They tend to be a bit unstable, if not outright unsupported, in several of the major distributions. VMware has been dropping support for them in newer versions and pfSense recommends against them.
Because FreeBSD's Realtek NIC drivers are pretty awful. So anything FreeBSD based, isn't going to work well, network wise. Linux and even Windows have much better Realtek NIC drivers. Windows just works, in Linux you can get away with turning off a few offloading features.
Considering Realtek has more marketshare in integration than every other manufacture combined, if anyone is dropping support for them, their priorities are wrong. That isn't Realtek's fault. That isn't even Realtek's problem; they seem to be doing just fine without these niche software vendors flailing support.
One generally picks a software platform first I am afraid, based on in-house expertise and such... Linux might indeed be by far the dominant player in this market but freebsd and its derivatives are not irrelevant. realtek should think twice before allowing others to come in and do better... because that is what happens if you screw up for too long.
Full disclaimer: I am not an expert on Linux, having mostly stayed with Windows and Apple. However, as I understand it, the responsibility for the absence of good Linux or freeBSD drivers for Realtek NICs probably lies with - Realtek. If a hardware manufacturer wants to make sure that their kit is well-supported by Linux or freeBSD, they contribute to the next distro by writing and submitting the drivers in question. Intel does it, AMD does it, NVIDIA does it, ARM does it etc. I think it's a case of Realtek figuring they don't need that market, and it's not worth their effort. If that's correct, than the selection of Realtek's NICs for these is a shame, as it will limit their use.
The STX form factor is a great place for Ryzen, but I'm guessing the price is going to be rather high since this is an embedded board. Consumer availability might be rather iffy too, but it's a step in the right direction.
Would have liked to see HDMI and CIR, would have been a nice alternative for NUCs in the HTPC arena. The 3.5mm jacks on the front are... confusing. Is one for headphones and the other for a microphone instead of a combo? If so, why are they on either side of the USB ports instead of right next to each other? Also, where's the fourth USB Type A port?
The audio jack on the far left of the bottom image appears to have the word "MIC" next to it. My old SunRay thin client and a number of other older models featured front 3.5mm audio jacks for headphones and microphones. Since this is an embedded board, I'm betting that thin client use is one of the intended targets, though it seems like quite a bit of CPU and GPU grunt for such a system.
Aww dang, I was hoping to use something like this to build a pfsense appliance. Hopefully PC-Engines will step up and make an APU3 with an embedded Ryzen APU and Intel NICs.
How does basic power on/off management on these boards work?
I only see a 4pin Molex connection to the board. If that's 12V in it could power the whole board with DC-DC converters making 3.3/5V for components that need them. but that'd leave the soft power on/off functions that are a few wires in the 24 pin atx plug. Are those carried by a separate header, or is that logic in the hands of the mini-STX enclosure instead?
It's got a barrel socket next to the 4 pin connector as well. Probably for external 12V or maybe 19V power supply. The 4 pin is likely for when you want to use an internal ATX/SFX... PSU. And all the necessary logic should be onboard the motherboard. The 10 pin header just to the left of the RAM slots is likely the normal front panel header with power button, reset button and activity/power LEDs. It has another header with a jumper labeled "auto power on", might be something to tweak the power on setting a bit. Unfortunately, while the sapphire website for this board is up, it only has the specs and some drivers packages, I didt not see any manual so far. :)
IF it's being powered via a barrel then the 4 pin molex could be 12V to the rest of the chassis. It is not sufficient to let the mobo control an ATX PSU. To control an ATX PSU you need the Green and Gray wires in the 20/24 pin connector to allow the mobo to turn the rest of the PSU on/off and the purple wire to provide standby power to energize the circuit that waits for the power button to be pressed and then signals to the PSU to turn everything else on. Theoretically you could send standby power via the 12V bus (and I suspect the OEM systems with 12V only PSUs may do so); but you're outside of ATX if you do so.
@DanNeely: it is reasonably common for these types of industrial PCs to always expect 12V supply - most of them are going to be "always on". They don't care about complying with the ATX standard - it's just a convenient keyed connector for which cheap off-the-shelf components already exist. If you want to switch power, you usually use an upstream device to do so. That way you can use the second (battery backed) device as an external watchdog, allowing a full power-cycle when your system inevitably goes belly up.
i'm pretty happy that these are SODIMMs where i could buy 2 2GB sticks... my 2400G is being used for distributed computing and i'm not even fully using the RAM capacity of the current 1x4GB stick
I don’t get it. I read that zen has 10Gb Ethernet on the die, yet absolutely everything I’ve seen uses Realtek or Intel to provide Ethernet. Why is that?
I just wish they'd release a version of these boards that actually has a audio good enough to use in a little miniature home theater setup. That audio quality with the same is not going to be very good
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
28 Comments
Back to Article
ERJ - Monday, July 9, 2018 - link
Too bad they are using Realtek ethernet. If they used Intel it would make for a great little firewall or even small lab VM server.jordanclock - Monday, July 9, 2018 - link
What's wrong with Realtek for those purposes?ERJ - Monday, July 9, 2018 - link
They tend to be a bit unstable, if not outright unsupported, in several of the major distributions. VMware has been dropping support for them in newer versions and pfSense recommends against them.ky41083 - Monday, July 9, 2018 - link
Because FreeBSD's Realtek NIC drivers are pretty awful. So anything FreeBSD based, isn't going to work well, network wise. Linux and even Windows have much better Realtek NIC drivers. Windows just works, in Linux you can get away with turning off a few offloading features.Samus - Monday, July 9, 2018 - link
Considering Realtek has more marketshare in integration than every other manufacture combined, if anyone is dropping support for them, their priorities are wrong. That isn't Realtek's fault. That isn't even Realtek's problem; they seem to be doing just fine without these niche software vendors flailing support.jospoortvliet - Tuesday, July 10, 2018 - link
One generally picks a software platform first I am afraid, based on in-house expertise and such... Linux might indeed be by far the dominant player in this market but freebsd and its derivatives are not irrelevant. realtek should think twice before allowing others to come in and do better... because that is what happens if you screw up for too long.eastcoast_pete - Saturday, July 14, 2018 - link
Full disclaimer: I am not an expert on Linux, having mostly stayed with Windows and Apple. However, as I understand it, the responsibility for the absence of good Linux or freeBSD drivers for Realtek NICs probably lies with - Realtek. If a hardware manufacturer wants to make sure that their kit is well-supported by Linux or freeBSD, they contribute to the next distro by writing and submitting the drivers in question. Intel does it, AMD does it, NVIDIA does it, ARM does it etc. I think it's a case of Realtek figuring they don't need that market, and it's not worth their effort. If that's correct, than the selection of Realtek's NICs for these is a shame, as it will limit their use.sor - Monday, July 9, 2018 - link
I’ve been waiting what seems to be an eternity for the 2400G based ZBOX MA551 that was shown in 2017. Maybe this V1807B would fit the bill.watlina - Monday, July 9, 2018 - link
Very Similar to the UDOO Bolthttps://www.kickstarter.com/projects/udoo/udoo-bol...
ayqazi - Monday, July 9, 2018 - link
Except no Arduino-like functionalityPeachNCream - Monday, July 9, 2018 - link
The STX form factor is a great place for Ryzen, but I'm guessing the price is going to be rather high since this is an embedded board. Consumer availability might be rather iffy too, but it's a step in the right direction.Xajel - Tuesday, July 10, 2018 - link
The top of range version cost about $450, of course you still need RAM, storage and so on.Sergio526 - Monday, July 9, 2018 - link
Would have liked to see HDMI and CIR, would have been a nice alternative for NUCs in the HTPC arena. The 3.5mm jacks on the front are... confusing. Is one for headphones and the other for a microphone instead of a combo? If so, why are they on either side of the USB ports instead of right next to each other? Also, where's the fourth USB Type A port?PeachNCream - Monday, July 9, 2018 - link
The audio jack on the far left of the bottom image appears to have the word "MIC" next to it. My old SunRay thin client and a number of other older models featured front 3.5mm audio jacks for headphones and microphones. Since this is an embedded board, I'm betting that thin client use is one of the intended targets, though it seems like quite a bit of CPU and GPU grunt for such a system.Sergio526 - Monday, July 9, 2018 - link
You're right, it does say MIC and other says Line Out. Still really weird that they put two USB ports between them.r3loaded - Monday, July 9, 2018 - link
> enabled by Realtek’s RTL8111G controllersAww dang, I was hoping to use something like this to build a pfsense appliance. Hopefully PC-Engines will step up and make an APU3 with an embedded Ryzen APU and Intel NICs.
Zok - Monday, July 9, 2018 - link
Is that a MicroSD slot between the M.2 and CPU?DanNeely - Monday, July 9, 2018 - link
How does basic power on/off management on these boards work?I only see a 4pin Molex connection to the board. If that's 12V in it could power the whole board with DC-DC converters making 3.3/5V for components that need them. but that'd leave the soft power on/off functions that are a few wires in the 24 pin atx plug. Are those carried by a separate header, or is that logic in the hands of the mini-STX enclosure instead?
Death666Angel - Monday, July 9, 2018 - link
It's got a barrel socket next to the 4 pin connector as well. Probably for external 12V or maybe 19V power supply. The 4 pin is likely for when you want to use an internal ATX/SFX... PSU. And all the necessary logic should be onboard the motherboard. The 10 pin header just to the left of the RAM slots is likely the normal front panel header with power button, reset button and activity/power LEDs. It has another header with a jumper labeled "auto power on", might be something to tweak the power on setting a bit. Unfortunately, while the sapphire website for this board is up, it only has the specs and some drivers packages, I didt not see any manual so far. :)DanNeely - Tuesday, July 10, 2018 - link
IF it's being powered via a barrel then the 4 pin molex could be 12V to the rest of the chassis. It is not sufficient to let the mobo control an ATX PSU. To control an ATX PSU you need the Green and Gray wires in the 20/24 pin connector to allow the mobo to turn the rest of the PSU on/off and the purple wire to provide standby power to energize the circuit that waits for the power button to be pressed and then signals to the PSU to turn everything else on. Theoretically you could send standby power via the 12V bus (and I suspect the OEM systems with 12V only PSUs may do so); but you're outside of ATX if you do so.overzealot - Tuesday, July 10, 2018 - link
@DanNeely: it is reasonably common for these types of industrial PCs to always expect 12V supply - most of them are going to be "always on".They don't care about complying with the ATX standard - it's just a convenient keyed connector for which cheap off-the-shelf components already exist.
If you want to switch power, you usually use an upstream device to do so. That way you can use the second (battery backed) device as an external watchdog, allowing a full power-cycle when your system inevitably goes belly up.
mode_13h - Monday, July 9, 2018 - link
Sweet! These chips seem to support ECC - I hope the board does, too.https://www.amd.com/Documents/V1000-Family-Product...
grahad - Tuesday, July 10, 2018 - link
2x DDR4 ECC/Non-ECC SO-DIMMsDual Channel speed up to DDR4-3200 (configuration dependent)
http://www.sapphiretech.com/productdetial.asp?pid=...
plonk420 - Tuesday, July 10, 2018 - link
i'm pretty happy that these are SODIMMs where i could buy 2 2GB sticks... my 2400G is being used for distributed computing and i'm not even fully using the RAM capacity of the current 1x4GB stickgrahad - Tuesday, July 10, 2018 - link
Pricing per email (and for small quantities) is:FS-FP5V1807B V1807B 35-54W 52093-00-40G - $450
FS-FP5V1756B V1756B 35-54W 52093-01-40G - $390
FS-FP5V1605B V1605B 12-25W 52093-02-40G - $340
FS-FP5V1202B V1202B 12-25W 52093-03-40G - $325
latentexistence - Tuesday, July 10, 2018 - link
I don’t get it. I read that zen has 10Gb Ethernet on the die, yet absolutely everything I’ve seen uses Realtek or Intel to provide Ethernet. Why is that?HideOut - Tuesday, July 10, 2018 - link
I just wish they'd release a version of these boards that actually has a audio good enough to use in a little miniature home theater setup. That audio quality with the same is not going to be very goodn13L5 - Sunday, August 5, 2018 - link
Meh, I want a Mini-STX board for Ryzen chips of my choice - not some embedded stuff for the IoT crowd.IoT will become hell on earth, once its all up and going - cause it'll be misused to the hilt - and not just by the Chinese government...