Performance Preview: 13-inch MacBook Pro Benchmarks Are Here

[AndrewFreedman]

Laptop Mag released its initial benchmarks for the 13-inch MacBook Pro, and its SSD numbers are particularly fast.

Performance Preview: 13-inch MacBook Pro Benchmarks Are Here : Read more

 

[jimmysmitty]

That review has a few flaws. The test seem suspicious and also not like for like.

If they want like for like here is a similarly cost MSI laptop:

https://www.newegg.com/Product/Product.aspx?Item=N82E16834154891

8th gen CPU, just like this one, PCIe SSD, like this one, 2x the RAM AND a dedicated Quadro GPU.

I guess they wanted to compare just 13 inch to 13 inch and not actual specs.

 

[velocityg4]

That review has a few flaws. The test seem suspicious and also not like for like.

If they want like for like here is a similarly cost MSI laptop:

https://www.newegg.com/Product/Product.aspx?Item=N82E16834154891

8th gen CPU, just like this one, PCIe SSD, like this one, 2x the RAM AND a dedicated Quadro GPU.

I guess they wanted to compare just 13 inch to 13 inch and not actual specs.

That's their standard way of doing comparison tests. People in the market for an ultraportable of a certain size aren't going to care much how it compares to a much larger and heavier computer. There are other sites for such comparisons.

 

[krelat]

That's their standard way of doing comparison tests. People in the market for an ultraportable of a certain size aren't going to care much how it compares to a much larger and heavier computer. There are other sites for such comparisons.

Can you link to some? I did a search for benchmarks on these machines and the sites that this one links to seem to be the only ones bothering at the moment, but I'd love more detailed opinions.

 

[heffeque]

Notebookcheck has several lists: postimg dot cc slash image slash qnzd5i9en

Surely one or two fit your demands.

 

[ssdpro]

The picture indicates it can successfully run Fortnite. How productive.

 

[therealduckofdeath]

That review has a few flaws. The test seem suspicious and also not like for like.

If they want like for like here is a similarly cost MSI laptop:

It's unfortunately always been like that when blogs "benchmark" Apple products. Apple deliberately put their components in between commonly used hardware speeds and then reviewers always compares with (much) cheaper, never more expensive.

 

[stdragon]

You're buying into the Apple ecosystem; that's the "value" they pitch. In terms of dollar/performance ratio, Apple will always lose to the PC market.

Apple used to mean "state of the art" in terms of a blended form and function. But that died with Steve Jobs. Now, it's just form. Apple just sell designer computers nowadays.

 

[nobspls]

I though Tom's had higher standards then just relaying Crapple propaganda from crap sites like laptop mag. If Tom's isn't doing their own benches there is no point in producing click baits to other sites. Keep doing this and Tom's will lose all of their credibility and destroy Tom's hard earn respect and reputation.

 

[razor512]

The areas where it fell short during the CPU benchmarks seems to be a thermal issue. The system did very well with short multi-threaded benchmarks, but for longer productivity ones, it did worse than older generation hardware. This means that apple released yet another generation of macbook where the cooling is inadequate, thus it thermal throttles after a little while. Such a thermal design also means that the system gets slower over time as any dust buildup no matter how small, will negatively impact overall performance.

 

[loveday.ben]

It was such an amateur review IMO. The benchmarks were hugely skewed in Apples favour and weren't comparing 'apples with apples' - no pun intended. Their SSD benchmark was like me saying 'wow, my all flash SAN can do a volume clone in 2 seconds!' because it never actually copied any data...

 

[Solandri]

Can you link to some? I did a search for benchmarks on these machines and the sites that this one links to seem to be the only ones bothering at the moment, but I'd love more detailed opinions.

From what I've been able to find, their benchmark results are not due to the SSD. They're due to Apple's new APFS which seems to create hard links (like a shortcut in Windows) when you copy files, rather than actually copy them.

https://www.pcmag.com/article/345519/what-macos-sierras-new-apfs-file-system-means-to-you

So basically, their benchmark is them copying a bunch of files from one folder on the SSD to another folder. The Windows machines dutifully copy the files bit-for-bit. The Mac says "you only want a copy? Well then I'll just create a second filename and make it point to the original copy of the file." The copy time then gets shifted over to when (if) you modify the copied file. Because only then is a new copy truly created and written to disk.

 

[stdragon]

The new MBP have an SSD read spead of 3.2GB/s and write 2.2GB/s. I'm not aware of the IOPS rating, but APFS was built for flash storage and other next generation feature. Encryption, shadow copies, time stamp granularity, and of course, de-duplication to better utilize space more efficiently.

Speaking of de-duping data, while it's heresy to call any file copy on a single storage device a "backup", the concept of de-duping and cross-referencing a single chain of incrimentals from a SINGLE source makes me uneasy. A lot of things can go wrong with that source. Corrupted data from a bit-flip (non-ECC) to a dirty shutdown to where not even the file system journaling can recover properly (but it does save the integrity of the APFS as designed). Yes yes yes, have backups and all that, I know. But a copy is supposed to be a copy for a reason. But whatever.

 

[photoew]

Wow, the cognitive dissonance here could be cut with a knife.

Here's my benchmark for my 15" MacBook Pro with i9 processor, 32 gigs of RAM and 1TB NVMe. (Apple usually uses the best ssds available, currently the Samsung 970 Pro.)

Write: 2580.7 MB/s
Read: 2682 MB/s.

External drive with four 256 gig NVMe WD Black drives striped RAID 0:

Write:846.6 MB/s
Read: 2303.1 MB/s.

I'm sure bigger drives would have yielded faster write speeds.

That's all the testing I've done so far. But personal impression is this thing screams.

 

[Non-Euclidean]

Did you disable the instant cloning so you can get an accurate measurement on file write speeds?

 

[TJ Hooker]

Did you disable the instant cloning so you can get an accurate measurement on file write speeds?

Doesn't sound like it. From the Laptop Mag review:

"I had to do a double take when I saw how quickly the new 13-inch MacBook Pro duplicated 4.9GB worth of data. It took 2 seconds, which comes out to a rate of 2,519 megabytes per second. That’s insane.
[...]
To be fair, Apple’s relatively new APFS file system is designed to speed up file file copies using a technology Apple calls Instant Cloning. But a win is a win."

 

[Solandri]

Wow, the cognitive dissonance here could be cut with a knife.

Here's my benchmark for my 15" MacBook Pro with i9 processor, 32 gigs of RAM and 1TB NVMe. (Apple usually uses the best ssds available, currently the Samsung 970 Pro.)

Apple has been trying to get away from Samsung SSDs due to bad blood over their court cases against each other. Initially they went with (slower) Toshiba. Lately they've been using (slower) Sandisk SSDs. There was a small "scandal" a couple years back where they sent MBPs with Samsung NAND to reviewers, but most of the production models used (slower) Toshiba NAND. People were opening the box, benchmarking the laptop, returning it to the Apple Store under warranty for an exchange, repeat until they got one with Samsung NAND.

https://www.ifixit.com/Teardown/MacBook+Pro+13-Inch+Touch+Bar+2017+Teardown/92171

(iFixit's teardown of the 15" model did find Samsung NAND. However, the laptop in the review was the 13" model.)

Edit: Also worth pointing out that Toshiba sold their memory division to Bain - a group of companies which includes Apple..

https://www.reuters.com/article/us-toshiba-accounting-idUSKCN1BA13K

Write: 2580.7 MB/s
Read: 2682 MB/s.

Here's an 850 EVO with Rapid mode enabled:

That's about as relevant a benchmark if you've got Apples Instant Cloning enabled. You're not measuring the read/write speed of the SSD. It's not making bit-for-bit copies of the files, it's making hard links (a new filename which points back to the original file). Hard links are only a few bytes of data that need to be written to the drive per file.

In both cases, you're not benchmarking the SSD. You're breaking the benchmark program by doing something it isn't expecting.

External drive with four 256 gig NVMe WD Black drives striped RAID 0:

Write:846.6 MB/s
Read: 2303.1 MB/s.

This is good! You're in a unique position here. Instead of running benchmarks, take a folder full of a bunch of different files, and copy it from one folder to another on your internal drive. Time how long it takes. It should be super-fast since it's just making hard links.

Now try copying the same folder from the internal drive to the external drive. See how long that takes. Instant cloning shouldn't work across different drives (unless Apple wants the file to become unavailable if you plug the external drive into another computer - if it does copy as quickly, see how long it takes to eject the external drive). So that should give you a real measure of the speed of your SSDs.

 

[stdragon]

That's the only real way to test it - copy from source to destination on two different physical drives. With the advent of software and hardware caching based schemes, it will throw off any benchmarking app that's getting an acknowledgement that data was immediately committed to storage when in fact it wasn't, or still pending a write-back in queue.