13-Inch MacBook Pro With M2 Chip Outperforms Base Model Mac Pro Despite Costing Nearly $5,000 Less

[PinkyMacGodess]

Not sure what your point is. My first Graphics computer was a Qauantel Paintbox and cost ( $250K - the 2022 equivalent £650,000). You can spec a single box Dell ever $250K now.

My Mac Pro cost me $16K and took less than a month to pay for itself.

I know that there were computers that sold for a lot more than what I was buying that house for then. It's just amazing how ubiquitous things have gotten, and how things were supposed to be lower priced, and yet how expensive some things still are. I prived a 'loaded' new new new mac pro and almost fell off my chair. But it's like some of the tools that I have. You get (usually) great tools for higher prices. *usually* I've morked my old old old Mac Pro to death. It's served me well. No issues with it either. Good stuff, but my butt still puckers thinking about it...

 

[jav6454]

Sorry… these are really stupid comparisons (thermals? Storage? Max RAM?). Processors aside, these are two completely different computers for different needs. Next thing they’ll be saying is the m2 MacBook Pro is far more portable than the Mac Pro, despite the Mac Pro having wheels.

Yes, they are computers for different needs. One has a server class processor and the other is a consumer processor. Geekbench is one way of viewing computational power, the correct way is to analyze actual program tasks.

 

[cocoua]

this only brings over the table the shameless situation Apple has maintain with it's real "Pro" lineup. Outdated chips.

So, Intel hasn't released most modern Xenon in this years? or even more competitive i9 desktop processors?

Mac Pro have been doomed since the G4 1,42Ghz Dual (which I had) Only worth the price the firsts months.

OK, MacPro arent just the processor, but has a ton of upgradable options for scientific tasks...

 

[jav6454]

There are 12 years between those 2, that's not "expected" it's long term churn and completely different chip architectures and software...you can't even run 64bit bios on that OG pro.

12 years of enhancements and node reductions.

 

[pasamio]

Well the key difference being that the other was released only 3 years ago. We are at the forefront of a significant change. I don't remember the previous mac pros circa 2008-2012 being beaten that fast. Definitely not 3 years.

Amazing what moving off Intel's chips over to Apple Silicon has done. A good chunk of the reason of the lack of performance improvement has been Intel's relative stagnancy over the last decade as they struggled with their manufacturing processes. It's at the point now that Intel, which has their own fabs, is now paying TSMC to make some of their chips to remain competitive. This is what has enabled AMD to catch up with Intel after Intel had been ahead for so many years and if you look at the PC media you have channels like Linus Tech Tips pointing out that a three or four year old CPU can mostly keep up with games still because desktop CPU performance hasn't grown that much.

One thing that will be interesting is when Apple decides to drop Rosetta support from their chips. I've wondered for a while how much space it takes up to support hardware accelerating those x86 instructions and if it is a meaningful gain to pull it back out again and allocate that space to something else.

 

[Sir_Macs_A_Lot]

"Value is in the eye of the beholder" (or something like that). You might not be able to justify shelling out that much coin for the Mac Pro, but as long as enough people do...Apple will continue to make them and keep them at that price-point.

 

[Hessian]

The new 13-inch MacBook Pro with the M2 chip appears to be faster than a base model Mac Pro in benchmarks, despite costing nearly $5,000 less.

In an apparent Geekbench 5 result that surfaced on Wednesday, the new 13-inch MacBook Pro achieved a multi-core score of 8,928, while the standard Mac Pro configuration with an 8‑core Intel Xeon W processor has an average multi-core score of 8,027 on Geekbench 5. These scores suggest the new 13-inch MacBook Pro, which starts at $1,299, has up to 11% faster multi-core performance than the base model Mac Pro for $5,999.

Higher-end Mac Pro configurations are still able to outperform the M2 chip, such as the 12-core model, but at the cost of $6,999 and up.

Given the Mac Pro has other benefits like expandability, configurable GPU options, larger built-in SSD storage capacity options, and much larger RAM options, this certainly isn't an apples-to-apples comparison, but the benchmarks are nevertheless a testament to the impressive performance of Apple silicon chips in more affordable Macs.

A sample of average Geekbench 5 multi-core scores for various Macs:

• Mac Studio with M1 Ultra: 23,366
• Mac Pro with 28-core Intel Xeon W: 20,029
• 14-Inch and 16-Inch MacBook Pro with M1 Max: 12,162 to 12,219
• Mac Pro with 12-core Intel Xeon W: 11,919
• 13-Inch MacBook Pro with M2: 8,928 (based on a single result)
• Mac Pro with 8-core Intel Xeon W: 8,027
• 13-Inch MacBook Pro and MacBook Air with M1: 7,395 to 7,420

The Mac Pro and the high-end Mac mini are the only Intel-based Macs remaining in Apple's lineup. During its March event, Apple teased that a new Mac Pro powered by Apple silicon is coming, with an announcement widely expected by the end of this year.

The new 13-inch MacBook Pro will be available to order worldwide starting this Friday, with deliveries to customers and in-store availability beginning June 24. Apple is also releasing a redesigned MacBook Air with the M2 chip in July that should likewise outperform the base model Mac Pro for an even lower starting price of $1,199.

Article Link: 13-Inch MacBook Pro With M2 Chip Outperforms Base Model Mac Pro Despite Costing Nearly $5,000 Less

What about adding an external graphics card? Is that an option for the MacBook Pro?

 

[Cognizant.]

What about adding an external graphics card? Is that an option for the MacBook Pro?

You can no longer use external GPUs on Apple Silicon Macs. The ended with Intel.

 

[Onelifenofear]

You can no longer use external GPUs on Apple Silicon Macs. The ended with Intel.

I expect that may change when they realise a mac pro With Apple Silicon. However I am actually thinking they have something up there sleeve. Either Apple Silcon that plays well with an external GPU - no reason they can’t - AMD supports Metal or possibly a secondary GPU Chip.

 

[Cognizant.]

I expect that may change when they realise a mac pro With Apple Silicon. However I am actually thinking they have something up there sleeve. Either Apple Silcon that plays well with an external GPU - no reason they can’t - AMD supports Metal or possibly a secondary GPU Chip.

I don't think so. I think we're just going to see an insanely beefy way-too-many-Apple-silicon-cores-to-count GPU. Apple is done relying on other manufactures for these kinds of things.

 

[pasamio]

I expect that may change when they realise a mac pro With Apple Silicon. However I am actually thinking they have something up there sleeve. Either Apple Silcon that plays well with an external GPU - no reason they can’t - AMD supports Metal or possibly a secondary GPU Chip.

Apple writes the display drivers for it's various supported graphics cards, it's why there are specific hardware requirements in the Hackintosh community to ensure that you have a device that the software supports. AMD doesn't support Metal, Apple supports Metal and implements it on top of the AMD and Intel graphics chipsets. GPU drivers on Windows, arguably the best support platform, have noted stability issues so I wouldn't expect quality macOS drivers from those vendors in the long run for an even smaller ecosystem.

I don't expect Apple to offer support for GPUs from third party manufacturers after the transition, it just isn't worth the cost to Apple. They've invested in their own display technologies to power their own mobile devices and now on the desktop. What I think could be interesting for Apple is their own compute card similar to Afterburner, it wouldn't take much for them to continue to repackage the existing chips onto upgradeable compute platforms with their own on board high performance memory that enabled offloading large rendering tasks. I'll be the first to admit that seems fanciful but a nerd gotta dream.

 

[HaddockW]

If we look at how Nvidia is approaching data-center ARM processors. Much of it mimicking M1 Ulta dual SoC Interposers and the fact Nvidia signed a 20 year architecture license. https://www.tomshardware.com/news/nvidia-big-cpu-plans-with-20-year-arm-license


Apple needs to be in the same lane. They need to support drivers and libraries that leverages Apple Silicone. For example, GPU computer like Machine Learning/TensorFlow needs investment by Apple. Without this,a Mac Pro Silicone is doomed. I am doing a lot of Machine Learning work and looking at the A1000 Tesla GPUs for these workloads. Apple is missing out.

 

[Tagbert]

If we look at how Nvidia is approaching data-center ARM processors. Much of it mimicking M1 Ulta dual SoC Interposers and the fact Nvidia signed a 20 year architecture license. https://www.tomshardware.com/news/nvidia-big-cpu-plans-with-20-year-arm-license


Apple needs to be in the same lane. They need to support drivers and libraries that leverages Apple Silicone. For example, GPU computer like Machine Learning/TensorFlow needs investment by Apple. Without this,a Mac Pro Silicone is doomed. I am doing a lot of Machine Learning work and looking at the A1000 Tesla GPUs for these workloads. Apple is missing out.

 

[binarysmurf]

What I think could be interesting for Apple is their own compute card similar to Afterburner, it wouldn't take much for them to continue to repackage the existing chips onto upgradeable compute platforms with their own on board high performance memory that enabled offloading large rendering tasks. I'll be the first to admit that seems fanciful but a nerd gotta dream.

THIS is a rather interesting idea. I wonder if that's the way Apple will go? Apple have to do something to satisfy the GFX needs of its AS Mac Pro users, and the current peak of 64 GPU cores isn't it.

 

[Onelifenofear]

Apple writes the display drivers for it's various supported graphics cards, it's why there are specific hardware requirements in the Hackintosh community to ensure that you have a device that the software supports. AMD doesn't support Metal, Apple supports Metal and implements it on top of the AMD and Intel graphics chipsets. GPU drivers on Windows, arguably the best support platform, have noted stability issues so I wouldn't expect quality macOS drivers from those vendors in the long run for an even smaller ecosystem.

I don't expect Apple to offer support for GPUs from third party manufacturers after the transition, it just isn't worth the cost to Apple. They've invested in their own display technologies to power their own mobile devices and now on the desktop. What I think could be interesting for Apple is their own compute card similar to Afterburner, it wouldn't take much for them to continue to repackage the existing chips onto upgradeable compute platforms with their own on board high performance memory that enabled offloading large rendering tasks. I'll be the first to admit that seems fanciful but a nerd gotta dream.

Sorry, I should have said "AMD chips support metal" rather than the company. I know Apple Writes the drivers - but AMD worked with Apple to optimise the code on the Mac Pro Cards.

Don't get me wrong I completely agree with you- They may make a GPU Compute card. As you say they are fully capable. They've proved they can make fast silicon...

AMD 6800x duo is 30.2 TFlops

Nvidia 3090 is 35 Tflops

M2 = 3.6 TFlops - half a 3050... and less than the RX6500XT, but great for an integrated card.

However, the M1 Ulta is 21 TFLOPS! So I suspect they are looking to blow Nvidia out of the water with the Pro GPU speeds.

What may be interesting is if they make their own Apple GPU card to go into the 2019 Mac Pro!

 

[binarysmurf]

What may be interesting is if they make their own Apple GPU card to go into the 2019 Mac Pro!

This would be an excellent development. Perhaps an evolution of the Afterburner card? We shall see.

 

[M3gatron]

Apple was one of the founders of ARM and has remained a close contributor. Apple has an architectural license from ARM that allows them to use the ARM ISA but do their own designs to implement it. For the first few generations of iPhones, Apple used ARM designs but soon started doing their own chip design and evolved that to our current Apple Silicon A and M-series chips.

Trying to say that Intel copied either ARM or Apple gets complicated in this case. Either answer would be correct.

ARM's big.Little architecture has nothing to do with Apple in the first place. Also the company was founded in November 1990 as Advanced RISC Machines Ltd and structured as a joint venture between Acorn Computers, Apple, and VLSI Technology. Not to mention ARM is a British company.
So again if Intel copied somebody it was ARM, Apple did the same, they also copied ARM's big.Little architecture. Apple has no technical contribution to ARM and can't get any praise for any of ARM's designs in the last +20 years as they have nothing to do with Apple so there's no complication whatsoever.

 

[M3gatron]

An Apple employee was the first CEO when the ARM company was founded 32 years ago and Apple also provided almost all of ARM's initial funding ($3m on day one - and more funding over the following decades).

ARM was supposed to design the CPU for the Apple Newton, which was one of many early failed attempts at creating a smartphone.

Apple didn't "copy" ARM. They paid ARM to spend decades designing a better CPU, and then Apple used that CPU design.

So what? ARM's big.Little architecture has nothing to do with Apple.
And Apple did copy ARM's big.Little style architecture, they didn't come up with the idea themselves. By the time Apple had a working implementation it was already a proven design.

 

[M3gatron]

ARM? What consumer processor had this before M1? I don't mean in a lab but for sale to the general public. To my knowledge Apple did this first.

LoL, what question is that?
Anyway, for your information: big.LITTLE was first introduced in October of 2011, and it was introduced together with two new core designs, the Cortex-A7 and the Cortex-A15. One of the first CPUs to launch with this core design was Samsung's Exynos 5 Octa 5410, which powered the Samsung Galaxy S4 in 2013. Apple introduced the first big.LITTLE SOC in last part of 2016, so 3.5 years later.

 

[M3gatron]

It's not copied it is ARM - M1 is an ARM Architecture as are all the iPhone / iPad chips... they have a perpetual Licence to use the general architecture - all the apple chips have used them. The key is how it all works together and apple are now years ahead of the competition now.

I was talking about ARM Holdings, you know the company and their chip designs, not about ARM instructions sets.

 

[M3gatron]

But Apple was not first.


Also ARM is not owned by apple. apple pays a license to use the ARM architecture and design and apple builds upon it.

Yeah up to the point when SoftBank bought ARM, Apple was holding less than 2% of ARM's shares, quite insignificant. Even from the beginning Apple owned like 20% of ARM's shares so it was never the biggest contributor or the actual owner of ARM.

 

[M3gatron]

I didn't mean just x86 because Apple Silicon isn't x86. I'd say the Surface Pro X falls under the tablet category. I'm not saying Apple invented ARM or the big little concept but Apple was the first to make them work on mainstream devices. Before Apple Silicon, ARM was generally in devices that required low power like phones or tablets or cheap devices like low end Chromebooks.

Before everyone was happy with the status quo like if it works why change it.

That is incorrect, again. ARM was making it's way into servers way before Apple launched a desktop ARM chip.

Amazon Launches ARM-Based Custom Graviton CPU - ExtremeTech

Amazon unveiled its first custom-built silicon today, and a new type of cloud instance designed around it. Meet Graviton, the company's first CPU.

www.extremetech.com

 

[russell_314]

That is incorrect, again. ARM was making it's way into servers way before Apple launched a desktop ARM chip.

Amazon Launches ARM-Based Custom Graviton CPU - ExtremeTech

Amazon unveiled its first custom-built silicon today, and a new type of cloud instance designed around it. Meet Graviton, the company's first CPU.

www.extremetech.com

Before Apple Silicon, ARM was generally in devices that required low power like phones or tablets or cheap devices like low end Chromebooks.

Yes, I know it was in servers. Now re read what I said to include the adverb "generally". I put the Webster definition below.


gen·er·al·ly | \ ˈjen-rə-lē
, ˈje-nə-, ˈje-nər-lē \

Definition of generally
: in a general manner: such as
a : in disregard of specific instances and with regard to an overall picture generally speaking

b : as a rule : usually

Source: https://www.merriam-webster.com/dictionary/generally

 

[M3gatron]

Yes, I know it was in servers. Now re read what I said to include the adverb "generally". I put the Webster definition below.

That quite a lot of effort to cover a flawed statement: Before everyone was happy with the status quo like if it works why change it.
ARM in servers wasn't an exception or something unique, it was a clear direction that has been underway long before Apple launched a desktop CPU with ARM cores, actually its simply absurd to suggest that because of Apple we now have high performance/power ARM CPUs, as we would no doubt have had them without Apple.
The Neoverse N1 ARM Core specifically designed for servers, was disclosed in 2019 so the design started a few years before that when nobody even knew for certain that Apple is going to go ARM in computers. The same way its absurd to suggest that intel copied big.Little from Apple.

 

[pasamio]

However, the M1 Ulta is 21 TFLOPS! So I suspect they are looking to blow Nvidia out of the water with the Pro GPU speeds.

What may be interesting is if they make their own Apple GPU card to go into the 2019 Mac Pro!

They're also doing this at a much lower power draw which is the other thing I'm interested to see is what happens when they put that into the Mac Pro case with the cooling it provides. Then if they did a GPU card that could potentially pull a few hundred watts on their own, you could end up with a beast of a machine. Crazy street cred if they did it but I do wonder how much of a market such a machine would have. One of the more powerful features of Metal is the tile based renderer design which allows them to easily scale all of this out, they're already working on various streaming technologies so their own GPU, with a heft memory package, would be readily placed to stream in data over a lower speed bus but then crank the visuals out once it's all there.