Another issue I am having is when booting from this SSK NVME drive on USB-A or USB-C (so far tested on ports 2 via USB-C and 5 via USB-A), I keep getting I/O lockup (100% utilization but no read/write, locked at 0/0 for read and write metrics and the disk queue grows exponentially) for that device when it hits max speeds. I am wondering if that could be a power delivery issue or something else. Don’t see this issue on my other computer so it’s just weird. Maybe the Framework firmware isn’t negotiating the correct amount of power to the device?
That info about the USB ports power is new.
Previously the page did not mention power.
1 Like
I’ve been on that page a bunch of times and don’t recall having seen that on there before. So that seems about right to me.
You could test the actual speed by copying large files from your laptop to your USB device, and watch the speed there, a 500MB/s speed is about USB3.0(5Gbps). Does the SSD with enclosure faster than the SanDisk?
The USB ports on the FW16 are a little buggy and have compatibility problems.
A) Port 1, 4
B) Port 2, 5
C) Port 3, 6
A, B, C each have a different make/model of USB chip driving them.
So, if you have a compatibility problem in C, try A or B instead.
I have found Port 1, 4 to be the most compatible with devices, so I hang NVME enclosures off port 1,4.
I put thinks like USB headphones in port 3, 6.
I put power brick input in port 2, 4.
Yes, sorry I didn’t post screenshot. I did speed test with crystaldiskmark and got 900mbps on both devices. SanDisk on port 6 and 5. SSK enclosure on port 5.
Yeah I have my dock on port 1, power on 2, headphones on 3. However I made the mistake of not ordering more than 2 USB-C I guess since I only have USB-A and HDMI on my right side 4-6 ports. I guess I need to order 2x USB-C ports and eliminate the HDMI on port 4 for USB-C instead, USB-C on 5 instead of type-a. Kinda annoying but is what it is
900Mbps is between USB2.0 and USB3.0. Do the speeds change when using port 1 & 2?
Oh sorry, typo! ~900MBps on read, ~1000MBps on writes, not mbps. doing the math of 900x8 = 7200mbps or 7.2gbps. Doing the write speed math of 1000x8 = 8000mbps or 8gbps. So the speeds are surpassing USB 3.0 (USB 5gbps) and are in the realm of USB 3.2 (10gbps) regardless of port 5 or 6. Speeds did not change when using port 5 on USB-A or port 2 on USB-C, have not tried port 1 or 2 but let me give that a shot now.
Yep, same speeds on port 2 with a USB-C interface, so nothing going wrong with SanDisk, it’s getting it’s max speeds on ports 4/5/6, but then the SSK enclosure won’t be recognized on port 6.
Just tested again on port 6 so I could get a screenshot of the SanDisk external. Only ran the SEQ1M to save time though.
That’s my power draw when using an SSD enclosure regardless of port used. However my write speed is 400~500MB/s (USB3.0) because the SSD is SATA not NVMe as I don’t have a spare NVMe SSD to test so idk whether the power draw will double (exceed the 900mA) when using a faster SSD
1 Like
Just tried port 1 on USB-C and same issue. I still get hang and lockup for the whole system when I load up the drive with an intensive process. Definitely sucks to have this issue when I didn’t with my last computer but oh well, I guess I can’t use this drive without a USB 3.1 hub in the way to throttle it.
Definitely tempting me to get my own meter now lol. Given that mechanical external hard drives generally cap out at 1.2A power draw, I’ll just use that as my reference and say these drives likely won’t exceed 900mA but might under certain circumstances or depending on the enclosure. SanDisk might have made their external more efficient than most enclosures, and most enclosures might look for a 1.5A power budget before powering up. I guess I’ll just write it off as that
You can often find the power draw information in the specs of the specific drive in the enclosure. For example the SSD I’m waiting on currently looks like this:
Seems like it could actually draw close to 2A on write, but I’m doubting it goes quite that high when limited to 10gbps usb speeds.
Yeah sadly this manufacture doesn’t list that data, but based on research NVME drives can pull around 4.2W max, so if we count the enclosure it’s entirely possible it could either be right below 900mA or right above it I guess. Windows seems to think it’s 900mA but maybe at a firmware level it creeps right above, idk.
Testing today on my desktop (a ryzen 5 3600 system) on a USB-A 10Gbps capable port I got full expected performance and no freezing/lockup. So it’s definitely something wrong with the framework USB controller or drivers that causes that 100% utilization bug with 0 activity… Hopefully future drivers or bios resolve…
1 Like
I tested the one I ordered in a USB enclosure before I installed it in the laptop and it… sort of worked, but trying to write to it would instantly cause errors. It works fine installed so I suspect it may come down to power limitations, even on the ports that allow more power. Perhaps some NVME drives draw more than USB is really certified to provide.
I have a couple of other 2230 and 2280 drives in enclosures that seem to be working just fine. They’re DRAMless, less storage, and seem to have less power requirements than the one that had issues.
1 Like
This was my USB storage issue investigation (with help and contribution of many community members as seen in the post) regarding the AMD13, which IIRC shares similar USB architecture as the 16 inch.
Maybe worth a read: AMD Framework and NVMe SSD Enclosure Compatibility Investigation
That’s very close to what I have, except mine stays connected and just locks up and lets the windows disk queue build. My lockup issue only happens when booted through it seems, not when in windows or linux on an internal drive OS. I also have no issues with my m.2 Sata enclosure booted through.