power users the way we are hearing, that lead will only grow.
This is pretty well in line with Intel's new CEO's attitude. I mean, Intel just CANNOT be outperformed by chips from a lifestyle company.
I get what you are saying, but I honestly do not think that's true. Apple today is lightyears ahead of what Apple was when all they were was a computer company. The average Apple consumer is not buying Apple because of what numbers say on a piece of paper. They buy Apple because they want in on the Apple ecosystem. Apple is not going to stop innovating or advancing technology, but they have positioned themselves where they do not need to be the best on a benchmark. Apple is selling you on an experience and they largely deliver on that in spades.
The way Apple is positioned and how they built their company is why no one should be worried about Apple falling back to the PPC vs. Intel days where Apple had to compete on benchmarks.
Last edited:
This was true in the PowerPC days.
Early, the 601 was a starting point, 603e (G2) was low power and heat for laptops and 604 series was good for desktops and processor upgrades. The G3 got faster and faster and hotter. Then they starting having scaling issues and over time they had to liquid cool the G5 towers and couldn’t offer a G5 laptop.
That's because you're not an Intel employee where (employee and apparently chip) performance is measured in how fast you can churn out Powerpoints in PDF format for PR distribution.
Intel didn't mention how big a heatsink the Intel processors needed in order to achieve their questionable results.
Intel is comparing its high-end chips to a chip that's unavailable to Intel customers. This makes no logical sense.
It implies that Intel is having an internal and external crisis of confidence. They didn't do this when a direct competitor came out with faster chips (AMD).
Apparently the M1 really hit them right in the nads. More importantly, it's the third perceived failure of Intel recently: the modem business, the process failure, and now they're getting spanked by a consumer products company.
It implies that Intel is having an internal and external crisis of confidence. They didn't do this when a direct competitor came out with faster chips (AMD).
Apparently the M1 really hit them right in the nads. More importantly, it's the third perceived failure of Intel recently: the modem business, the process failure, and now they're getting spanked by a consumer products company.
It's annoying that Intel is reduced to this. And telling. I don't have a strong preference for MacOS over, say, Windows. I can pin-point several things the Mac does better, but I'm a high-tech person who can make use of any system.
I don't have love for Apple or Microsoft. Or any OEMs, really. I worry most about the strange concentration of power and influence these companies have, frankly. I wish that there was real collaboration in the engineering world on open systems on which we could install the OSes and softwares we preferred. Turns out that is not likely to be x86.
Apple upped the game, pretty decisively, though, with M1. The M1 laptops right now are so much better for low-power computing than the available options by a definitive margin. It's enough to prompt me to stick with Mac for any likely purchases this year. As it is, it's considerably re-assuring that Mac still continues to be a configurable Operating System, unlike iOS, in that Apple at this time respects my right to install alternative tools, both open-source and commercial software, and change their OS defaults to my preferences and needs.They also make good PR noise about respecting customer privacy, which I take as, while still framed in their interest, is not nothing, as they say.
I don't have love for Apple or Microsoft. Or any OEMs, really. I worry most about the strange concentration of power and influence these companies have, frankly. I wish that there was real collaboration in the engineering world on open systems on which we could install the OSes and softwares we preferred. Turns out that is not likely to be x86.
Apple upped the game, pretty decisively, though, with M1. The M1 laptops right now are so much better for low-power computing than the available options by a definitive margin. It's enough to prompt me to stick with Mac for any likely purchases this year. As it is, it's considerably re-assuring that Mac still continues to be a configurable Operating System, unlike iOS, in that Apple at this time respects my right to install alternative tools, both open-source and commercial software, and change their OS defaults to my preferences and needs.They also make good PR noise about respecting customer privacy, which I take as, while still framed in their interest, is not nothing, as they say.
Until Intel faces up to the fact that they need to drop some portion of backward x86 compatibility, they are never going to be able to compete with Apple in terms of power to performance. I don’t see that happening and they obviously are still screwing around from the process standpoint. Intel are fools for engaging in this idiotic PR stunt. This reeks of desperation and the Mac only has 8% marketshare, so who cares if a niche OS platform is faster than them? Someone at Intel should be fired instantly for this fiasco.
This.
“Fonds” are handled quite differently in MacOS and Windows and it’s no surprise Word on Windows is faster loading the fonds than Word on MacOS. Which is your point.
And why compare Chrome and only Chrome? One wonders...
Well both are true! Apple was vague because the M1 is not always faster, so their vague description of the
M1 was truthful. And its true that Intels cherry picking is dodgy.
As a writer I spent years using iBooks on my lap.
Comfy chair with foot rest. Laptop. A drink. Hours of time a day.
maybe, maybe they are working on a server chip? I think server chip would be much more beneficial for them, but I agree that there is competition coming, and I think QCOMs acquisition of Nuvia and stating that they will use them to improve their client CPUs is much more of a "threat" than MSFT ...