Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Just at the comment section here!

And BTW, nice game they used to "test" the GPU's just to give the advantage to Windows machines and make the Mac look bad.

They chose a shoddy Mac port from 2011, that running DirectX emulated through OpenGL, versus updated native DirectX graphics, nice, such a crappy move, you can CLEARLY see in the bottom that they are sponsored by Huawei, so they had to make the Huawei win.

This is why Apple is moving to Metal, and not looking backwards.

And the "it's because APFS" thing? Last time I checked Windows doesn't have APFS, or something similar, it's not Apple's fault that Microsoft chose to save money and don't develop a modern file system and it's still using NTFS from the 90's. In fact, Microsoft was developing ReFS, and they stopped because... their costumers were not complaining about NTFS.

And not, it's not an APFS clone. Because if it were, it would be even faster, in fact, instant.

Apple's SSD's are that fast, and other guys SSDs are that slow, it's no surprise, the older generation already had a crazy fast SSD:

ssd.png

So, those benchmarks make sense, and it's "a win is a win".

Thank you. It’s nice to see somone actually post proof and stand up for their resources and ability to outpace the rest. I think we forget that Apple has been the innovator or at least out front the majority of time. I was pissed at Apple blacklisting my ole’ 9.1 iMac when it could still run Sierra, until I the requirement for metal. Any new technology takes time to get to full speed, takes a bit to get up to speed. I’m not how metal actually works, but given metal is used in almost every pert of operations, not anything to do with graphice, I’m going to assume Apple is incorporating the GPU with the CPU for more power and responsiveness and making it easier to achieve even more realistic color. I could be wearing my b*******t on the outside of my boots, but Apple will once again walk away in front.

There’s a reason Apple is #1. They build higher quality products, and no one makes hardware like them. No company is perfect, and Apple has made some seriously bad calls in the past, but ya’ll acting like entitled trust fund brats, complaining you don’t like the maid because she made a mistake and broke her foot is pretty darn hilarious. Do you complain this way about your wife’s cooking, or do you put up with it because she has "other assets," you love, if you will, lol? Point made.
 
  • Like
Reactions: TimmeyCook
Yeah my thought exactly. It could be some caching or buffering mechanism that allows a file to appear to be moved or copied but it could be just accelerated by the cache. The test didn't specify the volume of the SSDs, which could affect the results as well. It would be more accurate to run tests with different file sizes and conditions, like what crystal disk mark does.

Come on people, lol, This talk is like back in the 1700/1800’s when a new medical advancement became available. People called it the work of the devil and yada yada. Yes the world is flat or any of the other stark, fundamentalist thinking and scare tactics used by the ignorant. You have the right to free speech, and free will, so don’t talk, cause talk is cheap. Go do the damn test yourself and prove your points. Armchair qb’s:). I’m thrilled Apple is moving forward. That’s good for my quality of life. These advancements reap manny more benefits than any mistake Apple has ever made, wouldn’t you say?
 
I now own a new 2018 15” MacBook Pro which I am quite happy with. This week I am in Shenzhen and stopped by a Huawei shop to see the new MacBook Pro competitor from Huawei and I must say I was amazed. Played around with the laptop for around 15 minutes or so and it’s very nice. If I had to pick another laptop aside from a MacBook Pro this would be it. Having an iPhone though I like how everything works so smoothly between the iPhone,iPad Pro and the MacBook Pro. The keyboard on the Huawei though I must say is much better compared to my new MacBook Pro although the Keyboard on the X1 Carbon Thinkpad beats both of them hands down. Hard to have it all I suppose.

The SSD numbers on the Huawei from what I read were not all that great but while testing it here it was very very quick. I suppose if one could swap the one it comes with to a faster model would be cool.
 
Snore.

For most people, the storage speed upgrades are meaningless. Improved graphics performance, on the other hand, would benefit most people.
 
Snore.

For most people, the storage speed upgrades are meaningless. Improved graphics performance, on the other hand, would benefit most people.

I, for one, hardly benefit from further graphics performance at all. More RAM and a faster SSD make a huge impact.

(The added cores don't hurt either.)

I don't see the basis for the assumption that "most" need graphics improvements. The MacBook Pro is never going to be a gaming beast (aside from eGPU), so that leaves us with… maybe audio/video and ML stuff?
 
  • Like
Reactions: G5isAlive
I, for one, hardly benefit from further graphics performance at all. More RAM and a faster SSD make a huge impact.

(The added cores don't hurt either.)

I don't see the basis for the assumption that "most" need graphics improvements. The MacBook Pro is never going to be a gaming beast (aside from eGPU), so that leaves us with… maybe audio/video and ML stuff?

Wonderful. You're one of the .04% for whom faster storage matters. For the people doing facebook, Word, and the occasional game it is utterly meaningless.
 
Wonderful. You're one of the .04% for whom faster storage matters. For the people doing facebook, Word, and the occasional game it is utterly meaningless.

How many "doing facebook, Word, and the occasional game" spend $3k on a laptop?
 
  • Like
Reactions: G5isAlive
Anyone got a new 2018 MBpro handy that they can run CystalDiskMark or similar benchmark?

I'm curious on the test on OP and how they came to those results. I'd like to see actual numbers other than "We copied a file!" since pure raw throughput often doesn't accurately reflect real world usage which has a lot smaller transactions

For example: the NVME in my laptop
 

Attachments

  • image.png
    image.png
    196.4 KB · Views: 131
  • Like
Reactions: sman789
The only other Mac that matches the speed reported is the iMacPro. The iMacPro uses RAW SSD chips in parallel with no onboard controller to achieve these crazy speeds. The custom T-2 acts as the controller instead of the CPU and the T-2 chip handles all of the encryption on the fly, again instead of the CPU.

Thats not really a new thing, SSDs have been using parallel NAND-channels pretty much since the beginning and many mainstream SSDs support encryption without relying on the CPU (see "self encrypting drive"). Apple has basically ripped appart a "Standard SSD" and put the controller part into the T-2.

PS: remember SandForce SSD-Controllers? They also always encrypted (and compressed) the data stored on the NAND-cips making data-recovery pretty much impossible if the controller failed.
 
With other laptops, should the device die for some reason, ultimately, the data can be recovered very easily by removing the drive and swapping it to a good unit, or for technicians, plugging them into external devices to read the data. if you solder down the storage, there's no way to easily retrieve it should there be a problem. In the MacBook Pro's case with soldered storage, if your MBPro becomes unusable for some reason, you cannot access your own data without going to Apple, who must take the laptop apart to get to a proprietary socket they've added just for this purpose, but it means potential of long term (as Apple repair can take several days to weeks) before you get access to your data back.

Didn't Apple remove this port? If so, even this option is unavailable.
 
I think the tester need to look into this issue.

If it was testing APFS's COW (copy on write), the MacBook Pro just wrote the metadata (maybe a few kb) to disk instead of 4.9GB actually data.
This isn't directly related to CoW. This is more like file-level filesystem snapshots. Would be a lot harder to do without CoW though, admittedly.
 
This isn't directly related to CoW. This is more like file-level filesystem snapshots. Would be a lot harder to do without CoW though, admittedly.

Neither. It’s a fast controller. That’s all. No fs magic in this case.
 
As others speculate these results are not really indicative of anything, due to how APFS influences these results.

When using APFS and duplicating a file, the file isn't actually stored twice on disk but only once, and both locations store a reference to the actual file on disk. Hence the high speed.

Changing even 1 byte in the duplicated file is what actually triggers a full copy of the file on disk. This is what they should've measured. Presenting it in the current way is deeply flawed.

Changing 1 byte does not trigger a full copy of the file on disk.
https://developer.apple.com/documentation/foundation/file_system/about_apple_file_system The section about how clones reduce the cost of copying goes into detail about how the deltas between the different copies are stored.
 
  • Like
Reactions: CarlJ
I think the tester need to look into this issue.

If it was testing APFS's COW (copy on write), the MacBook Pro just wrote the metadata (maybe a few kb) to disk instead of 4.9GB actually data.

Guys, the MacBook Pro runs 2 SSD chips on separate PCI lanes similar to RAID0. The MacBook uses an equivalent of NVME. The max speed for a PCI express NVME is 1400MBps. Running RAID0 gets you close to double that so 2600 is correct. 2800MBps is the max and 2600MBps allowing for processing overhead is accurate. Many other laptops are fitted with SATA SSDs which max out at about 550MBps. Copying 4.9 GB in 2 seconds is correct for a MacBook Pro.
 
  • Like
Reactions: waigx
Guys, the MacBook Pro runs 2 SSD chips on separate PCI lanes similar to RAID0. The MacBook uses an equivalent of NVME. The max speed for a PCI express NVME is 1400MBps. Running RAID0 gets you close to double that so 2600 is correct. 2800MBps is the max and 2600MBps allowing for processing overhead is accurate. Many other laptops are fitted with SATA SSDs which max out at about 550MBps. Copying 4.9 GB in 2 seconds is correct for a MacBook Pro.

Just want to update some corrections on numbers:

PCI-E v3.0 single lane is a thoeretical max of 985MB/s. Most NVME operates with an x4 lane giving a max thoeretical for a single NVME drive 3,940MB/s (31520mbps). Max speeds of NVME based drives currently are capable of 3000+ seq reads, 2000+ seq writes. which is far higher than the 1400MBps you've claimed

this is about on par with Apple's offerings in most of their lineup. in fact, it's basically the same tech Apple is using.

but yes, there are still lots of low end manufacturers who are putting SATA based drives instead of NVME, and that is limited to the SATA III bus's limit of 6Gbps (600MB/s)


if you scroll up above to my earlier post on this page, you will see a benchmark I have done directly on the NVME SSD that came in my laptop.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.