Prototype of RTX 5090 Appears With Four 16-Pin Power Connectors, Capable of Delivering 2,400W
Technology
67
Beiträge
41
Kommentatoren
300
Aufrufe
-
when you do need power you need a special circuit.
We also have a standard socket and a high power socket.
Expect our normal outlets provide 230V 16A 3.5kW (3kW sustained) and the typical high power outlets outlets provide 400V 30A 11kW or 400V 60A 21kW.
Which is why typical electric stoves here use 11kW and typical instant water heaters use 21kW.
Though probably the most noticeable advantage is in electric car charging.
Yeah, in Sweden I charge our plug in hybrid off 240, it's pretty quick and you can use any outlet.
The giant round connectors are weird BTW, with all the holes, trying to sort that out for faster charging.
I don't think we should run 100+ volts everywhere, we need to standardize on lvdc in most places (basically usb-c or so) with 100v only in kitchens and places you need it, because it's more dangerous and can cause fires more easily.
-
That's why I started my sentence with "In Europe"
But I live in America so naturally you're referring to US outlets, right??
-
Yeah, in Sweden I charge our plug in hybrid off 240, it's pretty quick and you can use any outlet.
The giant round connectors are weird BTW, with all the holes, trying to sort that out for faster charging.
I don't think we should run 100+ volts everywhere, we need to standardize on lvdc in most places (basically usb-c or so) with 100v only in kitchens and places you need it, because it's more dangerous and can cause fires more easily.
That's a common misconception. It's the Amps that cause fires, not the voltage.
The 5090 uses 600W, at 12V that's 50A, but at 120V that'd only be 5A and at 240V only 2.5A.
50A melts cables and burns your PC down, 2.5A won't. The only risk of higher voltages is that they can jump across small air gaps much easier.
-
Amps are voltage over resistance (I = V/R), volts absolutely matter, the human body has a decent resistance and the higher voltage helps burn through that.
There's a reason we talk about lethal current and not lethal voltage...30mA can kill you, even at something ridiculously low as 9V, but 5-10kV will not necessarily kill you, e.g. fences for horses will not kill you if you're electrocuted by them because there's basically no amperage. Voltage is not the determining factor in lethalness.
-
You would still need to run the computer off multiple plugs, as almost any 240v plug is 10a.
You'd likely need a dedicated breaker and plug, similar to a stove plug.
Here, plugs are 230V and 16A = 3680W. Not quite as much as I thought (most extension cords seem to be rated for a bit more, which makes sense), but definitely enough to run monitors of the same breaker.
-
That's a common misconception. It's the Amps that cause fires, not the voltage.
The 5090 uses 600W, at 12V that's 50A, but at 120V that'd only be 5A and at 240V only 2.5A.
50A melts cables and burns your PC down, 2.5A won't. The only risk of higher voltages is that they can jump across small air gaps much easier.
No it's not, I'm an ee.
P = I^2R, so power squares against the current, while it's linear to voltage.
This means current causes more heat dissipation in the wire, which has risks, potentially fire if you really go too far, this is why breakers trip.
But what really causes fires (again, outside of crazy overcurrent) is Arcing, from basically either bad connections or bad insulation, OR, from an inductive load that gets disconnected, so the current tries to stay constant in the coils, which leads to massive voltage spikes.
-
There's a reason we talk about lethal current and not lethal voltage...30mA can kill you, even at something ridiculously low as 9V, but 5-10kV will not necessarily kill you, e.g. fences for horses will not kill you if you're electrocuted by them because there's basically no amperage. Voltage is not the determining factor in lethalness.
There is no current without voltage, what you're saying makes no sense.
Current is voltage over resistance, you get as much current as you have voltage, unless there's an artificial effect limiting voltage, like a voltage regulator or zener diode or just fet.
When you say 'it's the current', that electric fense has x volts before you touch it, and the fact that it doesn't kill you means either the voltage is too low to produce a decent current in your body, or, there's a voltage regulator/limiter that means when you touch it the voltage drops to some lower level, which I could calculate using the nominal resistance of the human body and the voltage of the fence.
In a way, the output impedance of whatever is driving the fence determines how much the voltage drops under load.
I'm speaking about the voltage needed to get the deadly current across the critical areas of your body BTW, which can be handled as a kirchoff circuit I'm sure.
-
But I live in America so naturally you're referring to US outlets, right??
Yes, your are right, here is a map of Europe where you have chad power plugs in a-Murika (outside these towns only virgin plugs):
-
No it's not, I'm an ee.
P = I^2R, so power squares against the current, while it's linear to voltage.
This means current causes more heat dissipation in the wire, which has risks, potentially fire if you really go too far, this is why breakers trip.
But what really causes fires (again, outside of crazy overcurrent) is Arcing, from basically either bad connections or bad insulation, OR, from an inductive load that gets disconnected, so the current tries to stay constant in the coils, which leads to massive voltage spikes.
The recent technology connections video cited a lot of statistics on this topic, and at least household fires are primarily caused by overcurrent, not by arcing.
You probably know more than me — I only studied compsci with ee as minor — but from my personal experience, I've seen many cases where overcurrent caused damage, burns or fire, but I can't remember a single case where arcing caused actual damage.
Even in cheap chinesium powerstrips, the primary cause of fires is overcurrent due to AWG 22 copper clad iron wire, not arcing. (Though the switches usually weld themselves together after a few dozen uses).
-
That's why I started my sentence with "In Europe"
God forbid I supply information. I'm fucking done commenting with all you over sensitive weirdos who think everything is a fucking argument.
-
There's a reason we talk about lethal current and not lethal voltage...30mA can kill you, even at something ridiculously low as 9V, but 5-10kV will not necessarily kill you, e.g. fences for horses will not kill you if you're electrocuted by them because there's basically no amperage. Voltage is not the determining factor in lethalness.
In most household shocks, you touch a conductor, and you are the resistor to ground. Your resistance is independent of the drive voltage, so if you touch a 110V wire, the current will be half of what you get with a 220V wire. So the voltage determines the current, and thus the lethality.
There’s lots of other factors that go into the effective resistance like the amount of moisture on your skin, what shoes you’re wearing, and what the floor is made of, etc, but in all cases twice as much voltage will cause twice as much current. You are by far the highest resistance element in the circuit, so your resistance will completely determine the current - most household circuits are capable of supplying 10-15A continuously, so your resistance is the current limiter.
It’s a bad idea either to go touching live wires either way, but the rule of thumb I heard was was that a 110V shock usually won’t kill you and 220V shock usually will.
-
In most household shocks, you touch a conductor, and you are the resistor to ground. Your resistance is independent of the drive voltage, so if you touch a 110V wire, the current will be half of what you get with a 220V wire. So the voltage determines the current, and thus the lethality.
There’s lots of other factors that go into the effective resistance like the amount of moisture on your skin, what shoes you’re wearing, and what the floor is made of, etc, but in all cases twice as much voltage will cause twice as much current. You are by far the highest resistance element in the circuit, so your resistance will completely determine the current - most household circuits are capable of supplying 10-15A continuously, so your resistance is the current limiter.
It’s a bad idea either to go touching live wires either way, but the rule of thumb I heard was was that a 110V shock usually won’t kill you and 220V shock usually will.
It’s a bad idea either to go touching live wires either way, but the rule of thumb I heard was was that a 110V shock usually won’t kill you and 220V shock usually will.
That's completely incorrect though. I've been shocked by 230VAC at least a dozen times, if not more. And the fuse for the circuit absolutely should not be the limiter, the RCCB should trip WAY before the main fuse.
-
Almost has to be. 2400W would put it completely outside the consumer market. Consumer PSUs don't go that high. Home power outlets don't go that high unless you have special electrical work done. I can hardly imagine what a cooling system for a nearly 3KW system would look like.
nVidia cares less and less about the consumer market every year. We basically only exist to buy the factory fourths so that the overall yield of any given wafer can be maximized.
2400 for a single component is still rather insane even by server room standards. But 12 or even 18 load balanced? That starts to "make sense" for higher end data centers or even on-prem server rooms at the more tech oriented companies.
-
In Europe, this is no biggie
I just saw a reputable 2400W kettle on a random online store for 50€
Looks like there are 3000W options too
I think the typical limit is around 3600W, with 16A at 230V
-
This post did not contain any content.
Good luck plugging that into an outlet without tripping a fuse under load.
-
If they put 2x 12pin HighPower connectors and they wouldn't be burning up because each would be delivering just 300W. But they explicitely don't allow board partners to do it themselves, because NVIDIA is bunch of controlling assholes.
They could even used it as part of marketing. “So powerful it needs 2 connectors”
-
Good luck plugging that into an outlet without tripping a fuse under load.
10A. That's fine.
-
This post did not contain any content.
Christ, not exactly a model of power efficiency is it?
Also, if it's drawing that much power, how could it possibly dissipate all the heat? It must sound like an F-16.
I expect this card will be a hard pass from me...
-
You would still need to run the computer off multiple plugs, as almost any 240v plug is 10a.
You'd likely need a dedicated breaker and plug, similar to a stove plug.
All UK plugs are 13A.
-
10A. That's fine.
I'm not trying to be a 1-upper, but you need to also include the power the computer and CPU use as well. Not to mention the age of the outlet and how many times a plug has been inserted/removed from it. The contact resistance can be pretty bad depending on how old the outlet is.
