Is anyone else dissapointed in single core performance of the M1 Pro/Max?

[SpecMode]

Is this a joke thread? The single core performance rocks. I think Intel has 1 chip (@135w/270w-thats a desktop or a really heavy laptop with low battery) that barely, barely, barely, ever so narrowly edges it out. So hard to argue with success

Also a benchmark is only as good as the software measuring it and the persons knowledge of how to properly measure it

Yeah, I'm not sure why folks are dragging on the lack of single-thread performance improvements on one of the fastest single-thread CPUs to date — M1 Pro and Max are using literally the same Firestorm (performance) and Icestorm (efficiency) cores as the original M1, just in a different overall chip design. Single-thread performance almost certainly won't change until the M2 generation, where new core designs (possibly based on A15) show up.

 

[Love-hate 🍏 relationship]

I have a Pentium 5 Windows machine somewhere in the attic that will take care of 125% of your needs.

There is No need to be patronizing
Besides i didn't say i need it (i don't have it in first place )

 

[JPack]

We knew a long time ago the M1X would be based on M1/A14.

There were no architectural changes, so single core performance wasn't going to improve.

Unless you had unrealistic expectations like M1X being based on A15, it shouldn't be a disappointment.

 

[Love-hate 🍏 relationship]

ill share with you some mbp 14 bench if i can ,but im just surprised at the results this guy got .this is mba m1

 

[yitwail]

We knew a long time ago the M1X would be based on M1/A14.

There were no architectural changes, so single core performance wasn't going to improve.

Unless you had unrealistic expectations like M1X being based on A15, it shouldn't be a disappointment.

Even if it had been based on A15, difference might not have been that large because A15 was focused more on efficiency than performance because of disappointing iPhone 12 battery life.

 

[mi7chy]

GPU is a greater concern considering most heavy workloads are offloaded to it. CPU takes a backseat to GPU for the past decade or so.

 

[huge_apple_fangirl]

Hopefully M2 Pro/Max is based on A16 (and hopefully A16 has big boost in single-threaded performance). Still, it's pretty impressive that Apple's core design from last year is, at the very worst, neck in neck with Intel's as-yet-unreleased designs!

 

[JimmyjamesEU]

Post removed.

 

[JimmyjamesEU]

View attachment 1882566
ill share with you some mbp 14 bench if i can ,but im just surprised at the results this guy got .this is mba m1

Strange results. Doesn’t tally with other tests I’ve seen, or my own experience. I tested the pro at apple today and consistently between 260-270. In any case, this is not a single thread benchmark. The do closely correlate, but you want to use geekbench for that.

 

[ecrispy]

My post was based on the YT review I linked, which is by a developer and seems to be a channel thats a lot better than the typical youtuber who has zero technical knowledge.

It seems his benchmark scores on the MBA are higher than others are reporting, and that there may not be such a gap anymore?

There is no need to compare to Intel or how much perf AS has, we all know that. Why its still based on A14 is probably to do with the fact that scaling M1 is not as easy as people expected, and Apple needed much larger production capacity to meet demand which they could do with the existing fab design.

 

[ogawa]

Unfortunately a lot of CAD software hasn't really evolved to take advantage of multi core, so I was in the camp of those hoping for a single core increase with the M1 pro/max chips. That being said, all M1's have a much faster single core speed than my iMac so I'm still thinking of making the upgrade.

 

[arvinsim]

Which is why if 16GB RAM is enough, then just get the M1 MBA instead of the pros.

 

[8CoreWhore]

When a single core is being tested, does it use the efficiency core, or a performance core? Also, in real world, if an app wanted 1 core but needed lots of performance, does it get the efficiency core or a performance core?

 

[project_2501]

I think the OP has a good point.

I am surprised the "pro" laptops have the same single core performance as last years "air" laptop.

 

[bgillander]

Why are people trying to turn this into Intel vs Apple? I'm simply making the point that for certain use cases the older M1 performs better, and that use case happens to be not some trivial benchmark but web performance and single core.

As for 'pro' a MBP should perform better than a MBA in every single category (except portability, size, power draw) no questions asked, not just multi core/gpu.

Have you never noticed that as you go to higher core counts with multi-core CPUs, the actual (not max boost) GHz drops as the core count goes higher? Go check the GHz on the Mac Pro at each price level and you will see that you pay the most for the slowest core speed, but more cores. The fact Apple seem to have kept single core speeds relatively the same is actually pretty impressive, as there are some rules of physics and heat that don’t like to break.

 

[bgillander]

I think the OP has a good point.

I am surprised the "pro" laptops have the same single core performance as last years "air" laptop.

It does make sense to me because this is still the same M1 generation core, and the fact that they have been able to increase the number of CPU cores and even further increase the number of (heat generating) GPU cores apparently without dropping the CPU core speed is fairly impressive.

 

[Pro Apple Silicon]

It's literally not possible to be disappointed with what ranks among the best single core performance in the world.

 

[bgillander]

It's literally not possible to be disappointed with what ranks among the best single core performance in the world.

I would say it should not be possible, but what is that saying?

Random MacRumors forum member: “Hold my beer…”

 

[huge_apple_fangirl]

Unfortunately a lot of CAD software

Doesn't run on Mac at all

hasn't really evolved to take advantage of multi core, so I was in the camp of those hoping for a single core increase with the M1 pro/max chips. That being said, all M1's have a much faster single core speed than my iMac so I'm still thinking of making the upgrade.

You use a macOS for CAD? That's really interesting. I'm curious what software you use, because it was my understanding that most CAD software either didn't run on Mac or lacked feature parity with Windows. And I imagine that whatever does run isn't Apple silicon native.

 

[thekev]

thanks .funny i dont get such a high score on my mba m1 ,in fact i reach around 230 ,while on the 14 and 16 i was averaging 280 so...

however i must point out that unless im missing something here,chrome and most modern browser do actually rely on multi threading and multi core ,a LOT . ive read a comment in the video you sent and someone points out that for browsing its likely the chip relies on the efficiency cores more than P cores ,which would explain the lower score on the m1pro . after all the m1 has 4E which equals to 1.3P while the m1pro has only half of that (2E=0,7P)

mind you ,i think the scores on both are actually the same ,not lower.idk what the hell is wrong in this video but again
1)i dont get such high scores on my mba
2)i didnt get such low scores in mbp 14 base and 16

still ,in the best case,they both perform the same ....which is weird for a pro chip indeed

Virtually all web browsers use some form of multithreading. Anything that would be a realistic choice for you uses it so that threads can sleep while waiting for non-critical network resources. They don't tend to do a lot of parallel computing in the higher level portions of their respective codebases. Many of them use the same or similar libraries for computationally expensive tasks, and that's where you tend to see a lot of tight parallel regions, where multiple threads are used to increase total throughput rather than just to hide task latency.

 

[GrindedDown]

Correct me if I’m wrong, but don’t lower clock frequencies result in better battery life? If that is the case, then it seems really clear why apple did t change the clock speed. This is the first thread I’ve seen to say outright that single core is disappointing (not saying you’re wrong), but I’ve seen countless comments on the amazing battery life of these machines. Those countless comments about battery life translate into sales, big time.

 

[fakestrawberryflavor]

I mean it’s the same Firestorm core crom M1 so yeah it’s the same, this is the large chip of many cores. X86 CPUs follow this method too, release the small consumer chip, wait a while, release the bigger more cores chip of the same core. More MT performance, similar ST performance. How is this even a thread? M2 or whatever will give you the new cores and the new ST performance.

 

[leman]

I would have preferred to have higher single-core performance, yes. But I suppose this is Apples approach to computing. I do admit that there is something refreshing about not artificially limiting the performance of your products.

 

[smetvid]

This is the nature of the M1 chip and why the Pro and Mac still have M1 in the name. Same core just more of them for each level you go up. Most of us the day after the M1 came out assumed the single core score would stay the same and Apple would just add more cores as needed. Just made sense from an engineering perspective.

The M1 single core score is already very impressive so I'm not sure what the issue is with the Pro and Max having the same speed.

We will have to wait for the next generation of Apple Silicon like the M2 to see each single core have an increase in clock speed or better performance with the same clock speed.

I do not find the single core score of the M1, M1 Pro, or M1 Max to be surprising or a disappointment at all.

 

[Arctic Moose]

What matters then is thermals (sustained performance)

Which I expected the bigger charger and thicker and heavier enclosure housing a new cooling system with big fans to accomplish, thus leading to better (sustained) single-core performance. I would have preferred a MacBook Air (a year ago) but since the SSD size was restricted to 2 TB I waited, and it would have been nice if the delay and added bulk meant better single-core performance. As I said though, I’m only mildly surprised, and only a little annoyed.

I think it's clear that Apple's plan for this chip (and subsequent ones) is not to be judged by clock speed, otherwise we fall into the 1990s Intel bull****. It's clear they've found a better way to design and execute a chip design so expanding upon and exploiting that will be the initial focus of this product.

I don’t think the processors should be judged on clock speed, they should be judged on performance. It seems reasonable to me that the highest-end machine would have better sustained single-core performance than the most basic one. Like it or not, there are lots of common tasks that for whatever reason are not multithreaded, and where faster execution is beneficial.

I am sure I will be perfectly happy with what I have ordered, but the my gut feeling about paying $5000 for it would have been better if it had been able to outperform an iPad, regardless of metric.