List of fastest USB4 ASM2464PD & TB3 JHL7440 SSD enclosures (May 2024)

TB3, USB-C, USB4  86 Comments

[2024/05/22 – Additional models]

  1. Introduction
  2. Fastest external M.2 SSD enclosures
    1. Thunderbolt 5 JHL9480 enclosures
    2. USB4 ASM2464PD SSD enclosures
      1. Retail models
      2. ODM models
    3. Thunderbolt 3 JHL7440 enclosures
      1. For TB3/TB4/USB4 hosts and NVMe SSDs only
      2. Backwards compatible with USB 3.2 host PCs and SATA SSDs
      3. Portable SSD docking stations
  3. SSD enclosure chipset list table
  4. Firmware
  5. Write Caching
  6. References

Introduction

There are four main vendors of controller chips used in external SSD enclosures: ASMedia, Intel, JMicron, and RealTek. Realtek and JMicron offer the only broadly available single-chip solutions that allow using either a SATA or NVMe SSD. These are commonly found in the $20-$50 enclosures like the Uni model I reviewed and are limited by USB 3.x speeds. Other vendor solutions only support SATA or NVMe SSDs, but not both. If you want the absolute fastest enclosure in 2024, there are two options for use with NVMe SSDs: USB4 or Thunderbolt 3.

Fastest external M.2 SSD enclosures

Thunderbolt 5 JHL9480 enclosures

Although Thunderbolt 5-equipped PCs are not yet shipping, there are a few vendors that already have SSD enclosures.

USB4 ASM2464PD SSD enclosures

These enclosures sustain higher throughputs compared to older Thunderbolt 3 designs which reserved bandwidth for DisplayPort and legacy USB. Real-world peak throughput is ~3.1-3.8GB/s with the latest Gen4 M.2 NVMe SSDs. Many enclosures since 2020 are marketed as “USB4 compatible” but are based on the older Thunderbolt 3 chips, so you need to pay attention to the fine print. Enclosures based on the ASM2464PD chipset typically exceed 3.2GB/s and should not have been listed for sale prior to July 2023. Some brands have had crowdfunding campaigns earlier. Few PCs available in 2023 are able to take advantage of these speeds. AMD CPU-based systems with built-in USB4 running the latest version of Windows 11 seem to yield the best results. Most of these enclosures use similar PCB designs and have almost identical performance characteristics. But some ship with old firmware that can affect performance in certain situations. See the firmware section below to see if updates to your enclosure are available.

Retail models

  1. AData SE920
    • ships with 1TB or 2TB SSD pre-installed
  2. BlueEndless BS-M280U4 (alibaba)
  3. C-Smartlink UD1101
  4. Hyper NextDrive HD5001GL (ebay)
  5. Jeyi TB2464/TB2464Pro/TB2464Fan (Amazon/Ali)
    • 3 different versions available: basic, larger Pro model, and Pro model with fan
  6. Maiwo K1695 (available at Amazon/NewEgg/BangGood)
    • possibly same manufacturer as Jeyi but with different chassis design
    • sold under various brands like iSheep with same K1695 model number
  7. Orico TCM2-U4
    • has cooling fan but heatsink not in full contact with SSD chips so they run hot
  8. OWC Express 1M2
    • available bare or with up to 8TB SSD pre-installed
    • all-metal chassis
  9. Satechi ST-EU4NPM
  10. Stardom UBOX-B4BP (has second USB-C port for power to work on older USB 3.x systems)
  11. ZikeDrive Z666 ASM2464PD (successful Indiegogo campaign)

ODM models

In the future, this list will be migrated into a distinct article comparing all the ASM2464 enclosure options.

Thunderbolt 3 JHL7440 enclosures

Most Thunderbolt 3 enclosures are based on older DSL6xxx/JHL6xxx chips (Alpine Ridge) and were limited to 22Gb/s out of an available 40Gb/s due to the way the controller reserved bandwidth for DisplayPort and USB traffic. But JHL7440 (Titan Ridge) solutions provide ~10% higher transfer speedswith real-world speeds maxing out at ~2.6-2.8GB/s. Many of these models are marketed as USB4 or Thunderbolt 4 but are actually just Thunderbolt 3.

For TB3/TB4/USB4 hosts and NVMe SSDs only

Backwards compatible with USB 3.2 host PCs and SATA SSDs

Portable SSD docking stations

SSD enclosure chipset list table

Depending on system configuration, (cables, hubs, mainboard, SSD enclosure PCB, SSD model, etc.) real-world speeds can be lower than what is noted below. The real-world speeds are the highest sustained throughputs that can be easily reproduced with high-end PCs. Some chips listed here are not intended for SSD enclosures and are included for completeness since since they can technically connect to an SSD.

VendorChipsetUpstream
interface		Downstream
interfaces		Theoretical
speed		Real-world speedNotes
ASMediaASM2464PD/
ASM2464PDX		USB4 with
PCIe Gen4x4		PCIe Gen4×440Gb/s~3200-3800MB/s– TB3/USB 3.x compat
ASMediaASM2364USB 3.2 Gen2 x2PCIe Gen3x420Gb/s~1900MB/s– faster than JMS586
ASMediaASM2362USB 3.2 Gen2PCIe Gen3x210Gb/s~875MB/s
ASMediaASM235CM/
ASM1352R/		USB 3.2 Gen2SATA36Gb/s~525MB/s– for HDD or 2.5″ SSD
ASMediaASM225CM/
ASM1351/
ASM1156/
ASM1153E/
ASM1153		USB 3.x Gen1SATA35Gb/s~450MB/s– for HDD or 2.5″ SSD
IntelJHL8440TB4 with
PCIe Gen3x4		PCIe Gen3x18Gb/s~775MB/s– TB3/USB 3.x compat
– for docking stations only
IntelJHL7440TB3 with
PCIe Gen3x4		PCIe Gen3x424Gb/s~2600-2800MB/s– USB 3.x compat
– controller limits speed
IntelJHL6x40/
DSL6x40		TB3 with PCIe Gen3x4PCIe Gen3x422Gb/s~2500-2600MB/s– controller limits speed
JMicronJMS586USB 3.2 Gen2 x22x PCIe Gen3x216Gb/s~1600MB/s– U variant supports SSD cloning
JMicronJMS583USB 3.1 Gen2PCIe Gen3x210Gb/s~875MB/s– A3 revision performs best
JMicronJMS581USB 3.2 Gen2PCIe Gen3x2
OR SATA3		10Gb/s~875MB/s
~525MB/s		– LT variant includes SDCard support
JMicronJMS580USB 3.2 Gen2SATA36Gb/s~525MB/s– integrates USB-C PD
JMicronJMS578/
JMS576		USB 3.1 Gen1SATA35Gb/s~450MB/s– for HDD or 2.5″ SSD
JMicronJMS562/
JMS561		USB 3.1 Gen12x SATA35Gb/s~450MB/s– for HDD or 2.5″ SSD
RealTekRTL9220DPUSB 3.2
Gen2 x2		PCIe Gen3x4
AND SATA3		20Gb/s~1900MB/s
~525MB/s		– dual ports
– RAID
RealTekRTL9210BUSB 3.1 Gen2PCIe Gen3x2
OR SATA3		10Gb/s
6Gb/s		~875MB/s
~525MB/s
RealTekRTL9211DSUSB 3.2 Gen2PCIe Gen3x2
OR SDCard		10Gb/s
8Gb/s		~875MB/s
~750MB/s		– supports PCIe & SDCard
RealTekRTL9200USB 3.1 Gen2SATA36Gb/s~525MB/s
TITUSB9261USB 3.x Gen1SATA35Gb/s~450MB/s– for HDD or 2.5″ SSD
ViaLabsVL717/
VL716/
VL715		USB 3.x Gen2SATA36Gb/s~525MB/s– for HDD or 2.5″ SSD
ViaLabsVL713/
VL711		USB 3.x Gen1SATA35Gb/s~450MB/s– for HDD or 2.5″ SSD

Firmware

If you notice your chassis is has poor write performance compared to benchmarks you find in various social media forums, consider downloading and updating the firmware.

Write Caching

To improve write performance, you can enable system RAM write caching in Windows. With any removeable storage device, there is a risk of data loss if the drive is disconnected during a cached write. But for certain SSDs, enabling this feature can improve overall system performance.

References

danchar

86 thoughts on “List of fastest USB4 ASM2464PD & TB3 JHL7440 SSD enclosures (May 2024)”

  2. Has anyone seen an ASM2464PD based enclosure that can accommodate 22110mm drives? Or, alternatively, know of a DIY solution using a transplanted or generic board? I’m not much into the maker space / 3D printing / CNC and couldn’t find anything just searching the internet.

    I’d love to upgrade my JMS583 based enclosures (IcyDock MB861U31-1M2B and Glotrends Sagittarius). I did see a Jeyi enclosure on Aliexpress that will take 22110 drives, unfortunately it’s based on the RTL9210B, so still limited to 10Gbps.

    Like

    Reply

    • Hi P.C. – which 110mm drive are you looking at? You have to be careful about the peak power as it’s unlikely a 110 would work with legacy USB-C laptops.

      Even with Thunderbolt/USB4, you only have 15 watts shared between the ssd and controller chip. And these chips run hot so I reckon you have at best 10 watts to actually use for the SSD. If your 110 is double-sided and has dram cache, I doubt it would work reliably.

      That being said, I am interested in taking on this project so I will consider disassembling one of my enclosures and designing a 3D- printable chassis for it.

      Most likely, it would be just dremmeling the original chassis to have a slit for the SSD to pop out with a cap to protect it.

      Like

      Reply

      • Hi Dan,

        After reading your reply it clicked for me why I haven’t seen any of these come to market: increased power draw for enterprise PCIe 4.0 and newer drives. You make a good point about compatibility concerns.

        I’m currently only using Intel Optane 905P 380GB drives in my JMS583 enclosures, but have used Intel Optane DC P4801X 200GB, Samsung PM983 1.92TB and an Intel/Solidigm DC P4511 2TB drive in the past. They are all PCIe 3.x drives and, according to documentation, have peak draws of 9.35w, 9w, 8w and 8.25w, respectively.

        I haven’t encountered any issues using them with Thunderbolt/USB4 or USB 3.2 Gen 2 Type-C (10Gbps) systems, just the protocol bottleneck that limits bandwidth when used on newer systems. I was hopeful that a ASM2464PD solution would help speed things up, particularly for reads.

        Like

        Reply

  3. Is it correct, that ASM2464 solutions can handle a maximum of 8TB only?
    Other solutions seem to be even limited to 4TB.

    Why is that?
    You might think that there’s no such limit, like in the USB/SATA controller world and considering 16TB drives NVMe should show up in the next 1-2 years.

    Like

    Reply

    • My understanding is that there is no actual 4TB or 8TB addressing limit for nvme ssd. Vendors just don’t test or warranty the higher capacities as there is little market incentive to do so.

      For portables and the m.2 interface, the bottleneck will probably always be power and cooling.

      IIRC, m.2 has 4x 3.3v power pins each capable of 0.5 amps. So a maximum of 13.2 watts by spec. With engineered-in headroom, it’s effectively never going to exceed 10 watts.

      For portables running on Thunderbolt you can be assured of 15W from the laptop, but from a regular PC, maybe 7.5W or 10W. And that power is shared between the controller in the enclosure and the ssd.

      So the question then becomes one of power efficiency of the Nand chips, Nand controller dram cache on board, and host controller. With current technology, you can create ssds exceeding 60TB in U.2 2.5″ form factor, but they consume close to 30 watts!

      Like

      Reply

  4. Which Enclosure with the ASM2464PD Chip is the fastest and can you recommend?

    Can you compare the ZikeDrive Z666 with the OWC Express 1M2 and what would you choose?

    Sorry for making multiple comments. Can you please delete the others?

    Like

    Reply

    • Hi don’t think the issue is the SSD enclosure – I suspect a particular module needs to be present in the system/mainboard UEFI firmware to support booting from USB4/Thunderbolt. You would have to ask the system mainboard vendor about that support.

      Have you considered a hypervisor approach to just boot a VM instead? If you haven’t tried this approach in a few years, there may be less performance impact than you think…

      Like

      Reply

      • Hi danchar,

        Thanks for your reply.

        The specific response OWC made to my inquiry was…

        “No, this OWC 1TB Express 1M2 does not offer boot support on Windows.”

        However, on the Amazon website for the 1M2 OWC writes…

        “The OWC 1TB Express 1M2 portable SSD is bootable, allowing users to install and boot their operating system from the drive for faster boot times and improved system performance.”

        Obviously they are in conflict with their own literature.

        As an experiment, I did the following…

        I restored my Windows internal SATA drive image that I made with Macrium Reflect onto a USB 3.1 Gen 2 SSD drive that I connected to my Thunderbolt 4 port and tried to boot it.

        The boot process brought up screen information that indicated that Windows could not start properly and it took me to diagnostics screens.

        I tried all of the available options and the only method that worked, was to do a “RESET” of Windows from Local Source.

        In essence, this very long process went through all of my Windows files on the USB drive and made some changes. Exactly what changes I do not know.

        When it finally booted, my icons were very scrambled on the desktop and many of my programs did not appear listed in the Start Menu. But it did run and work properly. I was able to boot up from power off and reboot as well.

        So it can boot from USB 3.1 Gen 2 enclosure, but very slowly and a bit disorganized.

        Therefore, I do not think there is any issue about booting Windows from the Thunderbolt port with Windows or the motherboard, at least on USB 3.1 Gen 2. I just don’t know if OWC’s 1M2 enclosure firmware and hardware will allow Windows booting since the info I got directly from them and from their online advertising literature are in direct conflict with each other.

        So I asked the question hoping someone had setup a bootable Windows on some other Thunderbolt enclosure successfully.

        I was not looking to boot a Virtual Machine, I wanted to run Windows from an external Thunderbolt drive the way that I currently do using eSATA, just with a faster data rate.

        Like

        Reply

  6. Hello,  I have some questions related to the USB4/TB3/TB4 enclosures discussion on the blog.

    I have a ASUS ProArt B550-Creator Motherboard.  I bought it as it was the first MB with integrated TB4 that I could find.  I am looking for the fastest TB4 NVMe SSD enclosure that I can buy.

    My motherboard only mentions TB4 (2x-USB TypeC) and USB3.2… or lower in its specifications.

    Looking over the Enclosure list of Chipsets I gather that to get the highest transfer rate from my motherboard I would want an enclosure that contained an Intel JHL8440.  I do not see an enclosure with that chip at its core. 

    So, what is currently the best solution for me and how do I know if will be functional.

    Thank you.

    Like

    Reply

    • HI – the fastest enclosures will be based on the ASM2464PD chips.

      JHL8440 is not a chip used for SSD enclosures and is only intended for docking stations. A docking station based on JHL8440 that has an m.2 slot built-in operate an SSD at less than half the speed of a dedicated ASM2464PD enclosure.

      Like

      Reply

  7. Hello Everyone from USB_Curious,

    As you can read from my prior post (April 21, 2024), way above this one, I have been having a difficult time with an external USB enclosure that uses ASMedia ASM235CM. I finally found a solution and it is partly applicable to the Thunderbolt 4 enclosures being discussed. So here is what I want to share.

    The specific enclosure is a OWC Mercury Elite Pro Mini which has eSATA 6Gbps and USB C 3.1 Gen 2 10Gbps connections. The enclosure has a power connector and I was using the OWC power adapter. When connected to power and ESATA or to power and USB C to USB A on the motherboard the SSD in the enclosure worked perfectly. But when connected to power and USB C to USB C on my Asus Z790 Creator WiFi motherboard, the enclosure would show as connected as if I was using a USB 2.0 port with slow speeds that correspond to the USB 2.0 protocol. I tried everything I could think of, posted in various forums looking for help cause I found similar complaints about the ASM235CM all over the net, contacted Asus, contacted OWC, etc. I also had an Oyen USB C enclosure that also uses the ASM235CM that worked fine connected to the same ports, no slow downs, no weird USB 2.0 indications on HWInfo64, so I was sure it was something to do with the OWC enclosure. I tried switching cables, downgrading firmware, etc. No luck. I asked ASMedia and OWC for the latest firmware cause I was nervous about using the Russia website many people have gotten their firmware updates from. ASMedia said they do not sevice end users, but they would give it to OWC to give to me if OWC requested it. OWC flatly refused to request the firmware from ASMedia and refused to give it to me. Oyen gave me updated firmware for my Oyen USB C enclosure with no hassle whatsoever, but the Oyen has no eSATA and has a VIA VL822 Hub built-in so I did not want to risk using it in the OWC enclosure. The Oyen firmware updated my Oyen enclosure with no issues whatsoever. Many thanks to Oyen for their help and great enclosure. OWC tech support was openly hostile. They led me in circles with nonsense. They could not understand that Thunderbolt 4 connectors on my motherboard were backward compatible with USB 3.2 At one point they asked me to open up my system box and photograph all of my edge connectors all around the motherboard. WTF ???

    A few days ago, I decided to update my motherboard firmware for other reasons. Updating motherboard firmware is not the insignificant task tech support people would lead you to believe. You have to first backup all of your hard drives / ssd’s. Make sure your backup is restorable. How many of you actually do this? Photograph all of the BIOS settings cause you know you can’t find the various slips of paper you scribbled down your mods over the years. Download the current and new firmware and prepare a reliable USB Flash drive. Include new drivers like Intel ME Interface, Intel Chipset, Thunderbolt driver, Serial IO driver, etc. After you rename your new firmware file and extract all of your drivers, you say a prayer and start updating everything and then re entering your BIOS mod settings. It takes time and is best not done all at once. After I succeeded with all of this, I decided to retry the problem enclosure. No change whatsoever. Same exact problem. I was ready to just abandon any hope of resolving this.

    I was looking at a PC Magazine review about OWC 1M2 and in it they remarked that they had problems with really slow, (slower than mechanical hard drives), writes. They got OWC to help them. A potential bad review from a major tech magazine was apparently far more incentive that my user requests to their tech support team. The help was to tell them to set the enclosure Policies in Windows 10 to Better Performance instead of Quick Removal. That solved the magazine’s problem. I tried and it did not work for me. But then I had a light bulb moment. eSATA connection on my enclosure requires the external power adapter and the power adapter worked fine with USB C to USB A. But what if OWC never bothered to check if it worked with USB C to USB C connection? So I removed the power adapter, ( I have several and had tried swapping them earlier on in my troubleshooting with no improvement), and tried again with the Better Performance setting in Policies. It worked. I was now getting 10Gbps connections and transfers to and from the enclosure. 3 months of trouble shooting and this was the answer. Argh !!!

    So, in summary: 1) If you have slow transfer rates and have tried all of the sensible things to test, try removing your power adapter and use bus power. 2) Set your Policies to Better Performance. 3) Assume that you will not get usable help from OWC unless you are a big tech magazine. Most of the many products I have bought from them have been good, but quirky. Usually I can figure things out on my own, but when I needed them, they were not helpful. Oyen on the other hand was extremely helpful.

    Now you know.

    Like

    Reply

  8. New USB4 enclosure (ASM2464PD)

    Hagibishttps://www.hagibis.com/usb4-m2-nvme-ssd-enclosure-p00286p1.htmlhttps://www.hagibis.com/usb4-nvme-ssd-enclosure-p00297p1.html

    Like

    Reply

  9. Has anyone tried the Hagbis MC40 TB4/USB4 NVMe enclosure? It’s a new enclosure that was released recently based on the ASM2464PD controller. I really like that is doesn’t have a fan but instead heat dissipation fins. Got lucky and bought one for $30 from AliExpress when I noticed it in the SuperDeals section. Just waiting for delivery.

    Like

    Reply

    • Hey @weirdg,

      Any update on the Hagibis enclosure you ordered?

      I’m also wondering about it, and especially if the fins work the best out of all the enclosures for heat dissipation.

      And I wish it was available for $30.

      Let me know what you think about it, thanks!

      Like

      Reply

      • @SK will do. I saw it listed for $30 and bought it right away on a whim. Didn’t know anything about it but knew that Hagibis makes accessories for Apple products along with some non-Apple products as well, and their products appear to have decent reviews.

        I’ve seen it drop to $50 a bunch of times on AliExpress, so hopefully it drops to $30 again.

        I think I’ll get it sometime this week, so I’ll put it through the motions then.

        Like

        Reply

      • Okay… I can finally write a bit of a review now. I received the Hagibis enclosure a couple weeks ago, but didn’t have any NVMe drives I could use with them. I ended up ordering two WD Black SN770 2TB NVMe drives this week and received them just today. I put one into the Hagibis MC40 TB4/USB4 NVMe enclosure and the other into an Acasis TBU401 TB4/USB4 NVMe enclosure. Both enclosures are brand new.

        Test environment:

        • Dell Latitude 7320 laptop with i5-1145G7 CPU, 16GB RAM, and Thunderbolt 3 port
        • Windows 11 Pro installed
        • Both enclosures are using a Western Digital Black SN770 NVMe drive with the latest firmware installed (731120WD)
        • The USB4/TB4 cable used was the one provided by Hagibis with the Hagibis MC40 enclosure

        Here are my pros and cons for the Hagibis MC40 enclosure:

        Pros

        • Well designed and excellent build quality. It’s built like a tank.
        • Alloy body and fin.
        • Fanless design. It uses heat dissipation fins to keep the drive cool.
        • Easy to dissemble the enclosure and install an NVMe drive.
        • Rubber feet so the enclosure doesn’t move all over the place.
        • They provided two thermal pads for the NVMe drive and two for the controller chip, so you have spares if you ever need to replace them.
        • A 0.6 ft USB4/TB4 cable is provided. The cable appears to be heavy duty and has an LED display for power. Great quality IMO.
        • Tools are provided. (Torx screwdriver and flat head screwdriver)
        • The screw to hold the NVMe drive in place is metal and has thumb grips.
        • The enclosure has a dual color indicator light.
        • Supports drives up to 4TB.
        • Uses the ASM2464PD controller chip.
        • For $1 extra on AliExpress, I also got a USB-A adapter when I ordered the Hagibis MC40. The USB-A adapter was an add-on and is not included with the MC40.
        • During my tests using Crystal Disk Mark and AS SSD Benchmark, temperatures peaked at 39 degrees celsius. Idle temps hover around 30 degrees celsius.

        Cons

        • Not toolless.
        • The enclosure casing uses the less common torx screws. Make sure you don’t lose the torx screwdriver.
        • The screw to hold the NVMe drive in place requires a flat head screwdriver.
        • The enclosure is bigger than the Acasis TBU401 enclosure.

        Benchmarks (I ran each test more than once and took the highest score):

        Here are my pros and cons for the Acasis TBU401 enclosure:

        Pros

        • 100% toolless.
        • Well designed, excellent build quality, and pocket-sized.
        • Aluminum build.
        • Fanless design.
        • Easy to dissemble the enclosure and install an NVMe drive.
        • They provided two thermal pads for the NVMe drive (1mm and 0.5mm).
        • A 1 ft USB4/TB4 cable is provided.
        • There’s a rubber nub or nipple to hold the NVMe drive in place.
        • Supports drives up to 8TB.
        • Uses the JHL7440 controller chip.
        • During my tests using Crystal Disk Mark and AS SSD Benchmark, temperatures peaked at 50 degrees celsius. Idle temps hover around 33 degrees celsius.

        Cons

        • Pretty average looking cable, but gets the job done.
        • No LED indicator lights.
        • No rubber feet.
        • No option to purchase a USB-A adapter as an add-on or extra.

        Benchmarks (I ran each test more than once and took the highest score):

        Size comparison:

        Conclusion:

        • Interesting that the AS SSD Benchmark score of the Acasis TBU401 enclosure was higher than the Hagibis MC40 enclosure.
        • I doubt real-world use cases would be significantly different between the two enclosures.
        • If you need something portable for on the go, the Acasis TBU401 is definitely the better choice.
        • If you need something that will stay permanently connected to you computer with lower temperatures, the Hagibis MC40 is the better choice.
        • If size doesn’t matter, then just get the cheaper one.

        For anyone wondering about pricing, I was patient when I purchased the Acasis TBU401 off AliExpress. I waited a bit and purchased it when I thought it was at the lowest price I had seen over the past few weeks. I probably could have waited longer for a lower price, but I don’t think I overpaid.

        I actually got lucky with the Hagibis MC40. I wasn’t looking to buy another NVMe enclosure and noticed it was on sale for an unbelievably low price considering it was a brand new enclosure, so I jumped on it.

        I purchased both off AliExpress for the following prices:

        • Acasis TBU401 – $62.79USD
        • Hagibis MC40 w/USB-A adapter add-on – $29.93USD

        Like

        Reply

  10. My experience is with the MPTool for the ASM235CM so the things I have observed may not apply to you, but I will share them anyway just in case they can help.

    I have numerous enclosures with the AS Media ASM235CM acting as the USB to SATA bridge. I only recently acquired a motherboard that has USB C connections for Thunderbird 4 and power charging. So even though I have had the enclosures for awhile, it is only recently that I can use a USB C to USB C connection. Prior I was using USB C to USB A cables without a problem at 5 Gbps.

    The enclosure from brand “B” has a firmware with date code for 2021. It works great with 10 Gbps transfer rates. The other enclosures from brand “A” have firmware date codes from 2017 and when they are connected to the same PC running Windows 10 Pro and using the same cable from the brand “B” enclosure, they connect at USB 2.0 speeds. So I felt that the firmware needed to be updated. I have not had success, but here is what I have learned about MPTool.

    The AS Media firmware update package that you find on station-drivers or usbdev usually is a compressed file that contains the MPTool, the firmware .bin file, the .ini configuration file, and often .log files from whoever uploaded it, and .doc and .pdf files in Chinese. AS Media does not release the tool or firmware to end users. These packages are uploaded by people who have managed to get ahold of it.

    The unlock code for this version of MPTool is asmedia

    The firmware filename is yymmdd_Dx_xx_xx.bin The numbers to the left of the D are the date of the firmware. It is not the date it is uploaded to the website which can be very different. That is, an older firmware may have been loaded later than a newer one. The numbers to the right of the D are the version of the firmware for the particular company’s enclosure. The firmware has configuration information embedded into it for a particular enclosure for a particular company. SOME but NOT ALL can be changed in the configuration .INI file. The .INI file has plain english description of preliminary settings and Vendor ID Code and Product ID Code of the bridge circuit. AS Media’s vendor code is 174C For reasons unknown, some manufacturers like to substitute their own manufacturer Vendor ID Code which makes it very hard for Windows to know what driver to load for the enclosure. The product ID Code is another headache. Nowhere can I find the correct code for the ASM235CM. Sometimes it is called ASM2235 or ASMT2235. The brand “B” enclosure I have which functions correctly uses Product ID Code 55AA which applies to about 4 different AS Media bridge chips, but none of them are the ASM235CM. So Windows lists the device as a different chip, but in brand “B” enclosure it works just fine. I tell you all this because you may think you have the right firmware, but is it really right? Sometimes you can look directly at the circuit board and see what is there. Other times, the chip is covered up and you have no idea what is really soldered in there. RESIST THE URGE TO POKE AROUND WITH YOUR FINGERS UNLESS YOU ARE WEARING GROUND STRAPS ON YOUR WRIST AND ANKLE. These devices are very sensitive to static electric.

    There are two parts to updating firmware with this tool, the configuation and the firmware itself. When you update the configuration, it seems to set the chip back to some preliminary version of the firmware. This may be some type of failsafe. When you update the firmware itself, it overwrites SOME of the configuration information based on what is embedded in the firmware .bin file. If you try to update both at once, it will fail. If you do not set the firmware filename and path and configuration filename and path both in the .INI file (use a plain text editor like NOTEPAD and save the original first), and also in the MPTool dialog, it will fail.

    If you are doing this on the AS235CM, you need to use a USB C to USB A cable.

    Update the configuration ONLY first by unchecking the boxes in the tool regarding the firmware and reflash. It should work quickly if you set the values in the .INI file correctly and loaded it into the MPTool dialog. Eject, power off/unplug. Wait 10 seconds, then power back on/plug back in.

    Then, update the firmware ONLY by checking the firmware and reflash boxes, but unchecking the update configuration box. ALSO UNCHECK THE ADVANCE SERIAL NUMBER BOX. Remember, this was intended to by a mass production tool. It will take longer to update the firmware. When done, pass or fail, eject, power off/unplug, wait 10 seconds, then power on/plug in again.

    Hope this helps.

    Liked by 2 people

    Reply

  11. I’m trying to to update firmware for the ASM-2464PD in the Satechi ST-EU4NPM. It looks good it the updater software but at ~17% I get a message with “FAIL”. There are no other messages not even in the log file. The device still works as normal. I have tried several times. Has anyone seen this or have any suggestions?

    Like

    Reply

  13. I can’t seem to get my JEYI TB2464 to connect to my Intel Titan Ridge (JHL7340) at TB3 speeds. I’m on Windows 11, I’ve upgraded the firmware on the AS2464PD controller to the latest one (March 8). On an M2 MacBook it works fine and I’m getting speeds up to 3000 MB/s. On Windows 11 the firmware tool shows it connects at “PLUSx1”, which seems to mean USB 3.1 SuperSpeedPlus (so 1 lane of 10Gbps). Sequential reads in Windows seem to max out at the 1 GBps mark so that about corresponds to that 10GBps lane. Has anyone had success in getting this to work at TB3 speeds in Win11 using a TB3 host?

    Like

    Reply

  14. Great article. Thank you.

    Question: These enclosures are listed as USB4. From reading, it seems Thunderbolt 4 specs exceed USB 4 specs. Are the enclosures with the ASM 2464 Thunderbolt 4 capable ? Or do they just meet the USB 4 specs ? Is this just a licensing issue ? Will a Thunderbolt 4 port on a motherboard work fully capable on one of these ASM2464 USB 4 enclosures ?

    Thank you for any insights.

    Like

    Reply

  15. Hi Danchar en everyone here,

    I am a noob when coming to this and have some questions. Is the enclosure not just an enclosure? What I mean with it, the board does all the work, so if I remove the board from the enclosure the board will still work, right? The reason for asking this is that I just want the board with the controller, no enclosure. I want you make an enclosure myself that is as thin as I can get it, with enough heat dissipation of course. Or is the firmware tied to the enclosure? So is it possible to order only the board? Or get the board? Hopefully my questions makes sense, if not let me know so I can explain it more.

    kind regards,

    Vids Yoh

    Like

    Reply

    • Yes you can safely extract the PCB and fashion your own bx with a superior heatsink with no ill effects. Many of these enclosures look nice but are not engineered with max cooling in mind.

      Like

      Reply

  18. If anyone is having trouble with the Jeyi (or similar ASMedia 2464 based) USB4/TB enclosure having lower speeds than expected, it may need a firmware update. Until I installed the 2024 January 01 Jeyi firmware update, my enclosure was being reported in System Information / PCI (or System Information NVME, click on the external SSD) as a x2 link width PCI NVME controller.

    Having only two lanes limited the bandwidth and the r/w performance of my (x4) NVME SSD, and for my use case it is significantly faster (almost but not quite twice as fast) after installing the new firmware. Your mileage may vary depending on the quality of your SSD–e.g. a fast PCIe 4.0 SSD might nearly reach USB4s practical maximums even when using only 2 PCI lanes, and a few SSDs might be x2 by design (rare nowadays, but some cheaper older SSDs were x2).

    Note that although this problem seems to (either primarily or exclusively) affect Apple Silicon Macs, the firmware update tool runs only on windows. The files appear to be from ASMedia, so it might be possible to install this firmware on other enclosures.

    Like

    Reply

      • Firmware updates generally do not affect the SSD data, however the vendor will always recommend a full backup prior to applying firmware. There is always a chance of bricking a device with any firmware change.

        Like

        Reply
  20. Pingback: mac: New recos for SATA to USB for SSD on Mac – Tong Family

  22. Just looking at the OWC Express 1M2 on BH Photo and found the following interesting from their tech spec description:

    “When used with a Thunderbolt 3 port on an Intel Mac or Windows computer, or a 10 Gb/s USB-C port on an iPad, Mac, or Windows computer, the Express 1M2 will run as a 10 Gb/s USB 3.2 Gen 2 device with speeds of up to 990 MB/s.”

    https://www.bhphotovideo.com/c/product/1801761-REG/owc_owcus4exp1mt01_1tb_express_1m2_portable.html

    Unfortunately, I’m seeking an M2 enclosure (and possibly hub) solution for my Lenovo X280 (Win 11) business laptop, which only provides TB3 and USB-C 3.1 Gen 2 ports.

    Click to access ThinkPad_X280_Spec.PDF

    Would love the faster speeds and beautifully designed Express 1M2 enclosure, but not sure the added price is worth it given the far less expensive USB enclousure out there.

    Thank you for the very helpful website.

    Like

    Reply

  23. I benchmarked both the Acasis TBU401 and the Zikedrive Z666 containing a Samsung 990 Pro 4TB attached to the thunderbolt port of my NUC-12 extreme.
    Although sequential read performance was a spritely 3111 MB/s in both cases, I was disappointed to find that the write speed was only 463 MB/s for the Acasis and 466 MB/s for the Zike. I was able to achieve faster write speeds via USB 3.2×2 (1007 MB/s) than Thunderbolt.

    danchar – do you have any idea why the write performance would be so much worse than read? I am wondering whether this is a prevalent issue affecting all 12th Gen Intel series PCs with thunderbolt 4 ports…

    Like

    Reply

  24. Hi,

    Thanks for your writeup. I want to add a bit more info from my experience with two enclosures.

    Computer is an M1 Mac Mini. SSD is a Solidigm P41 Plus 2TB which is a PCIe 4.0 Gen 4 NVME. Brand new.

    I first purchased the Orico TCM2-U4 since I have owned other Orico enclosures and have been happy with them. Decent looking case. I like the crystal clear plastic, though the LED was a bit bright for my tastes. Fan noise was acceptable; about as loud as my Mac Mini’s fan, but silent would be better. Used the included USB cable.

    Performance looked very good at first. 3600MB/sec reads and 2800MB/sec writes in Blackmagic Disk Speed test. However the speeds would drop off dramatically after a few minutes – down to below 1000MB/sec. I had a theory it was heat throttling, so I used smartctl and watched it. At idle, the SSD reported 55C (about 68F/20C room temperature). At load it would shoot up to about 70C after 30 seconds to a minute, and then throttling would happen, and it would be apparent in the smartcl report. Examining the drive carefully, it’s just a poorly designed enclosure. The fan circulates air from outside to inside the unit, but the heatsink doesn’t contact the drive directly (so why even bother including a thermal pad?), so it’s not doing an efficient job cooling as it’s relying on airflow past the bare SSD, not conduction through the heatsink.

    I contacted Orico’s support team to inquire about the cooling and they pretty much said it was normal and expected behavior due to the design. Here’s what they said:

    there is no contact between the primary heatsink on the enclosure and the thermal pad,Please allow me to explain the design principle for you :
    This enclosure is a turbo fan active cooling, through the air duct design, if the heat dissipation silicone and the metal plate completely affixed to the air duct will be blocked, affecting the fan heat dissipation

    I’ve since returned the enclosure.

    I then purchased a Maiwo K1695 enclosure. It does not have a fan, and uses a fairly standard approach of a thermal pad on the SSD, which touches the heatsink which is the outer aluminum case. It’s simple and effective – the case is warm, but not burningly hot. Performance is identical to the Orico (from System Profiler it appears they both use the ASMedia 246x chipset), but it doesn’t suffer from the same thermal throttling. It idles at about 40C, going up to 50C if I run benchmarks on it for 10 minutes straight to really heatsoak it. This is the one I’m keeping.

    tl;dr Don’t bother with the Orico TCM2-U4 as it’s kneecapped with a poor thermal design and will throttle performance within minutes.

    Like

    Reply

      • I actually need to buy a new enclosure anyway, and prices don’t seem much different between TBU401/405 and the latest generation so figure I may as well go something future-proof. I’m looking for something as small as possible but with decent thermals. Have you tested any of the models with a fan? If you were buying now which one would you get? Thank you for any advice.

        Like

        Reply

  27. Thanks for your reply and inspirations, but its a bit too complicated for me :-). The reasons behind are mainly for that issue https://arstechnica.com/gadgets/2023/02/samsung-releases-firmware-fix-for-rapid-failure-issue-in-new-990-pro-ssds/
    or things like that https://www.techpowerup.com/310053/seagate-releases-directstorage-firmware-for-the-firecuda-530-ssd

    It seems nowadays kepping a basic storage product also in the update loop is important, but what to do if you just have a notebook in place 😉 or a SSD in a PS5…

    I hoped with actual SSD enclosure chipsets a direct connection to PCIe is possible, via ?Thunderbolt? and for the SSDs vendors update tool suite is now transparent, but it seems there is always a “bridge” in between :-(.

    Like

    Reply

  28. Hi DANCHAR, this website is pretty much the best site for enclosure chipsets. Many thanks for that. I am looking for an enclosure that allows me to update the firmware of SSDs. Unfortunately, this is now necessary again and again and not possible for me due to the lack of a desktop PC. Is there an enclosure out there that makes this possible? Possibly ZikeDrive USB4 NVMe SSD Enclosure Z666 with latest chipset? Thx.

    Like

    Reply

  29. The HyperDrive Next USB4 NVMe SSD Enclosure I received a few days ago does not seem to have an ASM2464 controller

    here is what System Info says:
    USB4 SSD DEVICE:

    Vendor Name: HYPER
    Device Name: USB4 SSD DEVICE
    Mode: Thunderbolt 3
    Device ID: 0x1701
    Vendor ID: 0x339A
    Device Revision: 0xE3
    UID: 0x174C210D55882570
    Route String: 3
    Firmware Version: 12.34
    Port (Upstream):
    Status: Device connected
    Link Status: 0x2
    Speed: Up to 40 Gb/s x1
    Current Link Width: 0x2

    Like

    Reply

    • Hi the Hyper does indeed have the ASMedia chip – but you won’t see the benefits on Current Mac models which are speed limited.

      Plug it into an AMD CPU based PC with USB4 and you’ll immediately see the difference 🙂

      Like

      Reply

  30. The HyperDrive I recently bought directly from Hyper a week ago did NOT come with an ASM2464 chip as far as I could tell.

    The model number on the box says its a HD5001GL-90, and that it’s the “HyperDrive Next USB4 NVMe SSD Enclosure”

    It benchmarks similar to what you’d get with a JHL7440 chip

    It connects at 8 GT/s at 4x link width

    System Information under macOS 14.2 beta 3 gives me this:

    USB4 SSD DEVICE:

    Vendor Name: HYPER
    Device Name: USB4 SSD DEVICE
    Mode: Thunderbolt 3
    Device ID: 0x1701
    Vendor ID: 0x339A
    Device Revision: 0xE3
    UID: 0x174C210D55882570
    Route String: 3
    Firmware Version: 12.34
    Port (Upstream):
    Status: Device connected
    Link Status: 0x2
    Speed: Up to 40 Gb/s x1
    Current Link Width: 0x2

    Like

    Reply
  31. Pingback: List: 2TB, 1TB, 512GB M.2 2230 SSDs (updated 2023/11/16) – upgrades for Surface Pro7+/X/8/9, Laptop 3/4/5, SteamDeck, XBox Series X, etc. – Dan S. Charlton

  32. Thank you Dan for providing such valuable information and keeping it updated. My main point of interest is portable SSDs and I have a YouTube channel on it in my local language (Turkish). I have just posted a video about USB4 enclosures and your blog has helped me so much during the research period. I currently own HyperDrive Next USB4 enclosure with ASM4242PD controller and had the chance to test it on a MacBook Pro M2 Max, an intel Z790 motherboard (Asus ROG Maximus Z790 Extreme) and on a laptop with AMD Ryzen 9 7940 HS processor (Asus TUF Gaming A15).

    To give you and the other people reading this blog a quick summary, I can say that on Mac it reaches R: 3406 MB/W: 3220 MB per second. On intel Z790 mobo with Thunderbolt 4 ports it resulted in R: 3096 MB/W: 2917 MB per second. AMD system is where the controller shines and I was able to reach R: 3813 MB/W: 3826 MB per second speeds. On AMD system, I was able to copy a single 100 GB file in 42 seconds with an average speed of 2470 MB/s.

    I would be glad to share further info with anyone who replies to my comment.

    Like

    Reply

    • Quite interesting – thank you.

      I suspect the bottlenecks you’re seeing on the Intel platform is due to the JHL8540 chipset not wanting to saturate the bus in case DP or legacy-USB traffic comes along.

      I think the AMD mainboard USB4 solutions will always be superior to Intel TB4 as they will not reserve bandwidth but just interleave as needed if/when DP and legacy-USB traffic show up.

      Mac? Who ever knows what going on there 🙂

      I think Thunderbolt 5 and USB4-v2 will make things even more interesting – in theory we can then run one of these SSDs AND have multiple DP monitors and USB without slowdowns all on one cable 🙂

      Like

      Reply

      • Dan! Thanks for this writeup! I made a blog post comparing two different NVMe on the HyperDrive and Zike Drive with my M2 MacBook Pro, and referenced your writeup as it was the start of my journey trying to build the fastest external SSD I feasibly could: https://benrabicoff.com/fastest-ssd/

        From my understanding, the MacBook Pro caps out because you need PCIe 4.0 link rates through the entire chain, and Thunderbolt 3/4 on the MacBook Pro uses PCIe 3.0 x 4 lanes, limiting to 3,500 MB/s after PCIe overhead.

        Much of this was over my head, but would love your take! Cheers!

        Like

        Reply
  33. Pingback: Plugable UD-4VPD USB4 desktop dock review – Dan S. Charlton

  34. What happens when you attach an enclosure based on the ASM2464PD controller to a Thunderbolt 4 port? A Thunderbolt 4 port supports USB 4 (up to 40Gb/s) on the Mac mini M2 Pro for example. Will the connection be negotiated to support the real-world 3600-3800MB/s ?

    Like

    Reply

    • The performance will vary depending on the Host PC implementation but in almost all instances, the throughout will exceed the equivalent Intel JHL7440 or JHL6xxx enclosures.

      My understanding is that the Intel CPU or dedicated Maple Ridge host controllers like the JHL8540 /JHL8340 do not implement DisplayPort/USB bandwidth reservation like the expansion-device-specific JHL7440. So its possible to get the “full” bandwidth.
      One example on an Intel platform: https://www.reddit.com/r/UsbCHardware/comments/16bqgtg/asm2464pd_usb4_throughput_testing_with_gpu_and/

      On Mac, it should be the same.

      I am not sure what happens when you connect the ASM enclosure to a Thunderbolt 3 host – I mean it will work, but I have no idea what the bandwidth would be.

      Like

      Reply

      • A TB 3 expansion device connected to a TB 4 host port establishes a PCIe Gen3x4 connection. Does the TB 4 host port maintain a PCIe Gen3x4 connection when connected to a USB4 device, which is offering Gen4×4 upstream?

        Like

        Reply

        • See this for some context: https://www.reddit.com/r/UsbCHardware/comments/16bqgtg/asm2464pd_usb4_throughput_testing_with_gpu_and/

          On Intel platforms currently available, only PCIe gen3x4 can be tunneled.
          On Mac I’m not sure but I suspect Gen3x4 is the limit as well.
          On AMD platform, presumably Gen4x4 is possible with ASM4242 but I haven’t seen any evidence of this. I’m pretty sure AMD’s built-in USB4 implementation is limited to Gen3x4.

          Upcoming Thunderbolt 5 and USB4 ASM4242 host controllers will be able to tunnel Gen4x4 from the host.

          At the downstream device end, it appears that the ASM2464PD chip sustains PCIe Gen3x4 upstream while having a downstream Gen4x4 link. The host to device throughput would never exceed the Gen3x4 (~32Gb/s), but there may be some advantages in IOPS, scheduling, latency, etc by using the Gen4x4 link inside the enclosure. I’m not familiar enough with PCIe to understand if there would also be some fixed overhead penalty for having to “translate” or buffer Gen3 vs Gen4.

          Like

          Reply

    • I’m out of the loop on this one – what’s wrong with JMicron? JM, RT, and AS are all based in Taiwan…

      Are you referring to the StuXnet device driver signature cert thing from 13 years ago???

      Like

      Reply

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.