climateprediction.net home page
Hardware for new models.

Hardware for new models.

Message boards : Number crunching : Hardware for new models.
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 3 · 4 · 5 · 6

AuthorMessage
SolarSyonyk

Send message
Joined: 7 Sep 16
Posts: 259
Credit: 32,069,627
RAC: 28,154
Message 67955 - Posted: 21 Jan 2023, 22:19:04 UTC - in response to Message 67944.  

And they don't lie about the current it will take like most people do. The standards are just plain wrong and very dangerous.


What do you mean, "lie about the current it will take"? Either it is, or it isn't, the specified cross sectional area of copper, or whatever else it claims to be.

4AWG is .2043 inches in diameter, for a cross sectional area of .03278 square inches. If it's not that, it's not 4AWG. Anyway, we're clearly either talking past each other or both being pig-headed here, so I'll drop it.

So? And that doesn't make sense anyway, if you use a cable twice as thick but only put half the strands into the connector, you're better off than if you'd used the thinner cable and left all the strands in, since the vast majority of the length is thicker, and a tiny bit is the same!


And now you've violated all assumptions made about thermal dissipation and such, especially in the critical wire-to-connector area. It's not a good idea. Fine, it can be done, and it's probably OK, but it's still sloppy, hack practices.

Think, don't blindly follow codes (whatever they are, we don't have such nonsense in the UK).


Doesn't help when the inspector asks what, exactly, you thought you were doing. You seriously have no national electric code or such in the UK and can "just do whatever" for home wiring? M'kay.

What would you be plugging in which draws 50% of the power you can get for your whole house?


Most US homes (at least the pure electric ones) are 200A service, so 25%. They're good for 40A continuous, and EV chargers, welders, air compressors, etc... any "Big 240V appliance with a plug" is almost certain to use that connector, because it's the standard, even if it's not needed.

Er... didn't that break your precious code? You're using a connector which can't handle what it's meant to?


Wrong way. The inverter can source 25A sustained at 240V, and the wiring and plug are rated for 50A. There's no hazard in doing that, it just means that something that needs 40A won't run - it'll trip the inverter off. You can put a larger outlet on a smaller breaker, but you can't put a larger breaker on a smaller outlet. If I wired up a 15A 120V plug to a 30A breaker, that would be a problem. Putting a 20A outlet on a 15A breaker isn't the issue. The goal is fire prevention, and "not overheating the weakest link" is the design criteria.

And who needs more RAM?


My 12C/24T/16GB boxes aren't too happy right now...

Here is an Intel Broadwell-EP (host 1534812) which has got a bunch of "double free or corruption (out)" too.


Ah, OK. I'd thought it was limited to just AMD, but I guess not.
ID: 67955 · Report as offensive     Reply Quote
Glenn Carver

Send message
Joined: 29 Oct 17
Posts: 815
Credit: 13,667,305
RAC: 8,252
Message 68269 - Posted: 11 Feb 2023, 22:10:45 UTC - in response to Message 67950.  
Last modified: 11 Feb 2023, 22:11:25 UTC

SolarSyonyk wrote:
I guess avoid AMD, that double free thing is being a royal pain on them for some reason, though it's happening on enough different boxes (including cloud systems I know very well are ECC) that I don't believe it's RAM errors, just something in whatever code path AMD chips end up on.
Here is an Intel Broadwell-EP (host 1534812) which has got a bunch of "double free or corruption (out)" too. (In this host's list of oifs_43r3_ps results with error status, look at those which finished on January 12 and took less than 70,000 seconds.)
The double free error affected all machines, Intel Core/Xeon, AMD. It occurred less often if the machine was less loaded. I suspect the AMDs showed up more because they are more common & maybe folk were making full use of the often higher core/thread count. Most of these errors have now been fixed, I'm still working on another one. But it should happen less often with the next batches.

Anyway, what matters to OpenIFS is single core speed & memory bandwidth (L3 cache size is not important). Choose your hardware accordingly... ;)
ID: 68269 · Report as offensive     Reply Quote
Mr. P Hucker

Send message
Joined: 9 Oct 20
Posts: 690
Credit: 4,391,754
RAC: 6,918
Message 68273 - Posted: 12 Feb 2023, 15:48:18 UTC - in response to Message 67955.  

And they don't lie about the current it will take like most people do. The standards are just plain wrong and very dangerous.
What do you mean, "lie about the current it will take"? Either it is, or it isn't, the specified cross sectional area of copper, or whatever else it claims to be.
The specified area of copper is not sufficient to take the current the standards say it will. It make the wire heat up and drop voltage. Always get a bigger one than you're meant to.

4AWG is .2043 inches in diameter, for a cross sectional area of .03278 square inches.
I'm so glad I don't use inches. mm^2 is much easier, in fact 1 mm^2 is very roughly 10 amps max. That's the getting warm number of amps I avoid.

If it's not that, it's not 4AWG. Anyway, we're clearly either talking past each other or both being pig-headed here, so I'll drop it.
The one I bought is decent 4AWG, or what you would call 0AWG. I prefer amps to Chinese amps.

And now you've violated all assumptions made about thermal dissipation and such, especially in the critical wire-to-connector area. It's not a good idea. Fine, it can be done, and it's probably OK, but it's still sloppy, hack practices.
What?! Making the wire thicker along the entire length except where it goes into the connector is obviously going to create less heat and drop less voltage.

Doesn't help when the inspector asks what, exactly, you thought you were doing. You seriously have no national electric code or such in the UK and can "just do whatever" for home wiring? M'kay.
It's my house, I do what I like. Do you actually regularly let the police or whoever come in to check on your house? I'd expect that in a communist country. Land of the free indeed.

Most US homes (at least the pure electric ones) are 200A service, so 25%.
At 120V or 240V?

They're good for 40A continuous, and EV chargers, welders, air compressors, etc... any "Big 240V appliance with a plug" is almost certain to use that connector, because it's the standard, even if it's not needed.
I can't think of anything which could use 40A 240V. That is a lot of power. My neighbour has a welder which uses 13A 240V. Plugs into the same standard socket all over anyone's house. One standard, much easier.

Wrong way. The inverter can source 25A sustained at 240V, and the wiring and plug are rated for 50A. There's no hazard in doing that, it just means that something that needs 40A won't run - it'll trip the inverter off.
Ah, I think I thought you meant you had a 50A outlet on a wire which could provide less than that.

You can put a larger outlet on a smaller breaker, but you can't put a larger breaker on a smaller outlet. If I wired up a 15A 120V plug to a 30A breaker, that would be a problem. Putting a 20A outlet on a 15A breaker isn't the issue. The goal is fire prevention, and "not overheating the weakest link" is the design criteria.
I dread to think how complex your wiring is, especially the breakers. A 240V breaker has to somehow measure the current in both legs. Then try to monitor earth leakage on that too. I can't even begin to think how that would work without a computer inside it to add everything up.

My 12C/24T/16GB boxes aren't too happy right now...
Ouch. I've got 64GB in everything except the antiques which only take 8, but they only have 4 cores. I'll be going to 128GB when I can afford it.
ID: 68273 · Report as offensive     Reply Quote
Previous · 1 . . . 3 · 4 · 5 · 6

Message boards : Number crunching : Hardware for new models.

©2024 climateprediction.net