Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Status
The first post of this thread is a WikiPost and can be edited by anyone with the appropiate permissions. Your edits will be public.
Note, I have reservations about the Samsung 980 PRO SSD. I had problems with it when I used it as the internal drive inside my MP 6,1 to boot macOS Monterey and Ventura. I have not undertaken extensive testing with it as a data disk.
I had this concern since I'm about to put my 970 EVO Pluses to work with my upcoming Mac Studio, and with some research found that it's old news -- Samsung updated the firmware a couple of years ago and they work great with Mac. Don't know if you can upgrade the firmware on Macs or not, however -- I'm currently on Windows until my Mac gets here so I was able to do it.
 
Last edited:
I had this concern since I'm about to put my 970 EVO Pluses to work with my upcoming Mac Studio, and with some research found that it's old news -- Samsung updated the firmware a couple of years ago and they work great with Mac. Don't know if you can upgrade the firmware on Macs or not, however -- I'm currently on Windows until my Mac gets here so I was able to do it.
I applied the (at the time latest) firmware update to my 980 PRO and it made no difference in the poor performance with the MP 6,1. I infrequently use the enclosure in which I have the SSD installed. I have encountered at least one instance where the disk required several minutes to mount which suggests to me the problem may also extend to using them with external enclosures. I do not have a 970 EVO so I cannot offer any experience with these SSD's.
 
Last edited:
These will not be RAID, thus following the advice here of getting a separate enclosure for each to get 4 lanes and the fastest speeds.
RAIDing the drives does not in and of itself determine the bandwidth. RAID and the number of TB controllers you have determines the potential max bandwidth. The Studio Ultra has multiple TB controllers so you will be able to achieve max speed by connecting each drive to a separate controller and creating a RAID0 disk. Running RAID improves throughput because it accesses both drives simultaneously, but it does run the increased risk of data corruption. Do not connect both of the individual drives to the same controller or a hub.


Any reason not to get two Acasis TBU401s for the kind of work mentioned above?
If you are using them continuously they tend to get warm. So, pay particular attention the the power dissipation of the SSD you install and use a low power SSD. If you have not already looked at it, SONNET sells an enclosure that has a cooling fan.
 
Last edited:
  • Like
Reactions: Headlands
Fascinating to watch Amazon prices change. 2 hours ago, most NVMe 4TB drives £4 cheaper.

Aside, also interesting to see speed/heat differences between 1GB test file vs 8GB on old SATA SSD's (830 & 850 EVO).
If you are using them continuously they tend to get warm. So, pay particular attention the the power dissipation of the SSD you install and use a low power SSD. If you have not already looked at it, SONNET sells an enclosure that has a cooling fan.
What is that? Not seeing a SONNET 40Gbps enclosure with a fan. Can only find https://www.amazon.co.uk/Sonnet-Echo-Dual-NVMe-Thunderbolt/dp/B09TVFNLJT/ but no mention of fan on the page.
 
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
If you are using them continuously they tend to get warm. So, pay particular attention the the power dissipation of the SSD you install and use a low power SSD. If you have not already looked at it, SONNET sells an enclosure that has a cooling fan.
Gotcha. Do you have recommendations for an SSD that generates less heat/uses low power? 970 EVO and EVO Plus use a fair amount of power, I believe.
 
Gotcha. Do you have recommendations for an SSD that generates less heat/uses low power? 970 EVO and EVO Plus use a fair amount of power, I believe.
I found this database helpful, I found the reference in a Dortania OCLP issue discussing SSD's.

 
I had this concern since I'm about to put my 970 EVO Pluses to work with my upcoming Mac Studio, and with some research found that it's old news -- Samsung updated the firmware a couple of years ago and they work great with Mac. Don't know if you can upgrade the firmware on Macs or not, however -- I'm currently on Windows until my Mac gets here so I was able to do it.
I added a samsung sata evo to my mini late 2012. Lasted a year.
 
Can you describe what you mean by this?

This is the latest card from Sonnet:


or


You won't get the full speed of PCIe Gen 4 for Thunderbolt 3 but it's good future proofing

There are Gen 3 versions that you can get cheaper

Stick it in one of these:


or


or

 
WD_BLACK SN850X 4000GB : Apple M1 - 1GB.png WD_BLACK SN850X 4000GB : Apple M1 - 8GB test.pngWD_BLACK SN850X 4000GB : Apple M1 8GB 1mm.png

Okay I am thoroughly confused. According to Amorphous Disk Mark my new SN850X 4TB is not performing as fast as it should (with 1 & 8GB test files), but black magic is reporting same speed for Read & Write (L 1GB, R 5GB)

DiskSpeedTest - 1GB test file - SN850x.pngDiskSpeedTest - 5GB test file - SN850x.png

Was hoping that the 4TB would perform as fast as the speeds reported here for the 2TB version. It seems like great read speeds, but slow for writes. (The 4TB has chips on top and bottom)

M1 MacBook Air, 8CPU/GPU, 16GB Ram, 1TB disk

Acasis TBU401 + WD SN850X 4TB

Was also hoping I'd be able to see the NVMe via Parallels for the WD software, but that doesn't seem possible either.

This is while plugged in to power, using the provided TB4 cable, with a 2mm thermal pad, currently trying the supplied 1mm on the top of the NVMe.

Am I best returning? Trying a thermal pad underneath as well?

Tempratures running around 50-58ºC during testing.
 
Interesting (?) ... further testing using AJA with 16GB and 64GB test files I am getting the speeds I would expect (I could confirm them in iStat Menus). Amorphous DiskMark 4.0 seems to be the outlier here. (L 16GB test file, M 64GB, R 1GB)

I haven\t yet found a combination that gives a file test larger than 64GB.
AJA 16GB test file.pngAJA 64GB test file.pngAJA 1GB test file.png
Output stats from larger test file:

Disk Test
Resolution: 8K
File Size : 64 GB
Codec : 10bit YUV
Drive : /Volumes/Ext_4TB_WD
Cache : Disabled
Video file: Movie

Disk Write Test
Number of frames = 728
Write rate = 30 frames/second
Write rate = 2743 MB/second
Minimum rate = 2661 MB/Sec
Maximum rate = 2822 MB/Sec

Disk Read Test
Number of frames = 728
Write rate = 32 frames/second
Write rate = 2927 MB/second
Minimum rate = 2870 MB/Sec
Maximum rate = 2958 MB/Sec

So not sure what to believe is going on 🤔🤷

---- EDIT interesting tidbit ... when testing with Amorphous, if I read a local folder (QuickLook) at the same time as the write test, the results for the write test are higher.

SEQ1MQD8 WriteSEQ1MQD8 ReadSEQ1MQD1 WriteSEQ1MQD1 ReadRND4KQD64 WriteRND4KQD64 ReadRND4KQD1 WriteRND4KQD1 Read
16MiB988.543021.76869.572458.71346.511067.2452.1751.98
32MiB1349.713087.121134.002470.85318.461268.3744.7853.12
64MiB1264.743114.66958.422578.25310.471299.1241.3251.06
128MiB1071.133131983.982589.84299.381310.9440.4150.09
256MiB1065.443130.65880.142593.35303.351330.5639.3151.13
512MiB1037.453126.86989.52580.59301.651320.5039.7953.02
1GiB1016.443129.91857.192588.41302.101334.4838.7751.17
2GiB2254.92*3130.591807.72*2572.40286.231222.4840.0651.85
4GiB2618.47*3021.76944.452564.37276.041307.0739.4051.25
8GiB1074.503118.512409.48*2644.36287.791329.1441.5753.35
16GiB1991.71*3133.742491.66*2572.19278.281337.2340.3259.72
32GiB1544.10*3130.741807.22*2557.71271.691310.4341.8250.93
64GiB1716.34*3129.671799.70*2580.38258.94*1312.3938.82*53.33
1GB - No activity
1001.83​
3130.02​
854.97​
2569.58​
299.01​
1336.78​
39.33​
51.16​
1GB QL on folders on primary driver, at same time as write to external
2163.79*​
3136.94​
1830.67*​
2578.35​
296.66*​
1306.20​
42.09*​
51.79​

* reading doing QL on a folder during write test to external disk (observed same increase in write speeds when looking at output of `iostat -w1 disk6` and the increase is ONLY on the SEQ1MQD8 & SEQ1MQD1 write results, no affect on RND)
 
Last edited:
Okay I am thoroughly confused. According to Amorphous Disk Mark my new SN850X 4TB is not performing as fast as it should (with 1 & 8GB test files), but black magic is reporting same speed for Read & Write (L 1GB, R 5GB)

Was hoping that the 4TB would perform as fast as the speeds reported here for the 2TB version. It seems like great read speeds, but slow for writes. (The 4TB has chips on top and bottom)

M1 MacBook Air, 8CPU/GPU, 16GB Ram, 1TB disk

Acasis TBU401 + WD SN850X 4TB

Was also hoping I'd be able to see the NVMe via Parallels for the WD software, but that doesn't seem possible either.

This is while plugged in to power, using the provided TB4 cable, with a 2mm thermal pad, currently trying the supplied 1mm on the top of the NVMe.

Am I best returning? Trying a thermal pad underneath as well?

Tempratures running around 50-58ºC during testing.

<snip>


So not sure what to believe is going on 🤔🤷

Forget the benchmarks. How does it perform in real world? Have you encountered any actual problems? What's your use case?

Your temps look totally fine. Nvme drives tend to go up to almost 70ºC during intense workloads so 50-58º is well below any kind of throttling limit.
 
Forget the benchmarks. How does it perform in real world? Have you encountered any actual problems? What's your use case?

Your temps look totally fine. Nvme drives tend to go up to almost 70ºC during intense workloads so 50-58º is well below any kind of throttling limit.

a) hard to ignore when its so visibily slower than others. Wasn't expecting jump from 2TB to 4TB to make such a difference. I’d have saved the money for now if I’d realised. (Or waited for results from others, or swapped things around and used Acasis with 2TB drive, and then a USB 3.2 10/20Gbps enclosure for a 4TB disk)

B) mixture of uses. Photo library, raw editing, Virtual Machines, sound libraries for Logic / GarageBand / Music Maker (via Windows VM), offloading dev stuff, video storage / access, backups, data access

C) then occasional backups from it to 5TB mechanical HDD
 
a) hard to ignore when its so visibily slower than others. Wasn't expecting jump from 2TB to 4TB to make such a difference. I’d have saved the money for now if I’d realised. (Or waited for results from others, or swapped things around and used Acasis with 2TB drive, and then a USB 3.2 10/20Gbps enclosure for a 4TB disk)

B) mixture of uses. Photo library, raw editing, Virtual Machines, sound libraries for Logic / GarageBand / Music Maker (via Windows VM), offloading dev stuff, video storage / access, backups, data access

C) then occasional backups from it to 5TB mechanical HDD

It's only "so visibly slower" in one benchmark app out of three. You have to keep in mind that these apps aren't perfect and they don't account for every kind of hardware/controller combination.

Have you tried doing an actual large sequential transfer and looking at the realtime transfer speed in Activity Monitor? Majority of your use case is more dependent on random I/O speeds anyway.
 
there are some good You Tube reviews around showing how Lightroom and other photo and video tools don't benefit much from disk write speeds beyond a certain level as the ssd speed is not the slow point in the chain. Your AJA speed tests show results way above a level where I would have expected you to notice any difference !
 
  • Like
Reactions: AAPLGeek
It's only "so visibly slower" in one benchmark app out of three. You have to keep in mind that these apps aren't perfect and they don't account for every kind of hardware/controller combination.

Have you tried doing an actual large sequential transfer and looking at the realtime transfer speed in Activity Monitor? Majority of your use case is more dependent on random I/O speeds anyway.

"these apps aren't perfect" Indeed, as I'm discovering. Although also my own fault for extrapolating that great 2TB results go up to 4TB results. (If I had more disposable income I'd be intrigued to buy and test out Firecuda 530 4TB, Silicon Power 4TB XS70 (which has now disappeared after its £244 price) in the same setup)

Yes I have been trying out SQ transfer and also standard transfers. Also just finished backup of 4TB HDD to the NVMe.

there are some good You Tube reviews around showing how Lightroom and other photo and video tools don't benefit much from disk write speeds beyond a certain level as the ssd speed is not the slow point in the chain. Your AJA speed tests show results way above a level where I would have expected you to notice any difference !

Good to know, thank you. Will check them out.

Is the TBU401 the only thing connected to the Thunderbolt ports on your Mac?

The other port has generally just got power connected. Sometimes a USB-C hub with power going through it / or a spinny HDD.

When I did the chart results, I only had power and the NMVe disk connected.
 
I'm using a Trebleet enclosure with a Samsung 980 Pro NVMe. Results with a white TB3 cable, connected to my new Studio M2 Ultra. See original post, for technical details.

Samsung 980 Pro NVMe connected to Studio M2 Ultra with Apple TB3 cable, it maxes the Trebleet specs:

1688160352537.png

Internal Studio M2 Ultra NVMe:

1687898223882.png
 
Last edited:
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
2 pci lanes per slot.
Means 1500MB/s per stick. (And, amazingly, they say this on product page.)
Means 24Gbit/s for the storage. When in full use, 16Gbit/s is then left for usb.
When you configure the two SSD's as a RAID0 array, you maximize the bandwidth using all four PCIe lanes, or 3000 MB/s.
 
How can I check the temperatures of my external NVME drive? I downloaded TG Pro, which was advised from a quick google search, but it doesn't show the temperatures of my Acasis TBU405.

Acasis TBU405 connected via Thunderbolt to a Mac Studio M1 Max. I only see temps for my internal drive:
0.png
 
Are there any enclosures with the JHL8440 nowadays? I'm also curious to see how the ASM2464PD (ZikeDrive review) works with 4TB drives on M1 machine MacOS. If you're going with purely USB4 controllers the VLI VL830 is probably one to keep an eye on.
 
Last edited:
Yes this is fine. Get the 1.5mm one. 100% working on TBU405 and WD SN850x.
I've just purchased a WD Black SN850X 4TB and I'm still in the market for a good external enclosure which should also work with USB4 (40gbp/s) devices in the future. Unfortunately on the controller side things seem somewhat lacking, but I have been looking at the TBU405. Given your earlier posts I'll make sure to buy a thermal pad sheet of 1.5mm thickness.

I have read a couple of comments that the supplied cable might be lacking, so I suppose I have some research to do on high quality TB4 cables or USB4 (40gbp/s). But this is where I'm getting a little uncertain about thing as I'm pretty sure it's not going to work with an actual USB4 cable.

TB4 however is backwards compatible with TB3 and any proper USB4 port implementation is going to support TB3 at least, so going with a TB4 cable seems like the logical choice. I would like to see someone test this though. I don't own any USB4 devices yet.

In any case I decided to order directly from their website which was easily the cheapest option for me.

I'm still looking for a solution to be able to update the firmware of the WD Black. I have an old windows computer without any USB-C or TB ports. I've read that the WD Dashboards will need a PCIe connection to read the disk, so in theory this should work with my M1 MBP using a TB3 connection. Unfortunately it seems like the software for this doesn't exist.

It makes me wonder if using something like Parallels might make it possible to sort out the firmware update?
 
Last edited:
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.