Thunderbolt 3 vs USB 4 External SSD | M1 Mac 

Fantastic! Really glad to help. I go a little deeper into it in my most recent video. Best of luck on your purchase.

Ha! This Gerald Undone level is a breakneck workflow. I think I'll go back to vlogging 😆 Good to see you brother.

That’s fantastic. A Mac Studio on the way! 🚚📦 You’ll be blown away by that machine.

That rabbit hole is deep. Glad to see you made it out. Awesome to have you here.

Awesome. Glad to have you around here on this one. 🤘

Thanks! And it certainly can go a little deeper than this, but hopefully they can resolve the issue in the next iteration and open up more PCIe lanes or more power to the controller.

Thanks! I have some more recent updates as well, but I do still use these enclosures.

Glad to help. If you need to sandbox the OS you can boot from it, or just hold the OS and apps in the internal and run your libraries, gaming data, etc off of the external as well.

You bet. And thank you.

Thanks friend! I appreciate you hanging out. It can be tough because reviewing tech (or anything) has the ability to upset some folks if the results aren’t leaning to the correct narrative. I’m certainly not trying to create buyers remorse, but sometimes I find that the items we purchase aren’t exactly as advertised 🤷‍♂️ Anyway, I’ll keep speaking the truth 😉

From Brazil

Thanks! Glad to have you here on this one.

Glad that it helped. Best of luck.

You bet. I like to spread the wealth around to these benchmarking programs, but Amorphous is what I personally pay attention to.

Thank you. This one took quite a bit of research.

It’s my pleasure. Glad to help where I can.

Thanks so much. Appreciate having you here.

@@KevinRossRN No problem! 💯

Hey that’s awesome! And, I did an update on the Acasis and 980 PRO, and they are a great combo. I use it all the time. Great to have you around the channel and I appreciate you sharing your workflow as well. 🤘

@@KevinRossRN Thank you 😊

So for external use right now, and likely in the near future it will be harder to get that full throughput of Gen4, however in my case (and probably yours), I was able to find a Gen4 SSD at the same price, so I just grabbed it. I also do like USB4 for that protocol that it works with USB-C, so I do think it’s worth it there.

@@KevinRossRN Thanks much I'm going the same way too!

Awesome 😎 More testing to occur. There’s just a backlog of tech I’m waiting on.

Much appreciated.

Many thanks. 🤘

You bet. So there has been some slight improvement with the communication protocol, specifically in my latest video where I tested the Samsung 980PRO. I was getting slightly better results with this enclosure as well as the Fledging with that SSD. I’d say that this isn’t the case with all SSDs on the M1, but I am still testing. Hoping to get some Sabrent SSDs to test because it sounds like some of them are providing better results now.

@@KevinRossRN Thank you Kevin!

My pleasure.

Really appreciate you being here friend.

That’s the idea (real use case). And, thank you.

Thanks! I do put a lot of work into these. Glad to help.

Hey, when you’re cutting video, you’ve got to have sharp tools 😉🤣

@@KevinRossRN 💯 !

I’ve got a similar setup and when I got the M1 it was as if I got a Mac Pro. It’s a little beast. Have you looked into the TB3 and TB4 hubs from Caldigit or OWC? Those will give you some extra I/O.

Of course! Happy to help. I think I’m going to build out my own testing lab 🤣

Hi Kev. It’s Kevin in Toronto writing. Thanks for this video, which was slightly beyond my scope. However, I was going to suggest to you previously that perhaps in a future video you could talk about or mention USB-C and Thunderbolt 3 - that although they share the same plugin, they are NOT the same thing. In some instances, I don’t think that some cable manufacturers even understand the differences between the two. I have found that finding a proper Thunderbolt 3 cable isn’t easy. On Amazon, they are listed as Thunderbolt 3/USB-C cables or adapters, to which I ask myself “well, which one are you?” The USB-4 configuration that you mention being something Apple has done with the M1 chip is not something I am too familiar with, so thanks for going to the trouble of enlightening those of us that may not know. By the way, I have been finding glitches with the Big Sur OS that is infuriating at times. It’s been ejecting my external drives for no reason while I’m working in them. Out of the blue I’ll get a pop up advising that the external drive was ejected improperly. This goes for the T7 as well as my SATA Drive and a Flash stick. Very weird. Perhaps I’ve got a ghost in the machine! Hope all is well.

Hey Kevin, great to see you over here again. You know, it’s funny. I had a section in the video about the cables, but decided to cut it out. I was trying to trim it down a bit. Specifically I discussed the passive vs active cables, and yes, you are right when it comes to the confusion of those cables. As far as the randy ejecting, are you plugged directly in I assume? I’ve had someone else share their issue with using a hub, which hopefully with the latest update as of yesterday I believe, this may fix that issue with some of those hubs. However, interesting indeed. I don’t have a T7, but do have a T5, which I haven’t had that issue. I also keep my NVMe enclosures plugged in pretty much all the time. Interesting. I’ll have to investigate further. Have you found any others with the issue?

@@KevinRossRN - Hi Kev. Only the T7 is plugged in directly. The other two are through a USB-A hub. I haven’t clicked on the latest update yet. I’ll do that tomorrow and keep an eye on the ejection rate. Haven’t found anyone else with this same ejection issue. With the Flash drive, I move over a music file from the SATA drive to the Flash drive, and the flash freezes during the write and then just ejects. That was today. But for previous days, just straight ejection out of the blue. It’s not an often occurrence, but definitely a few times in each session. Weird. Never had these issues before. I’m not doing anything unusual or different. All the drives are formatted properly. Anyhow, I’ll figure it out Kev. Just thought I’d mention it.

That is weird, and honestly, frustrating. I wish I could point to a fix. At the very least, let’s hope that the latest update provides some benefit. It is supposed to fix the issue with third party hubs that were bricking the Mac, but I wonder if there’s a few other updates that are snuck into it that may address anything else. I’ll be curious to hear.

Yeah it seems the Ginsu knives benchmark is a hot request. 💀

Thanks! And so to clarify, what would be your main workflow and what’s a rough budget that you have?

No problem! I’d say $150-$200, mainly copying documents and video import and export from and to my m1 MacBook Pro.

So I’d say the SanDisk extreme may be your best bet. The one linked in the description that I tested is plenty fast. They have Western Digital NVMe SSDs in them. Otherwise, if you’re building an enclosure you could get the WD SN750 SSD and then a USB-C Enclosure like this one: amzn.to/37OmwFH It’s not TB3, but on par with the SanDisk speeds, and you can always upgrade the SSD in it. That would bring you in at around 150-160USD

@@KevinRossRN thank you so much! Did not know the sandisk extreme had the WD nvme in them! I will definitely check them out

Yep, the companies actually merged. Western Digital bought SanDisk.

That is correct. You can only get the speeds that either the enclosure supports and the throughput that TB3/TB4 supports. And based on the information I’ve received the Sabrent SSDs have sometimes performed slower in these enclosures, and just seem to do better directly on the motherboard.

Thanks my friend, and yes, I did run those temps... I think I accidentally cut those out of the video...let me go to my notes and I’ll update that in the description. BTW, how are you liking your QNAP?

@@KevinRossRN I really love the QNAP, it is very robust so far (knock on wood) I am currently editing with it with spin disks, in raid. Using it over thunderbolt. I am considering upgrading it with SSD disk, but again, like I tried to say, for working, it´s more than enough. It´s robust, and if a disk fails it costs me minimum. But of course as prices of SSDs keeps dropping and they do also become bigger it is becoming more and more tempting. For 1080 and 4K spin disks in raid is just fine. I did however also of course had to try some 8K files. Then it told me to F ...off. :) But I must really add, by old rabbit I transcode all my files to proxy before working on larger project anyways. And when I did that with 8K it had no problem. I feel really many misunderstands often when looking at reviews the difference in speed, and what formate one should be working with etc. Here I digress once more...sorry! I hope you are well and the family is well to ? :)

Ha! It’s okay to digress. Great to hear. I’m eyeballing a TB connected option to be able to edit from. I’d love to edit off the SSD and then archive on the spinning drives. I do currently back up on a Synology and can edit with Symlinks via FCP, but would love more throughput. All is well here. Hope you’re hanging in there these days.

Yep, the one linked in the description is the one I tested. And no, I haven’t heard that about looking like Rodney Mullen 😏😉 Actually yes, yes I have 🛹 Appreciate you hanging out 🤙

@@KevinRossRN haha, Maybe the hat+hair combo makes the similarities. Interesting, so ~500mb per sec drive is enough for such demanding export and does it pretty much at same speed as ~5000mb per sec SSD. Almost went nuts today with these numbers and research :D That real life comparison was really needed. Going with Sandisk Extreme and moving on with my life haha. Cheeeers! Keep it up!

So the one I linked up...it should say it parentheses as far the product number, it’s the 1050MB/s read and write...but yes, real world use netted those results and I use all of these SSDs constantly and don’t really notice any difference. And yeah, I’m flattered. Quite a few have mentioned the resemblance. He’s a cool dude and an even better skateboarder than I am.

@@KevinRossRN Got it! ;) Cheers!

I have a video coming out today on further tests with two of the top SSDs in this category. For video editing, it really doesn’t matter because the program can only read and write so fast. Are you going with a DIY setup?

Always looking for more tech to compare it to what I have. My wallet doesn’t agree though 😀 This one is definitely on the list though.

Kevin Ross based on what I’ve seen from that drive, I would think OWC would want to provide you a unit to compare!

They did send me an external SSD, no questions asked and no expectations to make a video, but would love the opportunity to test more of their products since as you can see I try not to just run a synthetic benchmark and call it good 😉

@@KevinRossRN well that explains why OWC doesnt send you anything else lol...They dont want the real world #s getting out ...

Ha! I have had those conversations to let brands know that I will be testing and I will share what my concerns are.

Yeah, this is something I’ve been testing on Monterey as well as Big Sur. It’s still drawing power even if the OS isn’t utilizing the drive, so hence the reason why it’s still warm. This seems to be something with these thunderbolt enclosures that I’ve found.

@@KevinRossRN Thanks for the reply! Have you heard of any 2TB SSDs that work with the Acasis and M1 mac with no issues? My 2TB WD SN750 is only getting ~1200mbps write which many other have noted too. I'm worried to try the Samsung 970 EVO Plus because i heard a lot of people having issues with that SSD and M1 macs.

Yeah the SN750 seems to work the best, except for that silicon power one I reviewed a while back. I think there’s been an update to the SN750 SSD components that may be slowing it down, which is related to the manufacturing issues globally. The video I did comparing it to the 970 EVO also resulted in the SN750 performing better, but it also comes down to what you’re using it for. As in this case, the program can only write so fast through TB3/USB4 and regardless of the speed of the SSD, it’s negligible. But again, just depends on what you’re using it for. This will be one of the first things I test on the M1X.

That’ll be interesting if it is. However, once the M1X comes out, I’ll be sure to test it against the M1 since the M1X is supposed to have a higher wattage chip. 🤷‍♂️

The same ones in the description. I have several of the SanDisk models. They’ve been great, and reasonably priced.

Thanks! And yeah, in my research I haven’t found any enclosure yet that can handle that. Up to 4TBs, yes, but I will say that this could be a thermal nightmare if not properly managed. I get sketched out by some of these enclosures, especially when it comes to the components they’re using. That’s why I try to put them under as much load as possible. I am working on something that is more NAS based as far as the hardware, but should be able to have that direct TB connection that would be able to handle those thermals and a few NVMe SSDs. My dream setup would be able to work on those SSDs through TB/USB4 and then archive within the same enclosure to some spinning HDDs...Such a niche workflow. But yeah, the two sided setup would likely cause failure or bottlenecks that these smaller manufacturers couldn’t handle on the customer service side, not to mention the expense because of the type of board they’d need in those enclosures. 🤷‍♂️

I bought the Wavlink too. I had the first replaced, and a Crucial P5 V3 specced not V4. Rated performance something like 3500/3000 yet I can only get around 1000 read and write. I sent the NVMe back to Crucial, and the new replacement performs exactly the same. Tried this on M1 Macs and Intel. Always close to the same. I am really disappointed in this. Is it perhaps an incompatible mix?

I actually did a follow up video that released yesterday touching on a couple of these questions. I’d have to determine which chip you’ve got on that board in the enclosure, but this could be an issue where the enclosure isn’t able to fully push that throughput, but there’s also the enclosure and SSD compatibility. I’m seeing more manufacturers list the SSDs they’ve tested.

Thanks! I still use both of these enclosures. I have two of the Acasis right now. They’ve been performing quite well. I also use them in my iPads.

I’m actually working on something that’ll be TB3 connected. I’m hoping that it’ll come to fruition. Tons of research happening right now.

That’s awesome! I made this one especially for you. 😉😴

I’ve actually updated these videos with the Mac Studio. The one thing I really like about the Acasis is that it’s USB4 so I’m able to use it on USB-C devices. I still use them both though, but the Acasis will have a slight edge on synthetic benchmarks. Real world however, the difference is minimal.

No problem with the Fledging, that is correct. The Acasis was the one that was missing.

The M1 has 16GBs of RAM and the Intel machine is 2017 13 inch MBP i5 with 16GBs of RAM as well.

As far as 2020 iPad Pro, I can’t speak to that, but it should be because my iPad Air 4 recognizes it. It was actually one of the first things I tested since it should be using the USB4 protocol.

I really need to get one of these for testing. I know it’s another go to option for folks. Are you on M1 as well? I assume.

@@KevinRossRN Allow me to correct myself. I switched the drives and the ports and had better "ventilation" than last time. There was no noticeable difference between HFS+ and APFS. Apologies for any confusion. Yes - I have a 4TB Maxed out 16" M1 Max and the experience is completely different. I work on my 2019 27" Core I-9 iMac and keep asking myself "Is this thing frozen?". Loading programs with the older SSD is particularly slow by comparison. My advice, unlike most on RU-vid, is buy AS MUCH of the new M1-Pro/Max as you can afford.

Oh gotcha. I think I responded before the edit. And honestly I think that’s reasonable for real world use…at least in my testing on this video as well as my updated test from last week I posted.

You got it! I’ve continued to test and update the community here on some of my most recent videos where we’re finding some better enclosure and SSD combos, but yes, it has been a pain with these M1s. The updated chips are “better” with this throughput, but we’re still seeing some issues with that communication protocol. Glad to have you here.

Ha! Not even close. Just someone who likes to know how things work I guess. From a very young age I used to take things apart and put them back together. Just sharing what I know, whatever that may be. 🤷‍♂️😉

Yeah, I think they were one of the first of these enclosures I came across with the nut, and as I picked up more I noticed that others started doing the same. However most of these brands are often using the same manufacturers to assemble these and just putting a different name on it. It’s been fine though. Haven’t had any issues.

Thanks! Years of podcasting, plus I suppose the “gift” of my vocal chords 🤷‍♂️ You never know, you might just hear me randomly on a voiceover for a documentary 😉

I insist…Thank you. 😉

I haven’t tested the TB3 version. Did you notice if it’s the same model number? I’d have to look into it, but I will say these TB3 enclosures are more than capable.

i just buy TB3 version, all is good i use WD SN750 running same speed with usb4 version

Yes. USB4 is compatible with Thunderbolt throughput and also falls back to USB-C Gen2. It’s just that the M1 TB ports are “slower” than the updated machines, but I’ve discussed this in previous videos as well that you can get that higher throughput if you’re really pushing them. But simple answer, I have two M1s and these work great.

So do you have access to the Acasis enclosure I reviewed? I have two of those now. I also have two SN750’s that perform well in those enclosures.

@@KevinRossRN no I haven't actually. It seems Acasis is not popular in my country (indonesia) which is why its a not easy to get your hands on it. and its way more expansive too. My friend offer me YOTTAMASTER. But he told me the results will be still around 2GB/S and not more, even if I put the SN850 gen 4 in it which suppose to has 6Gb/s read speed as in their adv claim. What do you think?

Well you’re going to be limited by the bandwidth of the enclosure and Thunderbolt. Now if you had that SSD installed internally, then you’d get closer to advertised speeds, but yeah, the boards in those enclosures only support just over 3,000 (give or take) and the bottleneck is the connection.

Yeah, the Samsung 980 Pro is a fav for f mine, but one of the questions I have is do you know how fast your particular workflow is reading/writing to the drive? I used an example in one of my recent videos how the synthetic benchmark speed often times doesn’t translate to real world use. Also, I’ve been meaning to follow up again with a 3.2 (Gen 2) drive test to show that with these Macs and really the protocol that occurs through Thunderbolt that the drives rated for 2,000MBs will just default to under 1,000MBs. Anyway, just thought I’d inquire as to the details of your particular read and write if you happen to have that info.

@@KevinRossRN Thank you so much for the reply! At the moment I'm seeing read & write speeds of about 950mbp/s using my Samsung Extreme Pro (2000mb/s USB 3.2 version) testing with Black Magic, which seems consistent with both your findings and my own. I'm not sure how to tell whether I'm getting those speeds in real situations using logic or not. What I do know is that the HD meter is usually pegged when my performance issues appear. If you've got any tips on how to test my real world read/write speeds while recording multiple tracks in logic I'll happily run some tests. If a TB4/USB4 (?) enclosure with a Samsung 980 pro will double my track count before I run into performance limitations, it's a no-brainer, but as you say, context matters, and I'd hate to spend even more money only to find my performance issues running logic persist. Thanks so much again for your reply! very kindly, Michael

So I personally use a number of apps to get this information, and iStat Menus is one of them where I can see what’s happening with the SSD as far as that read and write while a program is in use. This is where I’ve seen the CPU or GPU being maxed out (you can also see this with iStat) and therefore the read and write wasn’t even close to the max speed of the actual drive because those were the bottlenecks vs. the SSD being the issue.

​@@KevinRossRN Gotcha! will try iStat Menus and test while running a logic session to see if I can provide more info. I would really love anything even approaching the 2-3k read/write speeds I'm seeing in the (synthetic) benchmarks.. but I would also love to avoid spending hundreds on something that will not outperform my current setup. I've clearly reached my limitations with the samsung extreme pro v2 drive (rated for 2k mb/s, but performing at less than half of that).

It’s the SSD. You can usually find that on the manufacturer’s specs listed on their site, depending on which one you’re interested in.

Yeah, I have several videos in this subject. But yes, if you need to go beyond that then these Thunderbolt or USB 4.0 enclosures are the way to go. I still use them all the time.

Thanks! So I was testing a Gen 4 vs a Gen 3 and there’s no real difference. It’s a limitation of the board in these enclosures (for one), and there is some limitation with these thunderbolt controllers. I did however release a video today discussing the “benefit” of using a thunderbolt hub, which can speed things up a bit with real world renders/exports.

@@KevinRossRN yeah saw it , it explains a lot 🤔

Coolio. Glad it helped and also thankful to have you around here. 🤙

I’m still using the Acasis enclosures with great success. They’ve been great. No failure so far and I’ve had them for a while now.

Really appreciate you weighing in here. Would be great if this could be remedied in an update. My thoughts were along the lines of that trace distance of the retimer and controller to be able to adequately repeat that signal from the port, and especially being concerning with such a low watt chip. But again, helpful. Appreciate you dropping by with that bit of knowledge.

@@KevinRossRN Yea I think it's growing pains Apple depends heavily on the BSD community for the the vast majority of their toolkit. But BSD developers didn't have any ARM based machines with Thunderbolt to work and test with so Apple most likely took the BSD PCI implementation and patched it to work as USB4 because the only real developer based ARM machines are single board computers and the Pinebook Pro and none of them have Thunderbolt controllers. But I truly expected the M1 to be really bad but it's been actually really nice. I enjoy your content and keep up the the great work.

Thanks friend. Appreciate having you here. And yes, as I had mentioned that even though we’re not getting that advertised throughput , the machine has really been awesome. Such a time saver on my end.

My favorite right now is the Samsung 980PRO and the Acasis. Those might be linked up below. But in one of my recent videos I do show that external SSDs (for video editing) are still on par with the internal.

@@KevinRossRN in your later video your using Silicon Power vs Samsung 980. The Samsung is ~ $80 more for 2TB. Is it worth it.

Only for very large file transfers, and actually performing this task often. Otherwise, save some cash and think about the Silicon Power or the Samsung 970 EVO Plus.

@@KevinRossRN Great service! Sorry to bother you but now that I’ve found your channel and subscribed I’m going through all your videos on DIY SSD. Thank you so much. Take care.

Totally glad to have you here. Happy to help wherever I can.

Welcome to the community. As far as the disclaimer, it’s kind of funny because they ended up embedding this video on their website…funny. I believe what was happening was that there was a combo of updates to the SN750 and 850 SSDs (due to the manufacturing delays), and some were experiencing these slow downs, but I didn’t experience these directly. Here is a video I did not that long ago that might help: Samsung 980 PRO vs WD Black SN850 External SSD | M1 Mac and M1 Max ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-kNUQOIt2oqQ.html

@@KevinRossRN Thanks for getting back. It seems your numbers on the video you recommended to me actually backs up Acasis' claim about slower write speeds with the WD850, certainly vs the 980Pro. I was confused as the WD850 is gen4 and the WD's claimed speeds appear to be in line with the 980pro. Is your sense that the WD850 is throttling because of the thermals perhaps?

Nah, I don’t think it’s intentionally throttling. I think it’s the communication protocol that’s having issues with some of these Gen 4 SSDs. It’s as if they’re trying to be backward compatible, which they would be, but they’re just not taking full advantage of even the Gen 3 speeds.

Welcome friend! Funny enough, I am running tests right now as I type this. I’m comparing the 970 EVO Plus to the SN750 in this enclosure on multiple devices. Which file format are you using and which benchmark software?

@@KevinRossRN I tried synthetic benchmarks Black Magic and Amorphous Disk Mark and the SN750 formatted as APFS and ExFAT but with pretty much the same results. Write tops at 1160MB/s and Reads at 2600+ MB/s. I noticed also that the acasis gets pretty hot to the touch and actually stays hot even when idle (DriveDX showing 60 degrees Celsius during benchmarking). It may be coincidence but it seems like the write speeds I get from the enclosure are pretty much identical to the write speeds I get from the internal 256GB ssd. Update . Just tested with a 10GB test file copying over from the internal ssd to the external enclosure. It was finished in ~8.8s so its around 1.135 GB/s , pretty much the same number I'm getting from the benchmarks.

So I need to run a few more final tests, but on my MBP (2017) running Catalina I’m getting about 2795 read and 2572 Write on APFS. I’m going to be performing a large folder transfer of about 260GBs. I’ll report back.

Great question. So I assume you’re talking about the TB3 version, which I did look at and would be interested to test. Where I think this dock could really shine for the M1 is that it isn’t bus powered, but rather has that power brick that can help push that throughput, and this is based off of some research that other creators have gotten better throughput if they run their SSDs through a powered dock.

@@KevinRossRN Correct! I'm referring to the TB3 version. Maybe put an SN750 and then a PCIe 4.0 drive in there to test out the response. Curious about that throughput as well. :)

I’ll see if I can get my hands on one. The options are ever growing😀

It’s been known to happen. That’s why I started a second channel 😉

Maybe someone can chime in here that’s tried it. The Acasis is compatible with the FireCuda and they (Acasis) confirmed this themselves. I can’t speak to a non-enclosure setup and if you’re doing anything intensive then I’d be concerned with thermals because of lack of thermal management, but maybe your workflow isn’t that intense.

Ha! I can certainly get a million views, but I’ve also got to enjoy what it is that I’m doing and feel good about it. And I certainly feel good about this comment. Glad to help, and have you here. 🤙👊

@@KevinRossRN Certainly ! thanks for all the good hints and workup for us ! Cheers ( btw i decided to buy either a sandisk extreme pro or samsung t7 by your advice , i think ill go for samsung t7 since it has better service support here)

Nice! Well feel free to let me know what you think of it.

The SanDisk would be cheaper in the short term. The long term advantage and savings with the enclosure and NVMe SSD is that you can use that SSD in another enclosure, or a computer that can accept the NVMe. It could also be placed in a NAS setup that accepts NVMe SSDs. So, customizable, cheaper in the long run, but performance in this use case is pretty similar.

I actually did a video on using a Thunderbolt dock to help increase the saturation in that throughput that can occur, however it really only helps if you are truly able to push these enclosures to the fullest. Basically, I was able to get these to perform faster, but only when I was performing multiple tasks simultaneously.

What file format are you using? And have you tested it on another device?

So I did a video here on those Thunderbolt 4 docks: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-Cl-W029t4LQ.html If you can saturate the ports then you'll get better (overall) performance. I've also been notified that the most recent SN750 SSDs seem to get throttled because the cache gets filled quickly. I have two of these SSDs and haven't experienced this issue, but others have. I'm looking into it, and I can't say with certainty that they may have had some manufacturing changes on these newer SN750s. As far as the HDMI trick, I meant to add this to my video as I found this out later in my testing. Have you tried to just plug in an HDMI cable right from the mini? It should actually get you some additional performance.

@@KevinRossRN thanks Kevin.. at the moment the HDMI is connected into the Mini where I think coming out the dock may perform better... as for saturating the ports.. sorry I am not that computer techie.. what does that mean?

So in my latest video I mention that these ports need to be pushed because these chips are under powering the TB controllers. The dock helps push that a little, but in my testing I found that the more I had connected and actively doing something the faster the throughput….basically you should get a much higher write speed.

Really appreciate you popping over for this one. Much appreciate your time and comment. I do love the Fledging setup and have also been in communication with them, which I appreciate so much from a brand. And yeah, even though you (one) may get similar results from a pre-built 3.2 device, these are still great because of the customization...especially since we can't customize/update anything on these M1 machines. Many thanks, and keep rocking those faces.

@@KevinRossRN Yessir! I've always built my own because I'm just a speed freak (and a little OCD). That part is predictable. What wasn't was all the testing I did with all those DIY enclosures/NMVe and what was concluded from it. I also was never happy with pre-built NVMe configurations that are offered. Don't like to be pigeon-holed. I tinker way too much. Keep rockin' those beanies!

You got it. And, I need a beanie sponsor 🤣

What other SSDs have you tried? WD SN850, or Sabrent Rocket 4 by any chance? Also, what speeds were you getting with the SP Gen 4 NVMe?

The SN750, which I did post those read/write values in the video, but the only reason I tested the Gen 4 silicon power was because I got an awesome deal on it, I wanted to test whether it made a difference in the enclosure (it didn’t), and I ended up using it in a PC build that I’ll be sharing on the channel soon, but the SP in the PC is reading at 4977 and writing at 4200.

🤫 don’t tell anyone. We’ll keep this between us😉

Naw... Aerosmith!

Rock on 🤘

Hi Julian, welcome to the community and congrats on your M1. So I recently did a test with a TB4 hub from Anker and there is definitely something happening with the communication protocol with these (M1) thunderbolt3/USB4 ports. These controllers on the SoC can definitely be pushed really hard, however this seems to be the issue (for most). The harder you push them, generally speaking, the better they seem to perform. The Gen 2 protocol seems to take a massive hit in synthetic benchmarks, however the write speed can also be tough on the TB3 NVMe enclosures as well, except when they get pushed a bit harder and then you'll notice that bump in performance. Here's what I'll say and I'm continuing to test. The M1 is really an amazing piece of kit for a lot of tasks, especially for editing. However, there does seem to be or has been some engineering oversight on the controllers and retimers when it comes to the full capability of what Thunderbolt 3 should actually perform at, but in this case isn't reaching out of the box without adding a hub or tasks that really push that throughput. Now when it comes to FCPX tasks the program is going to be limited by that read and write and it's not going to fully saturate that bandwidth. I've gotten peak read and write speeds when rendering of anywhere from 1,000-1,300 MB/s on the internal drive....I'd have to double check my notes, but that is also not sustained because the CPU and GPU are going through that timeline and settle somewhere in the neighborhood of let's say 400-500MB/s and sometimes lower. So in my testing I might shave a minute or a couple when using the internal. Sometimes it's more, but not a massive jump. However the interesting thing is that the more I pushed these drives when connecting the Anker hub, I was however able to get almost exact render/export times on the TB4 enclosure as I was on the internal because I was trying to saturate that throughput as much as possible with other tasks. But again, I'm still trying to find a USB-C Gen 2 external drive that actually gets close to the advertised performance, but haven't found that yet. The SanDisk extreme is a drive I use quite often, but synthetic benchmarks are much lower, however in real world it doesn't really impact my work. I would say however that if you have the SSD and if it's within your budget, I have continued to have great luck with these thunderbolt enclosures, but I can only speak about the Acasis and the Fledging. Every now and again the quality control on a cable or an enclosure may have missed something because I had someone reach out about a slower Fledging enclosure, but once they got the replacement sent to them then their results were the same as mine. A very long winded reply, and I wish I had THE answer here, but until we get these next chips I don't really have a similar (ARM) device to test this against. The early bluetooth issues certainly got more attention because it impacted more folks day-to-day whereas us "tech" enthusiasts seem to be focused on these controllers on the chip, which hey, I'm right there with you. WIth two dedicated lanes we should be flying on these machines. I was really hoping for a software update that could be address this communication protocol, but there is definitely something within the hardware that isn't allowing this throughput unless you completely stress the machine. Not sure if this is helpful, but wanted to just drop a bit of what I know so far.

@@KevinRossRN thank you so much for the detailed answer😊 although i am still not sure what I am going to get now, i surely am closer😁. Thanks alot

Happy to help where I can. Best of luck on what you decide and that it works for you. 😉🤙🤘

So on TB3 ports you’ll find that the rated 20Gbps enclosures will only be supported at Gen 2 speeds of 10Gbps, which is just how they fall back on the communication protocol. And USB4 is supposed to give you theoretical speeds of 40Gbps on TB3 ports, but the actual throughput will be dependent on the machine’s TB controller. So in the M1 Macs we’re finding that these are a bit “slower” than advertised due to the low power chip, although they can be pushed much harder and ultimately faster. Now, depending on the board/controller in a PC, USB4 should still give you TB3 speeds, but I have updated videos on this subject and what we’ve been finding is that it really depends on the enclosure and the particular computer it’s connected to. Are you mostly looking for something for large file transfers?

I just got the replacement enclosure from Acasis, but need to bust it open and run some tests, which I'll update the community here on those results. I'm also waiting to see if I can pick up one of those updated Docks from Caldigit so I can also test the throughput as it seems like these M1's are just hungry for more power to clean up that signal with the retimer. I've poured over forums to find similar results when it comes to these enclosures, but some have just stopped at running a black magic test and calling it good from there, which you know my thoughts on that 😆 Per some of these forums many are reporting this to Apple, which some have said they've gotten responses indicating the engineers have been notified, which I'm unsure of how a software/firmware update will help if it's the issue of the retimer's distance to the controller (on the chip). Hard to say. But, now that I have two USB4 enclosures, let me at least run those tests to see if something has changed with this latest batch. I know other manufacturers have typically defaulted to trying another SSD in their enclosures, specifically the SN750 since it seems to be the one that's widely tested, which I picked that up to confirm and compare to my Silicon Power Gen 4. I'll circle back around.

@@KevinRossRN Hi Kevin, thank you. When you connect the Acasis USB4 NVMe enclosure to your M1 MacMini (MacOS 11.2.2), in the Mac Sys Information->Thunderbolt/USB4 tab, do you see "Device Name: Tahini Creek" and "Mode: Thunderbolt 3" ? I expected "Mode:" should be "USB4" for the Acasis USB4.0 NVMe model TBU401. When I connect a OWC USB4 hub to the M1 Mac TB/USB4 port it says "Mode: USB4". I think this "Mode:" info was added in MacOS 11.2.2. FYI, I did a couple more tests with S/W RAID0 across both TB/USB4 ports and just reached 4600Mbs reads, 3200MBs writes. But writes are not constant.. Soon I'll reformat the RAID drives chunk size down to 128K - ie: the TB packet size. But promising... Thx Jeff

I am getting Tahini Creek, which is also aligned with Titan Ridge, and TB3. Let me update the OS. I’ll circle back around, and do appreciate this data by the way.

@@KevinRossRN A little update: I tried two ADATA SX8200Pro 2TB each in a ACASIS enclosure - one connected to the M1 left TB port and one to the M1 right port. I used macOS software RAID 0 with 16K slice and got interesting performance most of the time: AmorphousDiskMark (at 4GiB) showed SEQ1M QD8 R/W: 5,750 / 3,200 MB/s while BMD (at 4GB) showed R/W: 3,245 / 3,010 MB/s. This speed should be enough for high end video editing. I also found the M1 has some smarts going on for auto adjusting thunderbolt/usb performance - the harder I pushed the Thunderbolt bus the better the throughput (ie inserted a Hub and added a 5K TB3 monitor) - I expect this is due to the new M1 storage controller. The down side is you sometimes see very poor performance until you exercise the Thunderbolt bus. The same 2 x ADATA NVMe's in a macOS Intel PCIe x8 slot (x4 switched for each NVMe) runs at a pretty constant R/W: 7,000 / 5,200 MB/s. I tried to isolate if certain NVMe or NVMe->Thunderbolt controller are better or worse, and this became a rabbits hole since the M1 definitely plays with each one differently - sorta learning how to max each storage device out over time, than cuts back for power saving or bandwidth control ? Even testing the WD Black SN750, famed to be the M1's most happiest NVMe, I also saw R/W: 2,800/2,500 MB/s somedays and a lousy R/W 2,100 / 800 MB/s other days. I am also battling a problem with 2 or more NVMe's connected via Thunderbolt to the M1 crashing a lot at boot up or connecting the devices. I tried different NVMe's and enclosures - very noticeable with 3 or more NVMe's - all crash the M1. Apple is kindly involved with this one. Hope this helps all.

This is very helpful friend. I’ve got an Anker TB4 hub that I’ve been testing to run similar benchmarks with and also pushing all of the ports simultaneously, and did notice a bump in performance when that occurred, but I suppose the question is how often that’s going to happen for most users 🤷‍♂️ This is a great breakdown though and I’ll continue to hang out in that rabbit hole a bit longer with my testing as well. 🤙🤘

Yeah, I’ve been meaning to review that one. The USB4 from Acasis is pretty awesome, and I’ll be curious to see what you find with the hub comes in. My update on the M1 Max should be coming out soon in regards to Thunderbolt speeds with these drives. Let me know what you find when you retest.

@@KevinRossRN I also tried that Acasis enclosure. I loved the build quality. Felt like a total beast. But on my machine the treebleet outperformed. I’ll check back in a couple weeks after the hub arrives.

Also thanks a ton for the quality in depth testing and review and truly no filler or fluff in the video. RU-vid needs more reviewers like yourself!

Yeah, I think it comes down to finding the one that works best with your setup because as you see they’re all a little different. Looking forward to the update. Whenever you get a chance.

@@KevinRossRN so I received my hub. Instantly noticed write performance is awful when plugged into the hub. However. Having the hub plugged in and the drive direct to computer i am seeing those higher write speeds and seeing them stay consistently high. Where as before they would drop down a tad and go back up. But direct to hub. I saw a drastic decrease in write speed with this drive down to 600… gross.

And thank you for your time. I’m still using both of my Acasis enclosures, if that helps 🤷‍♂️

@@KevinRossRN Thanks for the reply. I was about to order one but ACASIS aren't compatible with SK Hynix SSD's apparently. From what I understand it will kernel panic a Mac if used in that combination. Ouch, time to do more research. Interestingly while I was doing research I found out that the ANYOYO uses the same controller board as the ACASIS enclosure, just with a different aluminium case. When I went onto the ANYOYO web page I clicked on their contact us link and it was for an ACASIS email address.

Yeah a lot of these manufacturers are using the same components and marketing/branding as something different. I looked into it a couple years back when I was going to launch my own. Do you already have that SSD, or just looking at it due to access and price?

@@KevinRossRN I ordered a 2TB P41 Platinum last week when it came up for a deal on Amazon Australia too good to pass up ($178 AUD or $120 USD). It's still in the ordered state and not shipped though. Somewhat of a concern as there have been reports of some people receiving an empty sealed box. Why is it that most of the issues we have with IT equipment these days is the ordering and delivery rather than operational or warranty issues !? All part of the roulette wheel of online ordering I suppose. 😉

Sorry to spam messages 😌 but I thought it might be of use for someone to know that I emailed SK Hynix directly and they said they do not recommend using their SSD's in an external enclosure regardless of the brand. Their recommendation is for internal computer use only. As a result I cancelled my Amazon order for one and will probably buy a Samsung 980 Pro instead.

Congrats! What are you mostly using it for?

@@KevinRossRN video Editing, back up, and would like to use the computer SSD for the APP/Software vs the external SSD to store and work on my files (photoshop, Lightroom, ADOBE PREMIER Pro AND Final Cut Pro)

So I did an updated video last month on SSDs. I think you might be surprised by what I found…or maybe not. This Acasis drive I reviewed is still going strong in my rotation. I typically use the SN750 in it with good results, but someone from the community got a deal on the 980 Pro: amzn.to/31j4xHa And it’s working quite well. I also love the Sabrent enclosure: amzn.to/3ob1XMs But it’s not Thunderbolt, but I’ve been working off that as well with no issues.

@@KevinRossRN thanks Again, really appreciate your generosity!

No sweat. Glad to help.

I think the only thing that others have stated is that it randomly disconnects, but I haven’t tested this one. I’ve been meaning to take this one for a spin though.

@@KevinRossRN Yeah it doesn't seem to be very stable now. Probably I'll wait for the next gen to see how it goes, because the enclosure + an ssd are still a bit expensive to me

It might be best to wait or choose another. But if you can wait, then I believe we’ll start seeing more stability here.

@@KevinRossRN Yes I'm not in a rush right now, just got a great deal for an used Mac mini so I'm looking into some storage options. Btw the new Apple event is happening in 3 days and I'm very excited to see if the new Mac mini rumors are true!

That would be awesome, right? I love the M1 mini.

Yeah, sometimes I do throw those visuals in. Sometimes after days or weeks of testing, the project gets a little over the top in the edit and those visuals take a bit of work to incorporate sometimes and but the end I get pretty shredded. However, I will be putting more of those visuals into upcoming videos and certainly appreciate the feedback 👍

@@KevinRossRN no worries! i know its more work so it's whatever you'd like. A little short voice summary i think would be an alternative too and could be easier too

All good friend. The feedback is helpful. I’ll likely be incorporating a spreadsheet into my notes that I can hopefully bring in those visuals much easier. We’ll get there 😉

If you're looking to save money in the short-term, then an external SSD that's already built (like a SanDisk Extreme, Samsung T5/T7) are a great option. I've used them for years. The advantage of these enclosures is that the NVMe SSD that goes in them can be customized. So for instance I tested the WD SN750 which is a gen 3 PCIe SSD that is supposed to be "slower" than the Silicon Power PCIe Gen 4 SSD, which theoretically is faster, but in this case doesn't make a difference. But as a long-term example. I've purchased that WD SN750 and Silicon Power for these enclosures. And, when an updated enclosure comes out, or I'm actually able to take full advantage of the advertised speeds as these chipsets get updated, then I've always got the SSD (separate) that I can move to another enclosure. Another benefit with these SSDs and enclosures is that let's say I wanted to take that SSD and put it in a PC or NAS/Backup drive to be used there. So, the long-term return on the separate NVMe SSD can be better, but again, the prices of these pre-built solutions are great, they're usually really reliable, and will still be fine in most workflows.

Yeah, this all has to be measured for each workflow and user. As the tests show, the differences aren’t really there for an export. And FCP has a really good option to work off of a NAS with spinning drives due to Symlinks. Scrubbing through and cutting definitely needs a bit more responsiveness from the drive, but again, that’s why I don’t just work from Synthetic benchmarks only because it comes down to how it’s working for me at the end of the day. As far as getting back time, I’ve been getting back way more time in performance from the actual machine.

That’s awesome you’re getting decent performance on that Sabrent. It was funny because originally those SSDs were the worst performers, but not sure if there was an update that helped that communication protocol. So as far as the heat, is that with the Sabrent in there? Also, when plugging in does the green light not illuminate, or is it lighting up but it’s just not mounting?

@@KevinRossRN Yes, enclosure is hot with the Sabrent Rocket 4.0. The green light stays solid while the drive is connected and it starts blinking when it mounts. Unfortunately this doesn’t always happen, it’s a bummer that it works so intermittently, it’s very fast when it works. I don’t really know which device to blame, I am on my second copy of the Sabrent NVME (still hot and not mounting) and the second copy of Acasis is coming on Monday…

Interesting. I think it’s definitely unfortunate that these sometimes feel as if we’re beta testing them. I’ll be curious to see what you find when the enclosure comes on Monday. I think some of these drives don’t do well in these small enclosures because they run hot as it is, but in a PC you’ve got plenty of options for cooling. I think this is also one of the reasons we won’t be seeing full support for these Gen 4 SSDs in enclosures because the thermal management would be a nightmare. Some just run really hot even when they’re not even under full load. It would likely take an enclosure with a built in heat sink.

@@KevinRossRN I see, maybe I am better off putting a slower NVMe in this Acasis enclosure? The Sabrent Rocket 3 was my number 2 choice by speed (I believe it’s a PCI 3 version). In my testing, the SN750 was a distant last place in terms of the write speed, the Samsung Evo Plus came in 3rd with write speed in 1300s…

Have you considered the Samsung 980? I need to find the comment thread , but someone on here said he got something like 2700/2800 on both read and write. I was pretty surprised to hear it.

It depends on what video editing software you’re using, but it depends. As an example a 1000MB/s has performed the same as my NVMe drives that are rated at 3000+MB/s because the editing program can only read and write the data so fast. You may notice a negligible difference in the 500MB/s as I do have these drives as well. On the M1 Mac I notice it more because the throughput is limited on the TB3 ports, so that 500MB/s is hanging in the mid 300s. For the price (based on what I’ve seen) many of the 1000MB/s drives are within a few dollars 20-30USD give or take. I consistently get about the same performance with editing on these 1000MBs drives as I do on the faster ones. I used the Samsung T5 for years without issue. Also, in my latest video, I was able to edit on an external spinning HDD without issue, but I wouldn’t necessarily recommend it for sustained work.

@@KevinRossRN Thanks for the super thorough and rapid response. You're the man! I am new to editing and I am noticing that when I have more than a few tracks of video in my timeline in Davinci Resolve that it gets really glitchy and freezes quite a lot. I wasn't sure if that was because its in 4k 10bit 4:2:2, h.265 or if it had something to do with the media files being stored on my 500gb samsung t3. While its probably because of the codec, I was wondering if upgrading my SSD to one of these NVME's would make a significant difference? Thanks again Kevin!

Happy to help. Are you editing in Davinci on the M1 Mac?

@@KevinRossRN I am! Macbook air. 16gb RAM

Gotcha. I have noticed that in resolve I am getting similar results as you. Interesting too that I was even getting some lag and chop when trying to cut BRAW 🤷‍♂️ Possible it’s a mix of the codec and the optimization of Davinci. I hope it gets better because I do like that editor.

I still haven’t tested that one. I’ve heard some mixed reviews on this one with M1.

Exactly. They are much lower. Depending on the enclosure, possibly 2800 read and write give or take.

Right back at you 🤘

I use them both equally. Do you think you’ll ever use either one of these on anything other than the Mac?

@@KevinRossRN mostly mac mini and occasionally pc, but speed is Important as I want to run parallels virtual machine from it

Both will have plenty of speed. If you’re going with a PC as well, and you don’t have Thunderbolt on the PC, then the Acasis is the way to go because it’s USB4.

If they were in the computer and on the board, as long as the board would support that throughput (example would be PCIe Gen4), then yes. But there’s a bottleneck between the controller, the retimer, and then obviously the fact that it’s external and has to travel through that connection to the board and NVMe in the enclosure.

@@KevinRossRN So, using a SDD with a transfer speeds of up to 5000-7000mb/s (ex Sabrent Rocket 4 Plus, Samsung 980 PRO) would be pointless since using an external enclosure will limit the speeds to about 1GB/s?

In a USB-C Gen 2 enclosure it’ll limit them to about those speeds, but a USB4 or TB3 enclosure will be double that and then some, but yeah, you still won’t get the full read and write on that SSD in an enclosure.

I keep my PS and LR media on an external because as you know, that raw data just sucks up all the storage.

@@KevinRossRN yes that's true. I brought it up, because og recent apple M1 MacBooks seem to memory swap so much disk space it wears out the disk. I was thinking of getting a fast tb3 drive to compensate the usage and lower the wear and tear. Not sure it's a bright idea. What do you guys think?

I mean those SSDs in the Mac should be fine for quite some time, and PS loves that RAM, which is why I got 16GBs, so there’s less swap happening. But that being said, a TB3 enclosure, or even a 3.2 Gen 2 drive would help to have as your scratch/media source.

If you don’t fully unplug it from the device, there’s still a power draw occurring even if you don’t see it in your drive utility. It’s just something we’re seeing more often with these DIY enclosures.

@@KevinRossRN Thanks for your answer.Do you remember if the led on your Acasis case goes off after some time and if it goes into sleep mode? It is about 35-40 degrees Celcius idle. Maybe it's caused by the Sandisk Extreme Pro (identical to the WD SN750 Black). In the Sabrent case it also stays warm, but at less temperature.

Yeah, the LED typically remains on and it’s possible that the particular SSD could be drawing more, but these are definitely a little different than those pre-built solutions….based on my testing.

@@KevinRossRN It seems there is also no auto sleep mode as on some pre built external ssd like the Samsung T7. But the speed is stunning.

Yeah, I use these for the speed and expandability.

Ha! I think I can work something out on that German steel. Always need to throw something else in besides the benchmarks. 😉 Great to have you here.

Well even though the Sabrent Rocket doesn’t get the highest performance, Fledging tested it with this drive. The SN750 is a good choice. They tested this one as well, and I have as well.

@@KevinRossRN I was perhaps thinking about the Sabrent Rocket gen4 4tb version as the WD 850 gen4 doesn’t come in a 4tb version

And in the Sabrent enclosure?

@@KevinRossRN no in the Fledging enclosure. Will that still work?

That’s the one that they should have tested and I believe is posted on their site. I’ve had great luck with their customer service. They’ve been responsive.

Yes, I’ve tested other export options such as exporting into a 422, which is super quick, but does create a larger file. And then just a standard h.264 will take a little longer, and the file size more reasonable. This test was using a compressed h.264 which does take quite a bit of time, but it’s a compressed form that typically allows RU-vid to process that 4K file almost immediately (in my experience). But as far as real world use, I can say that even having a 1TB of internal, I still have several external drives as my scratch disk and project bundles to work from.

Have you tried any other synthetic benchmark tests other than Black Magic? Also, are you in APFS file format? I’m addition, the 2TB models are rated to get about 100MB/s slower than the 1TB due to the 500GB NAND dies on this model whereas the 1TB has 256GB NAND dies on it.

@@KevinRossRN Thanks for your feedback! Yes, I'm getting 877MB/s write and 2587MB/s read on the AJA test. Yes, in APFS format. I'm on Monterey. Not sure if I need to send anything back. But I'm bummed that I'm getting about half of the write speed that I should be getting.

It’s possible, and I can’t guarantee this would be the issue, but because of the supply chain issues and all the other manufacturing bottlenecks there are some claims that some of these SSD manufacturers are having to use different components based on these shortages, but I can’t be certain. I’ve discussed it heavily as well that these M1s are underpowered when it comes to these ports. They have the capability of getting to that higher throughput, but you need to put more demands on the machine. Try running a geek bench or other Benchmark test while also running the SSD Read/Write and see if that gives you a bump…just curious.

@@KevinRossRN I'm suspecting it may be the difference between the 1TB SN750 and the 2TB SN750. I've ordered another certified TB3/ USB4 cable that's shorter to see if that might help. If it doesn't, I'll try the 1TB SN750. Thanks for your help! Subscribed!

Glad to have you here and unfortunate to hear about your experience, but please feel free to keep me posted. Sending my best on those change ups.