Thunderbolt 3 M.2 NVMe SSD Enclosures

[joevt]

Do they risk disconnection (by exceeding the power limit on the TB port) if you put in a higher-performance (and thus higher-powered) NVMe drive?

I haven't heard about such a problem. I have a Trebleet bus powered enclosure. It has never disconnected. I have USB power testers but I don't think any of them will pass a 20 Gbps per line Thunderbolt signal.

Why doesn't anyone make a 2800 MB/s wall-powered version of these single-NVMe enclosures that, as a result of not relying on bus power, avoids this problem? Sabrent and OWC only make non-powered 2-lane versions limited to 1500 MB/s, and while Glyph makes a powered enclosure, it's also limited to 1500 MB/s:

I don't know. Adding a power supply makes the enclosure less portable?

 

[theorist9]

I haven't heard about such a problem. I have a Trebleet bus powered enclosure. It has never disconnected. I have USB power testers but I don't think any of them will pass a 20 Gbps per line Thunderbolt signal.

It may be the sort of thing that happens only occasionally, and only with sufficiently high-power drives, but where even the chance of this has caused Intel to refuse to give TB certification to bus-powered 4-lane NVMe enclosures (unless they certify the drive and enclosure together). For whatever reason, all of the TB3/4 single-NVME enclosures from the "established" manufacturers (the ones that have all their devices TB-certified), are two-lane only (and thus limited to 1553 MB/s).

Only their dual-NVMe enclosures are four-lane, and they are all powered. But they still reserve two lanes/SSD, and thus will only use all four lanes if you configure the two SSD's in RAID 0. But now, instead of 2750 MB/s, it's now 3000 MB/s max. It's all so mysterious...there really needs to be a white paper to clear this all up.

I don't know. Adding a power supple makes the enclosure less portable?

Ah, that's probably it!

 

[joevt]

But they still reserve two lanes/SSD, and thus will only use all four lanes if you configure the two SSD's in RAID 0.

12 Lane bridge chips might be expensive.

But now, instead of 2750 MB/s, it's now 3000 MB/s max. It's all so mysterious...there really needs to be a white paper to clear this all up.

The USB4 spec has some notes about how the the amount of buffering is balanced between bandwidth and latency.
"The amount of buffering at the PCIe Adapter is implementation specific as it balances the tradeoff between PCIe tunneling performance and PCIe link latency. It is recommended that implementations make the amount of buffers configurable."
It says the same thing for USB3 tunnelling (replace PCIe with USB3 in the above statement).
Section E has some calculations on how to choose the number of ingress buffers. It's not clear if the ingress buffers is the same as the buffering mentioned in those implementation notes.
It's complicated so I won't try to summarize it (also maybe because I don't understand it all).

 

[theorist9]

Where does it say 8 Gbps is reserved for video? I think it's more likely that 8 Gbps cannot be used for data (because the discrete Thunderbolt 3 controllers were limited to 31.5 Gbps) which has a different meaning.

It's not much of a standard if no-one is allowed to read it. For that you might try the USB4 spec which is open.

OK, I found an official source saying that 8 Gb/s is preallocated for video:

https://www.thunderbolttechnology.net/sites/default/files/Thunderbolt3_TechBrief_FINAL.pdf

"...to fill this Thunderbolt link, the silicon extracts and routes up to 4 lanes of PCI Express Gen 3(4 x 8 Gb/s) and up to two full (4 lane) links of DisplayPort out over the Thunderbolt cable and connector to the device(s) attached downstream from the host system."

Thus, since its max bandwidth is 40 Gb/s, but the parts that can carry data are limited to 32 Gb/s, the remaining 8 Gb/s is video only.

Now what about the 2750 MB/s limit for TB3/4 data devices? The doc doesn't say, but it could be that, of the 32 Gb/s available for data, TB additionally reserves 10 Gb/s for USB Gen 2 devices. That would leave 22 Gb/s = 2750 Mb/s for the TB3 device.

Finally, the document explains that TB gives priority to video over data (makes sense--a frozen display would be much more irritating than having to wait longer for data transmittal) and, from one of their graphics (screenshotted below), it appears it can allocate up to 32 Gb/s for video (less headroom), which means it uses the 8 Gb/s from the Display Port and borrows an additional 24 Gb/s from PICe. This allows it to drive two 4096 x 2160 displays @30bpp*/60 Hz, without compression (*30 bits per pixel = 10-bit color with fulll color depth, i.e., 4:4:4):

 

[joevt]

OK, I found an official source saying that 8 Gb/s is preallocated for video:

https://www.thunderbolttechnology.net/sites/default/files/Thunderbolt3_TechBrief_FINAL.pdf

That's the crappy marketing material I linked previously.

"...to fill this Thunderbolt link, the silicon extracts and routes up to 4 lanes of PCI Express Gen 3(4 x 8 Gb/s) and up to two full (4 lane) links of DisplayPort out over the Thunderbolt cable and connector to the device(s) attached downstream from the host system."

Yup.

Thus, since its max bandwidth is 40 Gb/s, but the parts that can carry data are limited to 32 Gb/s, the remaining 8 Gb/s is video only.

Just because your car can only go 32 mph on a 40 mph highway doesn't mean the other 8 mph is reserved for someone else's car. It just means your car is not as fast as it could be.

All we know is, for a discrete Thunderbolt controller, the upstream for the controller as a host is PCIe gen 3 x4, and the downstream for the controller as a peripheral is also PCIe gen 3 x4. A single PCIe device can do up to 31.5 Gbps (after 128b/130b encoding) or ≈28 Gbps (after PCIe protocol).
That tech brief (too brief) was written before integrated Thunderbolt controllers. The upstream of a integrated host Thunderbolt controller is not PCIe gen 3 x4. For integrated Thunderbolt controllers with two ports (Ice Lake and Tiger Lake) it is known that the upstream of the controller is greater then PCIe gen 3 x4. So is there a way for them to get more than PCIe gen 3 x4 from a single port if there's multiple devices connected to the port? Probably not but but it would technically be possible since the Thunderbolt connection is 40 Gbps.

Now what about the 2750 MB/s limit for TB3/4 data devices? The doc doesn't say, but it could be that, of the 32 Gb/s available for data, TB additionally reserves 10 Gb/s for USB Gen 2 devices.

If anything is reserved for USB, it is reserved as tunnelled PCIe data since the USB controllers are PCIe XHCI controllers built into the Thunderbolt controller chip.

Thunderbolt doesn't reserve 10 Gbps for USB. Apple can use up to 38.9 Gbps for the Apple Pro Display XDR. You can connect two displays that use up to 34.56 Gbps total. When I tried, the max display bandwidth I could extract using two HBR2 connections over a single Thunderbolt cable was 31 Gbps (4096 x 2304 x 68.595 Hz = 681.76 MHz pixel clock x 24 bpp = 15.5 Gbps).

That would leave 22 Gb/s = 2750 Mb/s for the TB3 device.

Except we've seen 25 Gbps benchmarks.

Finally, the document explains that TB gives priority to video over data (makes sense--a frozen display would be much more irritating than having to wait longer for data transmittal) and, from one of their graphics (screenshotted below), it appears it can allocate up to 32 Gb/s for video (less headroom), which means it uses the 8 Gb/s from the Display Port and borrows an additional 24 Gb/s from PICe. This allows it to drive two 4096 x 2160 displays @30bpp*/60 Hz, without compression (*30 bits per pixel = 10-bit color with fulll color depth, i.e., 4:4:4):

Yes, 32 Gbps is about the limit of two HBR2 connections (34.56 Gbps total).
But Apple can do two HBR3 connections up to 38.9 Gbps.

For my dual HBR2 tests, instead of using 30bpp, I used 24bpp so I wouldn't have to worry about macOS switching from 10bpc to 8bpc. Then I just increased the refresh rate from 60Hz until it would stop allowing the resolution (68.595 MHz).

Anyway this talk of reserved or allocated bandwidth is overly complicated and not very descriptive of what we're seeing. Isn't it simpler to say that DisplayPort uses what is required for the current display mode of each connected display, then whatever remains up to ≈22 - 25 Gbps (less than 26 Gbps so far) can be used for PCIe?

For Thunderbolt 4 and USB4, new tests need to be done and the tests should add cases that include USB tunnelling. For example, what's the max total bandwidth that can be achieved using USB tunnelling and PCIe tunnelling as measured by a benchmark like ATTO Disk Benchmark.app that can measure multiple devices at once? Maybe this has more of a chance of exceeding 26 Gbps since the tunnelled USB is totally separate from tunnelled PCIe so there's no squeezing (however unnecessarily) of multiple Thunderbolt devices into a PCIe gen 3 x4 bottleneck.

 

[mackiemesser2]

Too much variables, too much speculation (though based on rigorous testing, this I have to acknowledge). Isn't it time to ask Intel for the definitive specs? And I'm talking TB3 and TB4 *not* USB4. This is simply not our job and afaik at least Intel's documentation is good. So they should have all the answers.

 

[theorist9]

In this video (

), the poster got this speed out of a 970 Evo in an ACASIS enclosure connected to a Windows machine:

To fix this, he had to go into Properties and enable write caching, changing this (which was the default setting):

To this:

Once he did, he got this:

Do we need to be concerned about this on a Mac? I don't know how to access those settings on a Mac.

 

[theorist9]

Blackmagic DST 5GB Stress on Acasis TB3/USB4 enclosure with a 2TB Samsung 980 Pro NVME. Both direct to 2019 27" iMac and 14" MBP M1 Pro, and with Acasis attached to OWC Thunderbolt 4 Hub.

27” iMac TB3 port direct to Acasis enclosure:
2521 W / 2646 R

27” iMac TB3 port attached to Hub and Acasis enclosure:
2370 W / 2637 R

14" MBP attached Acasis enclosure via TB4 port:
2795 W / 2710 R

14" MBP with HUB and Acasis enclosure:
2739 W / 2707 R

14" MBP attached Acasis enclosure via TB4 port, BUT with USC-C (non-TB/USB4) cable:
973 W / 888 R (someone asked a few pages back about USB4/TB3 enclosure speeds with USB-C cable)

27” iMac 512GB internal blade:
1857 W / 2155 R

14” MPB 1TB internal blade:
5869 W / 5333 R

I'm curious what you'd get with Amorphous Disk Mark (free on the App store) for the random read/writes in comparing a direct TB connection, the OWC hub, and the 5 Gbps USB A port on your 2019 iMac.

The reason I ask is because, even with the direct TB connection, three of the figures will likely be ~<300 MB/s (see screenshot below of a 980 Pro in an ACASIS connected to an M1 Max). [N.B.: He's using an older version of Amorphous (3.1 vs. the current 4.0)—not sure how that affects things.]

That raises an interesting question: Will those, in particular, be just as fast with the 5 Gbps USB port, because they can't saturate its ~500 MB/s cap? Or will the USB-A port cut the speed of all the reads/writes down vs. the TB port, even those well under the USB-A's speed cap?

Source:

 

[whodiini]

Looks like ACASIS just released a new model several days ago

TBU405

Acasis 40Gbps Tool-free M.2 NVMe Thunderbolt 3/4 USB 4 SSD Enclosure

TBU405 VS TBU405ProM1TBU405 is base version (Tool-free Installation), while TBU405ProM1 (Tool-free Installation + Built-in Silent Fan) is an upgraded version of TBU405. The main difference is that TBU405ProM1 is additionally equipped with a built-in silent cooling fan. According to tests...

www.acasis.com

I got this, mechanical design is mediocre. The main problem is that the cover which is held by 4 screws, is recessed into the enclosure. It means that when you put the thermal pad on, there is no easy way to remove the cover since it is stuck to the thermal pad which in turn is stuck to the NVMe. I put the cover on and realized that the screws didnt line up. The cover can be inserted 2 ways, but only one way the screws are lined up. So I had to shake the cover loose until hte thermal pad released to rotate the cover 180 degrees. Poorly thought out mechanical design. There should be only 1 way to insert the cover, and have a slot for a screwdriver to lift the cover off. Mechanically, the envoy express is the best thought out design, but as people have noted, only 2 PCIe lanes. I havent come across any NVMe thunderbolt enclosure I would recommend yet that runs 4 PCIe 3 lanes.

I read the speculation that 2 PCIe lanes was because of potential power limitations. The TB3 spec is 5V 3A or 15 watts. I looked at the power consumption of NVMe. NVme can support up to 25W in power state 0. 15W max is power state 3. 18W is power state 1 or 2. Looking at firecuda 520, it uses 11W. Samsung 980 pro uses 8.5W, Those are within the 15W limit, even including the 2W used by the thunderbolt controller. Practically speaking, consumer NVMe are within the power spec of TB3. But maybe no knowing what NVMe is being used, intel wont certify unless it is half power? Hard to say. Certainly, the enclosure will get very hot - dissipating 15W in a small passive enclosure is not a great idea.

 

[theorist9]

I got this, mechanical design is mediocre. The main problem is that the cover which is held by 4 screws, is recessed into the enclosure. It means that when you put the thermal pad on, there is no easy way to remove the cover since it is stuck to the thermal pad which in turn is stuck to the NVMe. I put the cover on and realized that the screws didnt line up. The cover can be inserted 2 ways, but only one way the screws are lined up. So I had to shake the cover loose until hte thermal pad released. Poorly thought out mechanical design. There should be only 1 way to insert the cover, and have a slot for a screwdriver to lift the cover off.

I was thinking of that one as well, since it's bus-powered, and is a combo enclosure and hub. When you get a chance, could you run Amorphous on it and post your results? Also curious if you've tried driving a monitor off of it.

 

[whodiini]

Sorry, the one I have is just the enclosure.

Amorphous using a P31 Gold 2 TB:
read write
seq1M QD8 3170 2407
seq1M QD1 2022 2019
RND4k QD64 1130 267
RND4kQD1 52 39

 

[theorist9]

Sorry, the one I have is just the enclosure.

Amorphous using a P31 Gold 2 TB:
read write
seq1M QD8 3170 2407
seq1M QD1 2022 2019
RND4k QD64 1130 267
RND4kQD1 52 39

Ah, no problem. I mistakenly thought you were referring to the TB42 (thats the enclosure+hub), since I didn't notice that on their site until a few days ago.

But maybe no knowing what NVMe is being used, intel wont certify unless it is half power?

Yeah, that's what I've been told. Of course, one workaround is to add wall power, which the TB42 does (though it's not Intel-certified). I emailed ACASIS, and they said specifically that the wall power is indeed used to power the SSD, so it's not driven by the bus.

I havent come across any NVMe thunderbolt enclosure I would recommend yet that runs 4 PCIe 3 lanes.

The two Thunderbolt-certified ways to use all four lanes are (note that these are all wall-powered):

1) Get a TB-certified dual NVMe enclosure and run it in RAID 0. That would give you access to all four PCIe lanes. Options include:
a) Sabrent Dual-NVMe enclosure ($150)—though it's small, so you've still got the heat issue.
b) Sonnet Echo Dual NVMe dock ($350). The Sonnet has a fan, so there's a potential noise issue there, but it should keep the cards from overheating. And you get a full TB dock.
c) SIIG dual NVMe dock (~$350). It's fanless, so would be silent, but it probably has more thermal throttling than the Sonnet. Like the Sonnet, you do get a full TB dock.

2) If you want to avoid RAID 0, you could get something like a Sonnet Echo Express SEL single-PCIe expansion box ($350) and install either a single-NVME or dual-NVME PCIe card (e.g., Sonnet M.2 2x4 Low-profile PCIe Card, $250). That would allow you to run one or two NVMe's SSD's, respectively, where each has access to all four lanes, without needing RAID 0. But that's much pricier—with the dual-card system, you're already at $600, and you don't get a full dock (but it at least has a TB output, so you don't lose a TB port when you use it).

 

[theorist9]

Here's a nice review that compares the 2-lane OWC Envoy enclosure and the 4-lane ORICO enclosure using an SN850:

He found that, for operations below the 2-lane OWC’s 1550 MB/s ceiling, it’s almost as fast as the 4-lane ORICO.

So the ORICO beats the OWC by nearly 2x for large sequential reads, but for everything else they're fairly close.

 

[whodiini]

The review makes one assumption that is incorrect - that using that WD NVMe drive should remove that as the bottleneck. I own the envoy express, an orico thunderbolt and an acasis thunderbolt enclosure. I also have several NVMe drives. I have tested these combos on my M1 Studio Max and my 2018 intel i7 mac mini. The speed is VERY dependent on the NVMe drive used. Some NVMe drives just dont perform well on a mac. That WD NVMe he used is one of them. I have put a compatible NVMe into the envoy and the other enclosures and get ~ 2400 W/R speeds on the other enclosures and ~ 1300 W/R on the envoy, close to 1/2 the performance using a Spatium M470 PCie 4 NVMe.

On the other side, I have a recent SX8200Pro that runs at 1200 MB/sec on a Mac (M1 and intel). But putting it all 3 enclosures gives me the same speed. In that case, one could simply put that in a cheap USB 3.1G2 case and get 1000MB/sec. Or better yet, I returned it.

Some angel could put together a database:
Mac OS version Enclosure model Enclosure firmware NVMe drive NVME firmware BM Read speed BM Write speed

 

[whodiini]

The two Thunderbolt-certified ways to use all four lanes are (note that these are all wall-powered):

1) Get a TB-certified dual NVMe enclosure and run it in RAID 0. That would give you access to all four PCIe lanes. Options include:
a) Sabrent Dual-NVMe enclosure ($150)—though it's small, so you've still got the heat issue.
b) Sonnet Echo Dual NVMe dock ($350). The Sonnet has a fan, so there's a potential noise issue there, but it should keep the cards from overheating. And you get a full TB dock.
c) SIIG dual NVMe dock (~$350). It's fanless, so would be silent, but it probably has more thermal throttling than the Sonnet. Like the Sonnet, you do get a full TB dock.

A dual NVMe enclosure is a bad idea if you are trying to maximize speed. The reason is that it uses a single thunderbolt PCie x 4 bandwidth to transfer data. in Raid 0, that effectively limits each NVME to 2 PCIe lanes of about 1.5GB/sec. 4 PCIe lanes should give you over 3 GB/sec. Thus 2 separate TB enclosures connected to separate TB busses run in RAID 0 should give you PCie x 8. or >6 GB/sec, minus overhead.

 

[theorist9]

A dual NVMe enclosure is a bad idea if you are trying to maximize speed. The reason is that it uses a single thunderbolt PCie x 4 bandwidth to transfer data. in Raid 0, that effectively limits each NVME to 2 PCIe lanes of about 1.5GB/sec. 4 PCIe lanes should give you over 3 GB/sec. Thus 2 separate TB enclosures connected to separate TB busses run in RAID 0 should give you PCie x 8. or >6 GB/sec, minus overhead.

You misunderstand. I was speaking specifically to the 2-lane speed limitation of TB-certified enclosures connected to a single TB port, and explained that a way to get all four lanes available from a single TB port, in a TB-certified enclosure, is to get a dual-NVMe enclosure and configure it in RAID 0.

Responding that it's a "bad idea" because if you use 2 enclosures connected to 2 buses things are even faster is just silly one-upsmanship, and misses the point. It's no different than if you were talking about your proposed config, and I responded as follows:

"2 separate TB enclosures connected to separate TB busses run in RAID 0 is a bad idea if you're trying to maximize speed, since that limits you to PCIe x 8. If you want to maximize speed, you should do PCIe x 16 which, gives you 25,000 MB/s (https://www.amazon.com/High-Point-SSD7505-4-Channel-Controller/dp/B08H55V2T4)."

And then the next your-idea-is-bad-mine's-better guy can say:

"No, that's a bad idea to maximize speed, since you should configure two of those as a Cross-Sync RAID array, giving >40,000 MB/s"

And so on.

So let's not do that. Just because there's a bigger, more expensive tech solution out there, it doesn't mean the less expensive solutions are a "bad idea". Something is only a bad idea if it doesn't work as designed and needed.

*Incidentally, there's at least one commercial package configured for the Mac that does PCIe x 8: It peaks at 5300 MB/s reads:

iodyne Pro Data Review

The iodyne Pro Data is a laptop-sized RAID enclosure containing 12 NVMe SSDs, primarily focused on creative professional applications.

www.storagereview.com

 

[whodiini]

Sorry if you mistook my comment as oneupmanship. That was not my intent.

Yes, I did misunderstand your intent. Now that you explained it, makes sense.

 

[theorist9]

The review makes one assumption that is incorrect - that using that WD NVMe drive should remove that as the bottleneck. I own the envoy express, an orico thunderbolt and an acasis thunderbolt enclosure. I also have several NVMe drives. I have tested these combos on my M1 Studio Max and my 2018 intel i7 mac mini. The speed is VERY dependent on the NVMe drive used.

Yeah, as you probably know, ACAIS has a table showing the sequential write issues with various drives, and these might apply to the ORICO as well. Though they say, with a firmware update, you can correct the write issue on the SN850. Have you tried that? To do the update, I think you need a PC:

What I'm wondering is whether it's a write caching issue. When you connect these drives to PC's, write caching is sometimes turned off by default, which kills the write speed, but this can be corrected in Properties. In the Mac you can't make that correction (at least AFAIK, unless there's a way to do it in Terminal), so I'm wondering if that is what's going on here.

 

[whodiini]

I ended up getting a $100 windows laptop that had an empty NVMe slot to both sanitize/wipe NVme using partmagic and update firmware. Nvme drives are going thru a bit of a learning curve in terms of compatibility with different systems and computers, so having a way to do firmware updates easily is important. I have found that often, firmware updates cant be done with an external enclosure. Otherwise, one could use a windows VM on a mac. The highly regarded Hynix P31 gold seems to have the compatibility issue with the mac fixed with the latest update. I like that NVMe a lot because it has really good performance for its low power usage - a nice combo for laptops.

 

[theorist9]

I ended up getting a $100 windows laptop that had an empty NVMe slot to both sanitize/wipe NVme using partmagic and update firmware. Nvme drives are going thru a bit of a learning curve in terms of compatibility with different systems and computers, so having a way to do firmware updates easily is important. I have found that often, firmware updates cant be done with an external enclosure. Otherwise, one could use a windows VM on a mac. The highly regarded Hynix P31 gold seems to have the compatibility issue with the mac fixed with the latest update. I like that NVMe a lot because it has really good performance for its low power usage - a nice combo for laptops.

....And that low power usage would also make it a good match for these external enclosures.

I think part of the reason we're seeing these issues is that these NVMe drives aren't designed to be use externally. So when they are, issues arise.

For instance, I have an SN850 installed internally on my iMac, and I don't get the big sequential read-write discrepancy seen when this is used externally on a Mac or PC (at least w/o a firmware update):

Note: "Catalina 10.15.7" is a typo. All benchmarks were run on Monterey 11.4.

 

[whodiini]

Ah, no problem. I mistakenly thought you were referring to the TB42 (thats the enclosure+hub), since I didn't notice that on their site until a few days ago.

Yeah, that's what I've been told. Of course, one workaround is to add wall power, which the TB42 does (though it's not Intel-certified). I emailed ACASIS, and they said specifically that the wall power is indeed used to power the SSD, so it's not driven by the bus.

Based on the stability and speed of the acasis TB enclosure (and not the design), I decided to get the enclosure+hub. It is listed as TB42 on their US site and TB412 on their non-US site, but it looks the same. It was $20 more than just the enclosure. I want to try something to replace a TB3 hub with this: daisy chain a LG 5k monitor and also use a decent speed NVMe, powered by a tiny GaN 30W charger, It should use close to the entire 40Gpbs bandwidth. (5k: 22Gbps leaving 18Gpbs for the NVMe, or slightly over 2000MB/s for the NVMe). I know that the spec is for 40W to power the hub, but since the 5k doesnt need power and the NVMe draws only 10W max, 30W should be enough. Will report back when I get it. I like the idea of getting rid of my huge power brick on my current TB3 hub.

 

[theorist9]

Based on the stability and speed of the acasis TB enclosure (and not the design), I decided to get the enclosure+hub. It is listed as TB42 on their US site and TB412 on their non-US site, but it looks the same. It was $20 more than just the enclosure. I want to try something to replace a TB3 hub with this: daisy chain a LG 5k monitor and also use a decent speed NVMe, powered by a tiny GaN 30W charger, It should use close to the entire 40Gpbs bandwidth. (5k: 22Gbps leaving 18Gpbs for the NVMe, or slightly over 2000MB/s for the NVMe). I know that the spec is for 40W to power the hub, but since the 5k doesnt need power and the NVMe draws only 10W max, 30W should be enough. Will report back when I get it. I like the idea of getting rid of my huge power brick on my current TB3 hub.

Looking forward to hearing what you'd find. I was thinking about that device myself.

Their recommended 65W charger looks fairly small. I take it you decided to go with a 35W charger because you wanted something even smaller (?).

Any experience returning things to ACASIS if it doesn't perform as expected? My concern is I'd need to pay the return shipping which, internationally, could be a signifcant fraction of the cost of the device itself. It would be nice if ACASIS gave you a shipping label and just charged you the same as for the outbound shipping, which is only $11.59.

If this were available on Amazon that would eliminate this concern but, alas, it's not.

****
In the 2 TB size, the new WD SN850X is the same market price as the Hynix P31 Gold you mentioned ($200). Wondering about the tradeoff between getting one of the fastest SSD's made, vs. one you know won't have the write compatability issue.

****
Right now I'm testing a SanDisk Extreme to see if that is fast enough for my purposes. If not, I'll go with either the SIIG or the ACASIS.

I'm actually in the process of comparing the SanDisk connected to a TB3 vs a 5 Gbps USB-A port on my iMac. Once you've encrypted the disk (which cuts down on the speed, b/c I'm using FireVault rather than hardware disk encryption), the performance difference is much smaller than you'd think. The advantage of using the USB-A port is I wouldn't need to get a dock (currently I'm using both TB ports for monitors). But I noticed the SanDisk spontaneously disconnected when connected to the USB-A port, so that might not be a good idea. Don't know if it's because of the cheap USB-C to USB-A adapter that SanDisk provided, or because it's not a good idea to connect a USB-C 3.2 SSD to a USB-A 3.1 port.

 

[whodiini]

The new 2 TB WD SN850X is the same price as the Hynix P31 Gold ($200). Wondering about the tradeoff between getting one of the fastest SSD's made, vs. one you know won't have the write compatability issue.

Their recommended 65W charger looks fairly small. I take it you decided to go with a 35W charger because you wanted something even smaller (?).

Any experience returning things to ACASIS if it doesn't perform as expected? My concern is I'd need to pay the return shipping which, internationally, could be a signifcant fraction of the cost of the device itself. It would be nice if ACASIS gave you a shipping label and just charged you the same as for the outbound shipping, which is only $11.59.

If this were available on Amazon that would eliminate this concerns but, alas, it's not.

Occasionally Amazon has the P31 gold on sale. I just got the 2TB late Aug for $160.

Also, 30W GaN chargers are cheaper than 45W or 60W.

For all these Chinese made stuff, the warranty/support is non-existent with very few exceptions. Adata (TW) has US support. Orico wont answer emails. You have to assume that if it breaks or doesnt work, you are out of luck unless the seller (like Amazon) will take it back within 30 days, That means your warranty is 30 days, not the advertised 1 or 2 years. I bought mine from the Acasis factory store on Aliexpress because it is cheaper. A little over $100 is my upper limit for potentially disposable stuff and this is a bit over my limit, buut I am intrigued

 

[whodiini]

Yeah, as you probably know, ACAIS has a table showing the sequential write issues with various drives, and these might apply to the ORICO as well. Though they say, with a firmware update, you can correct the write issue on the SN850. Have you tried that? To do the update, I think you need a PC:
View attachment 2052579

This is the prior version of their chart. The WD850 got recommended with their new firmware. I also confirmed the Hynix P31 gold works with their latest firmware.

 

[EncryptedUser]

I get far better numbers on my M1 with my SN750 in their chassis than what they state there.