Live data from Hacker News

Up to 600 watts of power for graphics cards with the new PCIe 5.0 powerconnector

igorslab.de

151–160 of 182 posts

Re: Up to 600 watts of power for graphics cards with the new PCIe 5.0 powerconnector

#151
post #10

Assuming you can install two GPUs using 600W, that's basically running a hair dryer in your desktop since most of the energy is converted to heat.

The RTX 3080 already uses 320W (peak?). By next GPU generation who knows.

Good lord, just 2 generation back GPUs like the 1070Ti consumed 180W. Even if you add a CPU (of that era) like the i7-7700 with it 65W draw, that just 245W. Now a video card alone can draw more power.

Re: Up to 600 watts of power for graphics cards with the new PCIe 5.0 powerconnector

#152

That looks all the world like a Micro-Fit 3.0 drawing, whose contacts are normally rated at 8.5A, but here they're claiming 9.2A each. Also, the largest Micro-Fit terminal is made for 18AWG, but the article mentions 16. https://www.molex.com/molex/products/family/microfit_30?pare... I suspect what's happening is Molex is making a new version of the terminal with higher clamping force or better plating, to keep the te…

At what point would it make sense to move to a 24v or 48v power standard in PCs? The dc-dc conversion world has improved a lot since ATX was invented.

Re: Up to 600 watts of power for graphics cards with the new PCIe 5.0 powerconnector

#153
post #139

I've always wondered why CPU AIO coolers are so prevalent, but GPUs almost exclusively use air coolers. We're nearing a point where GPUs will have a TDP >2X the hungriest CPU (Threadripper), but the cooling used by most is far less effective. It's especially noticeable in smaller builds like those using mini-itx. Most of them GPU coolers draw air in from the bottom and exhaust out the sides, which doesn't really fit…

The difference is that with GPUs nearly every model has a different layout of components around the actual chip, and those components need to be cooled too, so making a universal cold plate for them can be really challenging. The AIO I bought for my last computer could be used with any CPU, AMD or Intel, released in the past decade. The water block I bought for the GPU I put in my desktop two computers ago could only be used with reference designs of the AMD R9 290 or 290x.

There are models of GPUs with AIOs, EVGA makes their Hydro series, Gigabyte has their waterforce series. They just usually aren't worth the cost since you have to design a new full cover block not just got every generation but also usually for different cards in the same gen too.

Lastly, TDP is not a defined standard and each manufacturer defines it differently. Intel for example defines TDP as the maximum allowable power draw under full load. How that translates to actual heat output depends on architecture and process node. Intel's most recent chips for example all max out at just over 100 watts TDP, but they're known to need 200+ watts of heat dissipation for sustained workloads.

Re: Up to 600 watts of power for graphics cards with the new PCIe 5.0 powerconnector

#154

Earlier quoted context omitted.

Every single home in the US has 220 already. They're taken from 220 to 110 at the breaker for common outlets, but many large appliances (and EV chargers) already are 220. Worse kettle performance is unfortunate. But better safety is a nice side effect (in particular as the NEMA connector the US uses is dangerous/poorly designed/bad, even with the ground pin).

If houses in the US are 240V, than houses in Europe are 480 (or rather, 400V because it's usually triphasic). The voltage arriving at the building is kind of irrelevant to the discussion. The question is whether a connection could even sustain the house network at double the voltage with larger appliances than it was designed for.

> The voltage arriving at the building is kind of irrelevant to the discussion.

My post talked about voltages arriving at appliances in US homes. Many appliances are designed and in fact receive 220 V. There's even a common (i.e. near every home) plug/wiring/breaker standard for 220 V.

You're the one trying to steer the discussion towards "arriving Vs. using" whereas the first line of my post was expressly about the active use of 220 V in near every US home today (tumble dryers, ovens, central heating, EV chargers, shop equipment, etc).

> The question is whether a connection could even sustain the house network at double the voltage with larger appliances than it was designed for.

I don't understand what you're trying to say. A full end-to-end circuit has to be built for a specific voltage and amperage. If you want to turn a 110 V into a 220 V, you have to re-wire from the breaker otherwise the wiring will overheat and catch fire.

Re: Up to 600 watts of power for graphics cards with the new PCIe 5.0 powerconnector

#155
post #10

Assuming you can install two GPUs using 600W, that's basically running a hair dryer in your desktop since most of the energy is converted to heat.

It helps save on the winter heating bill if you live in Wisconsin.

Re: Up to 600 watts of power for graphics cards with the new PCIe 5.0 powerconnector

#156
post #139

I've always wondered why CPU AIO coolers are so prevalent, but GPUs almost exclusively use air coolers. We're nearing a point where GPUs will have a TDP >2X the hungriest CPU (Threadripper), but the cooling used by most is far less effective. It's especially noticeable in smaller builds like those using mini-itx. Most of them GPU coolers draw air in from the bottom and exhaust out the sides, which doesn't really fit…

A big part is that CPU sockets are very standardized and you are only cooling the CPU and not the memory and power regulators. For graphics cards the coolers need to cool the gpu along with the memory and power regulators which can be in different places depending on the manufacturer. This usually means you need a block made specifically for your model of card. You can buy those blocks and do custom water cooling but the cost of that is incredibly high. Also it finally comes down to the AIO coolers really aren't better than good air coolers. Sure AIO tend to be quieter and move the heat to a different location but on pure performance they don't crush air coolers.

Re: Up to 600 watts of power for graphics cards with the new PCIe 5.0 powerconnector

#157

Earlier quoted context omitted.

It's not hard to understand, the energy required to heat up food vs energy consumed by the appliance. https://www.energy.gov/sites/prod/files/2015/06/f23/conventi... TLDR: they measure with a metal block.

> It's not hard to understand Please don't patronise! My oven, on full power, draws 2KW (at least, that's what the labels on my fuze-box suggest). If only 12% of that energy is used to heat the food, then the oven amounts to a 200W food cooker, combined with an 1800W electric heater (I rounded down to 10% to make the reckoning easier). I live in a studio flat; my oven is in my living room. How come I can survive with…

Most of the energy created by the oven doesn't actually go into the food, it heats up the air which heats up your food (and also the walls of the oven). While a microwave pretty much only heats the water molecules in your food.

To roast a chicken to 165f you'll probably set your oven to 400f. And then when it's done, the air and interior of your oven are still very hot. When you microwave food, you can touch the side-walls of your microwave right away and they won't be hot. The only excess heat in a microwave comes from the food steaming and contact with the platter.

Re: Up to 600 watts of power for graphics cards with the new PCIe 5.0 powerconnector

#158
post #27

I’ve always thought of gaming as a virtuous low-impact form of entertainment, insofar as it displaces real-world pursuits like foreign ski holidays, gasoline-powered road trips, or land-hogging golf courses. But burning nearly a kilowatt on a gaming rig feels like a step too far. Sure, the power could be sourced from wind or solar plants, but there’s still a level of excessive conspicuous consumption going on here th…

Why so negative. I think gaming is the future. We can heat houses and have fun. If only will be able to store the heat. Why make better chips when the user do not care ?

Re: Up to 600 watts of power for graphics cards with the new PCIe 5.0 powerconnector

#159
post #92

Earlier quoted context omitted.

I understand where you’re coming from, but just finding things can be a lot of effort. If someone wants the images with no watermark, they can look them up

Oh, so I can go to your blog or science article with those nice, detailed posts, take all the photos you made yourself and then make a shitty article on the Internet including your hard work. I will just in case slam my own watermark, so it will look like I did it myself and will not include you as an author. Apparently this sounds fair to you!

You're posting this on a site where the admin encourages users to post archive links to bypass paywalls. Wrong place for arguments that content creators get their fair share.

Re: Up to 600 watts of power for graphics cards with the new PCIe 5.0 powerconnector

#160
post #27

I’ve always thought of gaming as a virtuous low-impact form of entertainment, insofar as it displaces real-world pursuits like foreign ski holidays, gasoline-powered road trips, or land-hogging golf courses. But burning nearly a kilowatt on a gaming rig feels like a step too far. Sure, the power could be sourced from wind or solar plants, but there’s still a level of excessive conspicuous consumption going on here th…

> But burning nearly a kilowatt on a gaming rig feels like a step too far.

How much energy does a golf cart use? Cause those golfers are using a golf-cart for miles when their legs are perfectly acceptable (and golf-carts are illegal in professional play)

Post reply on HN