I'm pleased to say that despite the rain of the last couple of days, the Zigbee sensor is showing acceptable levels of humidity from inside the box. I think the small but warming wattage being consumed by the three items in there is helping.

Perhaps it's time we cloned Phyllis from the material.

I did probably have the volume a bit high for something that's supposed to be background. Especially given the percussiveness and repetitively of the track in question.

You're going to find that many of the truths we cling to depend greatly on our own point of view.

And, what orientation should one use when deploying in the Southern Hemisphere?

Whatever happened to Quenchers?

The original "Get It From Source" source relates to a small coppice on the map I had in the game Autonauts during my first playthrough. Sometimes you cannot wait for automata to do their job, and you have to get it from source yourself.

The era of Xeon E5v4 is not a good choice for low power use. Any of the 16+ core chips were ≥120W, and so that's already a bad starting point for a dual CPU machine running 100% load 24×7. Pair that with the fact these Chinese X99 boards generally implement poor power management, and you're going to see a better value return on your power very quickly with a more modern machine.

@@andrewsturmey Agreed and certainly not with a Milliband on the scene. He alone doubled the price of a kWh in a week. Great fun building and experimenting though.

Let's be honest, this isn't a sensible build. Going with an infinity mirror front panel instead of a mesh is definitely not a choice of practicality. The front fans pull in a small but adequate amount of air via the tiny side grille space on the front panel, but more importantly they give home to blue LEDs! I can confirm both onboard NVMe slots are full direct-to-CPU x4 PCIe lanes, which is nice. And I guess with a full four x16 slots of the same, you could in principle go with 18 total NVMe drives in a NAS. But yes, if you don't want an all-flash NAS, because you want higher capacities or lower cost-per-GB, front bay disks would work better. I can confirm indeed that virtualisation is an excellent use-case on this box. I cannot fault it, 20 months later; very reliable and pretty good performance.

Notable. Although I've been playing Civ games since the early nineties, I never really got into multiplayer much. I have occasionally had embarrassing moments where I've used Cheat Engine in small indie single-player games, only to discover later that they have scoreboards, which I'm now at the top of by orders of magnitude.

Lol ya i didn't think it was you. He was using a method to get more goodie huts and locate city states faster in the early game which gave him a big advantage. How was civ in the 90s? Fun?

I think the experience of Civ 1 can be summarised in one very familiar phrase... "Just one more turn"

I didn't have any CityFibre services prior to getting Yayzi; my other BT connection is VDSL based. Based on what I was offered, then I assume this postcode area is non-XGS and limited to ≤2Gb/s services. My assumption is that they wouldn't entrust your average person with swapping out something they screw to the wall and that has a dust-sensitive connector, but don't know for sure. I still haven't opened the parcel the router came in!

I have plugged in both EPS power connectors, but I can confirm it's not electrically necessary, if the PSU can deliver enough current on a single connector. I'd suggest both, if you have enough PSU connectors and cables.

You appear to have beaten both regulation and unlimited responses. Come to Zombieland immediately. As someone who operates a fuchsaw when doused with hard adult beverages these days, I would say it is worth a receive, but not an imbibe.

On the topic of stability, then all of my three builds; being the Jiahuayu, the Machinist, and the Huananzhi; have been extremely good. One runs ESXi with a mix of BSD, Linux, and Windows VMs, including one TrueNAS VM with PCIe-passthrough of 6 NVMe. One of the other two has been used for heavy FFmpeg transcode work, and the other a daily-use remote Windows desktop for software development. All very stable, and in line with the performance expectations of that era's Xeons. However... idle power draw is not as good a story. They all run a bit hungry, seemingly not implementing much by way of power state management. I think over the long term, ongoing electricity costs might make these Chinese boards less compelling.

Almost all was new. The motherboard bundle, although coming with a brand new board, did use refurbished processors and memory.

@@andrewsturmey Right, get that. Thanks for sharing.

The case is a GameMax Abyss. I think any EATX form-factor case would be fine, just depends if you care about aesthetics or cooling performance more. This one isn't exceptional for cooling, but plenty for me in the environment I have it, with plentiful case fans.

Yes. It's still working well. No hardware problems to speak of. I also upgraded it since the video with 2.5Gb/s ethernet.

@@andrewsturmey Thank you for replying to my message. I was also looking at the Huananzhi F8D Plus as they look very similar to the Machinist dual socket motherboard you are using. Just out of curiosity would you happen to know how many VRM mosfets per socket are on your motherboard? I can’t seem to find a clear enough picture to see for my self. They would be the little black squares just in front of the little motherboard heatsinks at the top and between the socket and heat sink.

I'm sorry, I've tried to have a look, but no luck, unfortunately. I got a stepladder out to see if I could get above it, and take some photographs through the side panel, but the heatsinks and fans are really big and obscure everything. I cannot readily shut the machine down to get it out either. I could see some mosfets and inductors, but no way I'm going to be able to give you an answer with any confidence. I am curious as to what you would consider a good number of them. I'd also be inclined to think that it remains a bit of a lottery anyway, with the quality considerations of the component sourcing a lot of these Chinese manufacturers do.

Thank you for reaching out. This sounds very promising. I notice the prefix announcement to me is unchanged, and as of right now, I still don't seem to get replies when I ping out over IPv6. Is there anything I need to do now it's enabled, before it will work?

Seemingly still not seeing return traffic. If you're sure IPv6 should be working for everyone by now, I might think about putting in a support email.

​@@andrewsturmey please do and we can take a look for you 🙂

I can confirm I now have IPv6 connectivity! Since my initial setup, the RA messages stopped advertising an on-link prefix; and rather counterintuitively, they also lack either the managed/other flag, which would cause the client to send a DHCPv6 request. However, if you override normal behaviour, as I did by using systemd-networkd's parameter "WithoutRA=solicit", to send a DHCPv6 request anyway in a non-standard manner, then you do get a prefix delegation back. Taking a /64 from the delegated /60 and using it internally, everything works brilliantly.

At least you didn't get stuck in concrete. Wood is confirmed.

Any chance you're trying to do this with a Windows 10/11 edition that is license limited to one socket? Most times people report this it's because they're not running the Pro edition. You could also boot of a Linux live image to check it sees both sockets.

@@andrewsturmeyThanks a lot . upgraded to pro and problem solved 😊

@@andrewsturmey that was it thank you so much : =) just doubled my performance after 7 months

I am pleased to hear you have enjoyed success. I'm interested to know what software you're running that enjoys scaling well across a very parallel core count.

@@andrewsturmey I am gaming 😎

I was busy talking to the switched-off camera when the firmware update finished, so I didn't really notice how long the first boot was that time, but after I cleared the CMOS, I did certainly notice some major waiting time.

I had it on my Jiahuayu board. For me, this occurred when I was using unregistered non-ECC memory. As soon as I swapped it all out for ECC, it worked. No idea if this might be relevant to your scenario.

Thanks for the info I will order some

Did both cpus work now with ecc memory?

The fabric oat tastes oaky and feels chewy. It doesn't have a nose.

Yes, and yes. You should have no problems with either of those requirements.

I recall, back when we both worked for C4L; it's probably over 15 years ago now, that we deployed IPv6 network wide. I remember thinking even then that we were a little late to the game.

It is indeed a non-sensical statement by them. Perhaps they're thinking you could get a router and terminate a tunneled IPv6 service on it. My experience of “affordable” tunneled IPv6 services is that 2Gb/s would be an absolute pipedream. I do already have a GRETAP-over-WireGuard tunnel set up to a 10Gb/s-connected server I have, so maybe I should just go ahead and set up an IPv6 egress from there and be done with it.

The power supply I'm using does have 2 CPU power outputs. From what I have seen of most Chinese boards that do have 2 CPU power connectors, they're wired in such a way you only have to connect 1 of them. But do consider in this case that this does mean a lot of current going via a single output, which at full load might be a problem for some power supplies.

A datacentre larger than the Inex Sipoo, Finland.