DomSmith

Regarding your screen shot; I had a very similar output from one of my test builds, it turned out to be a problem with the 'connection' to the display. My test bench runs through a DP/HDMI KVM switch and sometimes the display is not correctly detected by the display driver. This was resolved by running a good quality DP cable direct from the NAS to the monitor. If you're using a switch or a DP/HDMI converter (or even just a 'bad' cable) I suggest trying a good quality passive/direct cable.

I found that lspci command very useful in my endeavours repurposing thin-clients, might put a post up about it...

Looks like an interesting build. If you're having speed issues you could try a large single file transfer with something like Blu-ray rip and try to see where it's going wrong. I've personal had the following issues: 1 - Bad drivers (Microsoft auto-installed) for the NIC on the windows machine I was transferring from. 2 - Insufficient PCIe bandwidth to the NIC and HBA on the NAS. Use the lspci command from the Linux shell to check for this. 3 - HBA and NIC overheating as they were designed to be aggressively cooled by server chassis fans. 4 - 'cache-less' SSDs that slow down (a lot) under extended write loads. I also found Crystal DiskMark quite useful for this as it isn't dependant on a source disk. You can look at the results and ponder how they might corrospond to things like PCIe link rates etc.

Yeah it looks like the slot on the mobo is a standard m.2 port intended for an NVME drive. I'd guess they've made a 'riser' for this instead of making a custom board, which makes a lot of sense. The CPU has only 9 PCIe lanes running at Gen3 according to this: https://www.intel.com/content/www/us/en/products/sku/231805/intel-core-i3n305-processor-6m-cache-up-to-3-80-ghz/specifications.html After you take off the the lanes required for the I/O, networking etc. there isn't going to be much left for storage. I think I'd also want to verify that there are actually 4 lanes going to that slot to avoid later confusion. You can do this from the linux shell with the lspci command. There are a couple of ways of splitting that signal off to the 4 storage drives, there could be a PCIe switch on the expansion board so each drive would get 4 lanes but but the bandwidth would be shared on a time slot basis, this would be expensive to implement and have little benefit in a NAS setup. I suspect that the expansion board just peels off 1 lane to each of the 4 drives, so each drive will get 1 dedicated lane. If you squint at the traces on the PCB you might be able to see this, otherwise use the lspci command... This was known as 'bifurcation' and required specific support from the CPU, mobo and BIOS on older systems, but I think that's not the case any more...

Ah more clues, what fun... That looks like a sata ssd alright, using a custom header from the mobo. In pics on the product page it looks like there might be 2 of those headers... You can also see in your pic a 'riser' card in the m.2 slot on the mobo, this presumably passes the 1x4 up to the expansion board where it's then presented as 4x1.

Unless I've misunderstood, this just raises more questions for me. There is no mention of SATA connectivity on the product page, so where would you be connecting that? Does the m.2 slot on the board exist? How do the 4 m.2 slots on the expansion board connect? Can you share a link to the teardown video you found please?

https://hub.hexos.com/topic/2026-1st-idea-didnt-work-2nd-has-potential/#comment-12390

Ok I recognise that straight away, mentioned it in someone else's post about getting everything to fit...

The product page does hint to there being an m.2 slot on the motherboard. It is visible in the picture: it also seems to mention it here: I'd guess that you'd be able to put a regular m.2 SSD in here to boot from. I'd also be interested to know if the board even supports booting from the four slots on the top... I'm very interested in the n305 version for another project involving VMs running legacy game servers... ...keen to know how you get on with this. Which CPU version did you get?

Hi @Ioannis, yeah this can seem like quite an issue until you really have a look at what's going on with SATA power. In very loose terms; the original SATA power spec has 12V for larger motors (like those found in 'Desktop' HDDs), 5V for smaller motors (like those found in 'laptop' HDDs) and 3.3V to power the ICs etc. SSDs have no 'high power' motors so there is typically no need for the 12V (or 5V*) line, but they need a lot more 'clean' 3.3V than the interface can deliver (to power all those flash packages). *Most modern consumer grade SSDs, derive their 3.3V supply from the 5V line, and often need nothing else. (This is nothing new, mechanical HDDs have been doing this for some time...) Anyway, before I ramble off tooo far, if you're using a 2.5" SSD you'll most likely find it only needs a 5V supply. Furthermore, because it's using an onboard regulator to bring this down to 3.3V; the 5V supply doesn't need to be exactly 5V for it to work. For 'test purposes' you can pinch a 5V supply from the USB header/port, then if it works out have a look at a better method. Most of these mini-ITX / Thin-ITX boards will include a single SATA or ATX connector, although for some reason they can be the wrong gender just to add to the confusion. If you look at the pics of the Fujitsu Futro S930 build above, you'll see a 'ketchup and mustard' cable coming off the motherboard. This is a mini-MOLEX connector carrying 12V and 5V for this purposes. Whatever you use, be mindful that the 5V supply is coming from somewhere and care must be taken not to overload the source. USB headers are usually pretty safe as they typically have over-current protection. SATA and MOLEX headers are capable of delivering much more power, but the overall power supply for the unit may be less than 35W, so don't be thinking you can run more than a couple of drives without external/supplemental power.

On the bench today I have an old Fujitsu Futro S930, running an AMD GX-424CC (2.4GHz quad core) and 16GB DDR3-1600. Boot pool is an mSATA SSD on the board and a SATA SSD in the 'expansion bay' area, both directly connected to the SOC via the motherboard headers. Storage pool is 4 x SATA SSDs also in the 'expansion bay' under the PCIE riser. HBA (connected to PCIE riser) is a 81Y4494 4-channel SATA 'IT mode' card. 2.5GbE (NGFF) mPCIE card. Power consumption as measured from the outlet: Idle - 15W Load - 32W (Writing @ 2Gbps) Challenges with this one: 1 - Manual BIOS modification required to change the PCIE speed (for NIC and HBA) from gen1 to gen2. For more info, check out this link: https://www.reddit.com/r/homelab/comments/1d722m5/first_homelab_fujitsu_futro_s920/?rdt=65533 2 - Not a lot of space so the SSDs had to be removed from their 'housing' and stacked on m2 pillars (like motherboard standoffs but smaller...) 3 - Powering the SSDs; I had to make up a custom cable for this, splitting off from the ATX header (uses same connector as a FDD) on the motherboard. I cut and soldered together a FDD power cable from a scrap PSU, and some USB Header to SATA cables like these: https://www.ebay.co.uk/itm/205143130740?var=505859057128 I'm currently testing Plex with no issues so far, it's happy trans-coding a 4K BD-'backup' to my phone while also streaming other media to a couple of clients, while being hit with Crystal Disk Mark at the same time. When streaming DVD-'Backups' to 2 clients the power consumption settled down to just 17W. This box is fanless and completely silent, but it needs to be stood on it's end to encourage enough natural convection through it's internals. In particular I noted that the PCIE SATA card gets rather warm, like many others I've seen it's designed to be fitted in a conventional server with quite aggressive case/chassis cooling.

Well, I had to log into the TrueNAS management console to find it but it does appear to be there. Storage / Manage Disks / (select a disk) Edit Disk. There is a 'spindown' option there, will have a play with these and see how they affect power consumption...

Can't disagree with you there, that clicking is not particularly pleasant. However I don't know how Hex handles HDD timeout and spin-down as I don't have any suitable HDDs to test with. Presumably if you only use your NAS periodically and the drives are set to spin-down the power savings from using SSDs would be negligible.

The controller on that particular card is a JMB585 from JMicron. https://www.jmicron.com/products/list/15 I mostly use this card for trouble shooting, drive cloning and data recovery. To he honest it's relatively slow but very consistent, and I've had no problems with it. I'm not reccomending this specific card, but it's worth considering that for NAS applications the NIC is likely to be the performance bottleneck. Therefore you can probably get away with using a cheaper/slower/low-power HBA.

Also have a look at 'modern' 2.5" SSDs as they often share similar internals to their m.2 equivalents, but packaged to fit the legacy HDD form factor. Pictured below are 2 512GB SP-A55 drives. The actual board is tiny and the plastic case is just there to make it fit a 2.5" bay The case could be cut down or removed completely. # Disclaimer: I am not recommending these specific drives. #