Samsung X5 1TB - 1560 write 2490 read - is it OK?

[kubaw]

Did some performance test with Blackmagic software on a freshly purchased Samsung X5 1TB and got results as in the screenshot (run on iMac 2019 i9). Numbers are similar whether formatted with APFS or MacOS Extended (checked just to be sure). This is not definitely bad, but on the other hand: far worse than promised and far worse then some benchmarks I see here and there on the Internet. What results do you get with your drives? Is my item faulty? Should I try different TB3 cable perhaps?

Edit: real life tests seems even a little worse, copying of 111GB file took 118 seconds so about 1400 MB/sec.

Thanks in advance

 

[Fishrrman]

Your numbers sound about like what one would expect from these drives "in real life".
How warm does the drive get under heavy write loads?

 

[kubaw]

I guess I was fooled by some reviews, which were getting better numbers (such as this: https://www.macworld.com/article/32...edly-fast-portable-thunderbolt-3-storage.html), but note these numbers are also reached by a few reviewers on Amazon.

I've also tried on other iMac (2017 i5) and to my surprise got better write results (1810 mb/s) which is closer to what most reviewers are getting.

As of getting warm - hard to tell yet since I've only did a few short benchmark. It seems I need to play a little more with this device to learn more.

 

[petsk]

Seems a bit low for sure. Do you have anyting else connected to the second thunderbolt port? What are you going to store on the X5?

Try some other benchmarking tools like AmorphousDiskMark from the app store. If you’re going to use the X5 as bootdrive, random 4k read and write speeds are more important. Blackmagic is only testing sequential read and write speeds.

I’m waiting for an 500GB X5 to be delivered, going to use it for booting my 2019 iMac.

 

[Juicy Box]

I have been using the X5 for a while now, but mostly on a Late 2012 iMac under TB1.

On TB3, my X5 gets slightly better read speeds than you, @ almost 2700MBps, but my write speeds are much slower than yours @ 1100MBps.

Just a note, I replaced the 500GB NVMe that came with the X5 with a 1TB Samsung 970 Pro Plus, and after doing this the write speeds dropped by about 15% on TB1. The read speeds were unaffected.

I suspect that using a different NVMe that came with the X5 is affecting my write speeds.

Read speeds are great, and similar to what I get on an Apple 1TB SSD on the M1 Macs.

How warm does the drive get under heavy write loads?

As of getting warm - hard to tell yet since I've only did a few short benchmark.

A while back, there was some claims that the Samsung X5 overheats, but this isn't true. The case gets pretty warm as it is passively cooled, but it doesn't overheat.

I did some tests, and at idle, the X5 is hotter than Apple's blade SSD in the iMac, but under load, it actually stays cooler than Apple's internal blade SSD. The internal temps of the X5 also is cooler than what others have been reporting on other TB3 NVMe drives.

But....

The X5 drives with older firmware had a big issue where the temperature threshold at which the drive starts to throttle was set too low leading to throttling to USB2 speeds under sustain writes of 100+ GB writes.

This would be really noticeable in situations like cloning a boot drive, or doing large back ups to the X5.

In situations like using the X5 as a boot drive, this throttling issue would most likely never happen, as it would only happen on long, not-stop sustained writes, and read speeds would never throttle.

It would also not happen on BMDST, even if testing it for hours due to the nature of the back and forth, read and write testing.

The newer firmware was supposed to fix this issue, and I think most X5 units today probably have the newer firmware, but I have read that people with older firmware were having issues updating it.

 

[GuilleA]

Did some performance test with Blackmagic software on a freshly purchased Samsung X5 1TB and got results as in the screenshot (run on iMac 2019 i9). Numbers are similar whether formatted with APFS or MacOS Extended (checked just to be sure). This is not definitely bad, but on the other hand: far worse than promised and far worse then some benchmarks I see here and there on the Internet. What results do you get with your drives? Is my item faulty? Should I try different TB3 cable perhaps?

Edit: real life tests seems even a little worse, copying of 111GB file took 118 seconds so about 1400 MB/sec.

Thanks in advance

Use ATTO Disk Benchmark, it'll give you far more accurate results. Also, a brand new, empty SSD will always be faster in benchmarks because the controller knows that the pages are empty and have never been used, so it will just return a string of zeroes without actually reading the block. Essentially sending you synthetic data and saturating whichever bus you're using (SATA, PCIe, USB).

 

[ght56]

Use ATTO Disk Benchmark, it'll give you far more accurate results. Also, a brand new, empty SSD will always be faster in benchmarks because the controller knows that the pages are empty and have never been used, so it will just return a string of zeroes without actually reading the block. Essentially sending you synthetic data and saturating whichever bus you're using (SATA, PCIe, USB).

This!

The individual attributes of the file(s) being written have a massive impact on speed. ATTO is significantly more comprehensive in testing this than Black Magic.

MBP 16 1TB Internal SSD (700TB used):

 

[petsk]

It seems like the slower sequential write speeds are a problem with the iMac 2019 itself. Here are my benchmarks for Samsung X5 512GB.

iMac 2019 (left) vs Macbook Pro 16" (right).

I don't care so much about the sequential write speeds anyway, as I'm going to use it as a boot drive. The 4k random writes and read are even faster than on my Macbook Pro. And even better, I got the drive on sale for only 100€ including shipping and tax.