Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
btw semi related to this thread: I have a Realtek and JMicron Usb3.1Gen2 enclosure. Crucial works expected 700-800mb/s in both, Samsung EVO 970 (nonplus) only works well with JMicron and speed drops to 30mb/s in the Realtek.
 
  • Like
Reactions: AAPLGeek
Thanks for all the helpful information in this thread. I’m a photographer that just got my first drone. I’ll be shooting 4k 30 and probably some 4k 60. shooting for fun. Nothing huge at this point. What kind of speeds do I need to edit 4k footage from a (future) Mac studio with an external drive. For example, I don’t want to pay for a drive with speeds of 2800 when all I really need are speeds around 1000. Thanks in advance.
 
Thanks for all the helpful information in this thread. I’m a photographer that just got my first drone. I’ll be shooting 4k 30 and probably some 4k 60. shooting for fun. Nothing huge at this point. What kind of speeds do I need to edit 4k footage from a (future) Mac studio with an external drive. For example, I don’t want to pay for a drive with speeds of 2800 when all I really need are speeds around 1000. Thanks in advance.

I'm assuming that you want to edit footage that's on the external drive rather than edit it on your internal drive after importing it. That's a common scenario, and indeed how professional film editors work. I haven't gotten my hands on a drone yet, but I shoot video with Blackmagic's Pocket Cinema Camera 4K, and Samsung's T7 drives are a deserved favourite on the Blackmagic forum. There are also a couple of serious competitor drives that others may have experience with and can comment on.

I was checking the price of 2TB T7s a couple of days ago. According to CamelCamelCamel, which tracks Amazon prices, in the U.S. the 2TB T7 is at its lowest price ever. It's also only US$10 more than a 2TB T5. I'm not at all suggesting that you buy a 2TB drive, it was just what I happened to be checking. The 1TB drive may also be at its lowest price, or close to it, you'd have to check. I'm talking about the standard T7, not the one with fingerprint security or the new ones with enhanced protection from water, dust, etc., about which there are a couple of posts earlier in this thread. T7 drives with these additional features sell for somewhat more money.

For editing, I use Final Cut Pro X and Blackmagic's DaVinci Resolve for colour grading. I don't know what you plan to use to edit footage, but DaVinci Resolve runs Apple's M1 computers natively, and the free version of the app may well meet your needs. It's extremely good software, including the free version.

This is a Blackmagic Disk Speed test that I ran with my Mac Studio and a 2TB T7 a few days ago, using a 5GB load. This is plenty fast enough for 4K editing, at the professional as well as amateur level.


test.png
 
Last edited:
  • Like
Reactions: B.A.T
Hey folks,
Just received 2 x Sabrent TB3 to M.2 NVMe enclosures and installed 2 x Samsung 970 Pro 1T into each that I previously had on 2 x PCIe cheapo cards in my 2019 Mac Pro as a single raid0 volume.

This is what I'm getting from Amorphous
Screenshot 2022-07-12 at 13.02.52.png

with these temps (obviously haven't hammered it yet)
Screenshot 2022-07-12 at 13.05.23.png


Might try a reformat of the Raid0 see if it speeds it up a bit....
 
  • Like
Reactions: F-Train
I have been following this thread because I thought my external SSD speeds seem a bit low, especially compared to the Mac Pro 6,1 with the same drives connected to it.

I have a Mac Studio M1 Max, 32 core GPU, 64 GB of ram, and 1 TB internal drive. Internal SSD speeds are great, but I am surprised at the speed of one external SSD. I have a Samsung SSD 960 PRO 1TB housed in an OWC Envoy case, which is Thunderbolt 3. Still, the system report shows the theoretical speed of connections to 40 Gb/s:

Port (Upstream):
Status: Device connected
Link Status: 0x2
Speed: Up to 40 Gb/s x1
Current Link Width: 0x2
Link Controller Firmware Version: 0.31.0

The Blackmagic Speed List Test scores for the drive are surprising. The write speed with the drive connected directly to the Studio TB ports are better, 944 MB/s, then the drive connected to a CalDigit Element dock, 629 MB/s, but really only slightly. Read speeds are comparable. However, these results seem to be inline with the results F-train is getting. Perhaps the Envoy enclosure is slowing the speeds down since it has been suggested that the enclosure can affect performance. Maybe a Sabrent case will give me speeds closer to those reported by Grif since the SSD is similar to the one I have. Thoughts?

I transfer very large samples files (100 GB) to this drive, so speed can be important.
 

Attachments

  • DiskSpeedTest-2.png
    DiskSpeedTest-2.png
    243.5 KB · Views: 154
  • DiskSpeedTest.png
    DiskSpeedTest.png
    243.2 KB · Views: 147
Thank you for the explanation @F-Train and for everyone else. Really good info in here for a newbie like me. When I upgrade I'm looking at an M1Max studio with 1TB storage 64 GBs of ram and a couple of external 2TB drives pretty much as described a couple of posts above here by F-Train. I'm also going to get a 2 bay enclosure with spinning discs for backups/redundancy. All of the info in here helps a lot :)
 
  • Like
Reactions: F-Train
btw semi related to this thread: I have a Realtek and JMicron Usb3.1Gen2 enclosure. Crucial works expected 700-800mb/s in both, Samsung EVO 970 (nonplus) only works well with JMicron and speed drops to 30mb/s in the Realtek.

I'm considering getting a RTL9210 based enclosure for my Samsung 970 PRO. Perhaps I'm better off getting one with the ASMedia or JMicron chipset then? EVO and PRO models both use the same controller as far as I know.

I currently use a T5 SSD and I'm basically looking for a external solution with a better random I/O performance. Could you post your AmorphousDiskMark test scores with the RTL9210 enclosure so that I have something to compare to?

T5 Random IO.png
 
Have a look at post #84, where I talked about the difference between how the two enclosures are constructed. You may have a preference when it comes to an enclosure that's held together with screws (albeit small ones) and an enclosure that's held together by friction.
.......

Not very helpful, am I :)
You have been very helpful 👍👍

My Acasis enclosure with my 2TB Samsung 980 is pictured below. For my heatsink I found on a solid copper one and it is held in place by clear bands, of which they give you a few extra. There are 2 heat sinks in the pack, with some cleaning wipes and 2 thickness of tape for $9.99. The thiner tape worked fine and my enclosure snapped back together with just the slightest resistance. Here is the link:


Besides the issue with the snap on design, the Acasis box has a rubber grommet that is a bit too big to fit the tiny screw hole that holds the NVME. (Why didn't they provide a simple screw?) I got it to fit the first time but then it fell out. While I have lots of computer screws, I did not have one small enough to fit the NVME thead. So simple solution was to cut a 3/8" length of a heavy duty toothpick and push it in place, and now the NVME does not slide around. Once the lid is on, it is held in place by friction anyway. Note that the heatsink is resting against the bottom of the enclosure, and as heat rises, I turn my enclosure upside down when in use and this way the heatsink is at the top.


Acasis w: Copper Heatsink.jpg
 
Last edited:
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
  • Like
Reactions: turk101 and F-Train
I ordered a TEKQ Cube from Amazon and it arrived today. The cellophane wrapper had been neatly sliced around the lid of the box. There was no heat tape in the box. Is there supposed to be a piece of heat tape? I’m trying to decide if I should send it back for an unopened replacement. Thanks.
I dropped of the enclosure at Whole Foods and let them handle the return. I paid for a new item not a return that I have no idea what was done with it.
 
Last edited:
One note to everyone because I just went through this with a similar enclosure, Do NOT buy a Gen 4 NVMe stick. Thunderbolt 3/4 can’t take advantage of the extra speed at all, yet the drive will heat up so fast in one of these enclosures that it will rapidly heat up and fail. Only use Gen 3 models. TLC being the fastest.
 
One note to everyone because I just went through this with a similar enclosure, Do NOT buy a Gen 4 NVMe stick. Thunderbolt 3/4 can’t take advantage of the extra speed at all, yet the drive will heat up so fast in one of these enclosures that it will rapidly heat up and fail. Only use Gen 3 models. TLC being the fastest.

Many of the posts in this thread, possibly most (I haven't counted), that provide information on NVMe performance are about Gen 4 SSDs. The point that Thunderbolt 3/4 drives can't take advantage of Gen 4 speed has already been made. Several people have chosen Gen 4 SSDs anyway, given what's available in Gen 3 and the poor write performance of Samsung's Gen 3 970 EVO Plus especially.

None of the people who have posted about Gen 4 SSDs have reported unusually high heat levels. This might be because the Gen 4 SSDs that people have discussed in this thread are more efficient than Gen 3 SSDs with respect to power consumption, and are running at lower read and write speeds than they are capable of precisely because the connection is Thunderbolt 3/4.

You say that your Gen 4 SSD got so hot that it failed. Nobody else has reported that, or anything close to it. Maybe you just got a bad SSD. It would be helpful to know which SSD you're talking about, which enclosure you were using and how the SSD was being used when it failed.
 
Last edited:
I'm considering getting a RTL9210 based enclosure for my Samsung 970 PRO. Perhaps I'm better off getting one with the ASMedia or JMicron chipset then? EVO and PRO models both use the same controller as far as I know.

I currently use a T5 SSD and I'm basically looking for a external solution with a better random I/O performance. Could you post your AmorphousDiskMark test scores with the RTL9210 enclosure so that I have something to compare to?

View attachment 2029305
The JM583 is an older 970 EVO (4years old)
sequential are around 700mb/s for the JM583.

Not sure if it's M1's fault tho, people tend to get lower external speeds with M1.

JEYI NVME = RTL9120, loaded with 1TB Crucial P1
Not sure why i get such abysmal random writes even on the INTERNAL!

Threw in the 2TB 980 Pro in the JEYI Thunderbolt enclosure (1st gen from 2018). It's still faster than the internal 512GB in every measurement.

so this is all of my working drives. I connected the USB ones directly to the port on M1 for the test, the 980 Pro i tested through the OWC Thunderbolt Hub.
 

Attachments

  • Screenshot 2022-07-14 at 23.00.33.png
    Screenshot 2022-07-14 at 23.00.33.png
    698.2 KB · Views: 147
  • Screenshot 2022-07-14 at 22.53.10.png
    Screenshot 2022-07-14 at 22.53.10.png
    616.2 KB · Views: 131
  • Screenshot 2022-07-14 at 22.57.41.png
    Screenshot 2022-07-14 at 22.57.41.png
    386.1 KB · Views: 147
  • Screenshot 2022-07-14 at 23.02.37.png
    Screenshot 2022-07-14 at 23.02.37.png
    401.7 KB · Views: 153
Many of the posts in this thread, possibly most (I haven't counted), that provide information on NVMe performance are about Gen 4 SSDs. The point that Thunderbolt 3/4 drives can't take advantage of Gen 4 speed has already been made. Several people have chosen Gen 4 SSDs anyway, given what's available in Gen 3 and the poor write performance of Samsung's Gen 3 970 EVO Plus especially.

None of the people who have posted about Gen 4 SSDs have reported unusually high heat levels. This might be because the Gen 4 SSDs that people have discussed in this thread are more efficient than Gen 3 SSDs with respect to power consumption, and are running at lower read and write speeds than they are capable of precisely because the connection is Thunderbolt 3/4.

You say that your Gen 4 SSD got so hot that it failed. Nobody else has reported that, or anything close to it. Maybe you just got a bad SSD. It would be helpful to know which SSD you're talking about, which enclosure you were using and how the SSD was being used when it failed.

It is certainly possible it was a bad stick. It was fast - nearly 3GB/sec, but heated up a LOT. Within a few minutes it would top over 70C and fail.

I replaced it with an Inland Gen 3 stick which is closer to 2500/sec but it runs relatively cool. I am running off both a Ineo Thunderbolt 3 NvMe enclosure as well as a "Konyead" TB 4 enclosure. I also tried an Orico. All of them seem to have identical boards on the inside.
 
  • Like
Reactions: F-Train
I replaced it with an Inland Gen 3 stick

Micro Center's parent company owns Inland and sells Inland SSDs as a Micro Center house brand through their stores and on Amazon U.S. Pricing is apparently attractive. It would be great if you could post AmorphoousDiskMark or Blackmagic Disk Speed read and write results for the Inland SSD that you have.

For those who aren't familiar with Micro Center, it's a U.S. retailer of computer gear and components with 25 stores in 16 states. It also sells online. Wikipedia: https://en.wikipedia.org/wiki/Micro_Center
 
Last edited:
Micro Center's parent company owns Inland and sells Inland SSDs as a Micro Center house brand through their stores and on Amazon U.S. Pricing is apparently attractive. It would be great if you could post AmorphoousDiskMark or Blackmagic Disk Speed read and write results for the Inland SSD that you have.

For those who aren't familiar with Micro Center, it's a U.S. retailer of computer gear and components with 25 stores in 16 states. It also sells online. Wikipedia: https://en.wikipedia.org/wiki/Micro_Center
Screen Shot 2022-07-13 at 2.00.29 PM.png


This was a first run out of the box. It does vary of course as he use it but it pretty much days in this range. Ignore the Gen 2 I meant Gen 3.

For comparison, here was the Gen 4 stick in the same enclosure. But it heated up and failed within a few minutes and I had to return it.

1657894737502.jpeg
 
  • Like
Reactions: F-Train
General question here. Anybody know if you have a four bay enclosure can you run RAID 0 on drives one and two and RAID 1 on drives three and four?
 
General question here. Anybody know if you have a four bay enclosure can you run RAID 0 on drives one and two and RAID 1 on drives three and four?
any of the OWC 4+ thunderbay models, but it requires their soft raid software.
(raid 0 can be done using the apple disk utility but there's not much support)
 
General question here. Anybody know if you have a four bay enclosure can you run RAID 0 on drives one and two and RAID 1 on drives three and four?

Assuming that we're talking about a Mac Studio computer, the utility of this is going to depend on what type of drives you want to use and how many Thunderbolt 4 ports the enclosure has. As a practical matter, each Mac Studio Thunderbolt 4 port won't support read and write speeds of more than about 2800MB/s, and I've only seen a couple of cases where that speed shows on tests. If you want more, you need to use more Mac Studio ports, which also means more enclosure ports. If you haven't seen them yet, there are a number of discussions about these limits in earlier posts.

For example, the read and write speeds shown in post #85 (3863MB/s write, 4374MB/s read) were only possible because I used a RAID 0 consisting of two NVMe SSD enclosures connected to two Mac Studio Thunderbolt 4 ports. With a dual enclosure and a single cable running from the enclosure to my Mac Studio, I would have been limited to something under 2800MB/s.

Re @Icubed's comment just above, I've had no problems using Monterey's Mac Disk Utility for RAID 0, and it also supports RAID 1.
 
Last edited:
  • Like
Reactions: anticipate
My thoughts are that I'll be buying the OWC Express 4M2 and using with a Mac Studio. It has max speeds of 2800 MB/s but I don't care to get that high. Really what I'm looking for is 4TBs of external storage at a decently high rate of speed (RAID 0 drives) and a couple of drives to back stuff up with RAID 1 all in one enclosure. I'm not a pro just a hobbyist. This seems to me (possibly naively) that it will last me for several years.
 
  • Like
Reactions: thoomp and F-Train
Re @Icubed's comment just above, I've had no problems using Monterey's Mac Disk Utility for RAID 0, and it also supports RAID 1.
i should rephrase that as 'don't expect any apple tech support for RAID 0 and RAID 1 disk utility issues'
 
The JM583 is an older 970 EVO (4years old)
sequential are around 700mb/s for the JM583.

Not sure if it's M1's fault tho, people tend to get lower external speeds with M1.

JEYI NVME = RTL9120, loaded with 1TB Crucial P1
Not sure why i get such abysmal random writes even on the INTERNAL!

Threw in the 2TB 980 Pro in the JEYI Thunderbolt enclosure (1st gen from 2018). It's still faster than the internal 512GB in every measurement.

so this is all of my working drives. I connected the USB ones directly to the port on M1 for the test, the 980 Pro i tested through the OWC Thunderbolt Hub.

Thanks. Looking at those scores, M1's USB implementation is severely gimped.

Random QD1 is the single most important metric in small sized transfers and everyday tasks and if we look at the write speeds, all those scores are lower than the T5 connected to my Intel iMac. That includes M1's internal SSD as well your thunderbolt connected 980 Pro!

The RTL9210 JEYI enclosure is also reporting much lower speeds. The sequential reads is almost 30% lower than it should be.

To get a better picture, I connected my T5 drive to my M1 MBA and it only connects as a USB 5Gbps drive. Consequently, the scores are much lower. In fact, the scores are even lower than the same drive connected to the USB 3.0 5Gbps port of my Intel iMac.

It's sad that Apple has completely ignored the whole issue so far.
 

Attachments

  • M1 Gimped USB IO.png
    M1 Gimped USB IO.png
    957.2 KB · Views: 164
  • T5 5G Intel.jpg
    T5 5G Intel.jpg
    239.8 KB · Views: 148
  • T5 10G Intel.jpg
    T5 10G Intel.jpg
    230.2 KB · Views: 130
Last edited:
  • Like
Reactions: Ploki
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
Thanks. Looking at those scores, M1's USB implementation is severely gimped.

Random QD1 is the single most important metric in small sized transfers and everyday tasks and if we look at the write speeds, all those scores are lower than the T5 connected to my Intel iMac. That includes M1's internal SSD as well your thunderbolt connected 980 Pro!

The RTL9210 JEYI enclosure is also reporting much lower speeds. The sequential reads is almost 30% lower than it should be.

To get a better picture, I connected my T5 drive to my M1 MBA and it only connects as a USB 5Gbps drive. Consequently, the scores are much lower. In fact, the scores are even lower than the same drive connected to the USB 3.0 5Gbps port of my Intel iMac.

It's sad that Apple has completely ignored the whole issue so far.
Yes known issue on M1 chips - USB 3 speeds are slower than the Intel units. Thunderbolt speeds are fine though. Thunderbolt 4 and USB 4 are the same thing, practically.
 
Yes known issue on M1 chips - USB 3 speeds are slower than the Intel units. Thunderbolt speeds are fine though. Thunderbolt 4 and USB 4 are the same thing, practically.

Unfortunately, that doesn't appear to be the case.

Random write performance is still abysmally low compared to the Intel machines. Here are a couple of scores I found online. The random write speeds on M1 machines are lower than 1/10 of these scores.

Intel Thunderbolt P34A80.png



Intel Thunderbolt SPG GAMMIX S50.png
 
  • Like
Reactions: DeepIn2U
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.