obviously the seller doesn't tested and reset password before ship. return.Power button is still not working, vga not working
Any thoughts on that?
obviously the seller doesn't tested and reset password before ship. return.Power button is still not working, vga not working
Any thoughts on that?
btw: how did you connect the supermicro frontpanel cable to the asrock motherboard ? they are not compatible.Thanks for all the help much appreciated
Part number CBL-0084L is the adapter cable.btw: how did you connect the supermicro frontpanel cable to the asrock motherboard ? they are not compatible.
Yeah, the lack of onboard networking is what makes one of these not worth more than $30 imo.~$40 for 2 isn't a terrible deal. The lack of networking is kinda a killer.
And also CBL-0068L , for folks that need something a little longer.Part number CBL-0084L is the adapter cable.
The Gigabyte MJ11-EC1 has a similar downside, you need a PCIe adapter if you want to use any expansion. To date, no EPYC ITX board for a good price without making big compromises.And at that point, you're already paying the extra $30 that these are now overpriced at.
I have seen it once for around 100-120 but I was stupid enough not to grab it :-(Waiting for someone to start dumping MJ11-EC0 boards
Literally the first Result on eBay Germany for 120 EUR, or am I missing something ?I have seen it once for around 100-120 but I was stupid enough not to grab it :-(
That's the EC1.Literally the first Result on eBay Germany for 120 EUR, or am I missing something ?
"It depends"If you posted that its overpriced: is there a source of economical Epyc embedded boards (in the U.S.) that I missed? Or are you saying that, functionally, there's more value in other platforms? Even with the added costs for networking and cabling, these are still hundreds cheaper (at least in the U.S.) than the next cheapest Epyc Embedded option I could find.
That's a really nice piece of kit! I really like ASRock Industrial. The gear I've had from them was well made and lasted forever."It depends"
Are you buying one for Reg ECC support? Then no, I have nothing better in the US. If you aren't buying one for Reg ECC support, then what is the appeal of an embedded EPYC? It's not like you get more PCIe lanes out of this board than a regular Ryzen.
Are you buying one because it's a cheap motherboard with integrated CPU that has ECC? If yes, then consider alternatives like: https://www.ebay.com/itm/188103200167
I had the IMB-V2000M and can confirm it supported ECC. Idle power consumption was not great, ~20W (and no proper IPMI, but it does come with DASH).
They are similar era CPUs except the EPYC has a higher TDP and performs slightly above the V1605B

If you're buying stuff based on impulse budgeting, you've already thrown away considering the worth of something... because now it's just "is this too much to pay for something I find interesting/useful/etc" and not about if it's actually worth what you pay for it.I won't spend hundreds on curiosity, but $50 is under my impulse purchase threshold so I got one. Plus, I already had everything else needed to get it up and running.
Do you have a use case the specifically needs an embedded epyc?If you posted that its overpriced: is there a source of economical Epyc embedded boards (in the U.S.) that I missed? Or are you saying that, functionally, there's more value in other platforms? Even with the added costs for networking and cabling, these are still hundreds cheaper (at least in the U.S.) than the next cheapest Epyc Embedded option I could find.
I'm not sure that anything is unacceptable regardless of price, because like I said originally I think at the original sale price these were just fine... and acceptable is a different metric to different people who buy based on different criteria.I can understand if the networking options (give up the entire PCIe slot or rely on a USB or m.2 adapter), and likely giving up four (or more) PCIe lanes as well, make it unacceptable regardless of price. But, I think that's different than the "bang for your buck".
…which makes it strange that you're going for 2.5G over USB instead of throwing in a server-grade NIC.I'm one of those "only use ZFS with ECC" guys
As for Xeon D-1500 or Atom C3000, you put a NF-A6x25 fan on the heatsink (or whatever fan size fits) with two strips of gaffer tape on the sides and call it a day.I'm curious how people cool these outside of a 1U screaming server box.
No, nothing that specifically required Epyc. As mentioned above, the goal was adding capacity to carry the family through until (hopefully) RAM and Storage prices stabilize a bit and I can undertake a top down upgrade with (much) larger SSDs. I think I'm like a lot of STH members, in that I already have at least some parts on hand that COULD work for a new build. I probably tend to limit my choices more than I need/should based solely on if they work with what I already have: in this case 32GB RDIMMs and connectivity for either 5-6 each of m.2 NVMe and 2.5" SATA SSDs, or 10-12 NVMe by themselves. Maybe the reuse goal itself wasn't the best approach?Do you have a use case the specifically needs an embedded epyc?
I can't think of one myself (especially since amd with ecc is available a lot more than you'd probably think), but if you do absolutely need epyc then sure, it won't be overpriced to you because you're willing to spend for a premium feature.
Haha, yep that's the way it goes.@jackfrost all good points! No matter how much I research hardware purchases, its inevitable that I run across something more appropriate the following week.
While I wouldn't limit myself based on what I already had, I definitely keep in mind if something I already have can be used... I think reusing itself is a great goal, it allows you to be flexible and more cost effective... especially now and for the next few years until SOMETHING happens with chip prices.I probably tend to limit my choices more than I need/should based solely on if they work with what I already have: in this case 32GB RDIMMs and connectivity for either 5-6 each of m.2 NVMe and 2.5" SATA SSDs, or 10-12 NVMe by themselves. Maybe the reuse goal itself wasn't the best approach?