750watts should be bumped up to at minimuim 850 watts.
The CPU won't use more than 142 W. While I haven't bought a video card yet, I intend to keep that inside of 300 W. Add in a generous 100 W buffer for everything else that sits behind a power supply (which notably excludes monitors) and it's unlikely that I'll ever draw 550 W from the power supply. A high quality 750 W power supply is plenty enough for that.
if you like living on the edge, that's up too you. how many usb peripherals you using ?
Maybe he's a hardcore Aerosmith fan and he lives their music.
750watts should be bumped up to at minimuim 850 watts.
The CPU won't use more than 142 W. While I haven't bought a video card yet, I intend to keep that inside of 300 W. Add in a generous 100 W buffer for everything else that sits behind a power supply (which notably excludes monitors) and it's unlikely that I'll ever draw 550 W from the power supply. A high quality 750 W power supply is plenty enough for that.
if you like living on the edge, that's up too you. how many usb peripherals you using ?
Most USB peripherals use only a fraction of a watt. Regardless, that's accounted for by the "100 W for everything else", which is generally going to be a huge overestimate.
Building a desktop that might commonly pull 400 W from a 750 W power supply and hardly ever pull more than 500 W is hardly living on the edge.
With a good PSU, 750 watts is probably 100 watts too many. If he was getting an RTX 4090, the 750 watts should still be adaquette. I have only experienced one setup using a single GPU that required 850 watts.
I've got all of the parts except for the UPS, so I started assembling the computer. I'm stopping for the night now because my back hurts, so it's not done yet.
Installing the water cooler for the CPU is a lot more work than I realized. It's four "unscrew this from the motherboard", then 28 "screw this into that", and then a maze of eight "plug this 4-pin connector into that one" steps. And after I've screwed everything into place, I notice that the radiator is blocking the motherboard's CPU fan header. I might be able to sneak it in underneath, but I'm likely looking at removing the radiator (eight screws) to plug in the the fan header. I'm sure that it's fixable, but it's more work.
Once upon a time I was running a mildly overclocked i7 4970 and GTX 980 on a 450W power supply. It pulled 420W from the wall. Worked that way for several years, although I did eventually end up upgrading the PSU in it. It also was a decent Seasonic unit, not a no-name or obscure PSU.
My current rig can pull 500W from the wall if I have the CPU and GPU cranked at the same time, but most of the time it's down around 400W. I'm running an 750W PSU. That is on a 5900X and a 3080 - the 3080 has some mild factory overclocks.
Checking that against my thumbrule: (GPU + CPU + 100)
320W (3080) + 105W (CPU) + 100W misc = ... 525W. Actual draw = 500W. Looks pretty close. 750W for me is plenty of margin, I probably could get away just as well with a 650W.
The biggest PITA was that I had to upgrade my UPS, since i also have 2 monitors, audio monitor speakers, a lamp and a network switch on it - my old 800VA unit wasn't big enough and kept throwing out on overload, I had to bump up a couple of sizes. Everything on my desk pulls 700W all together if I have the computer running burn-in stuff.
You don't need to go batshit bonkers on PSU ratings, unless you just have a habit of going with the cheapest thing you can find, or just like to throw money at your rig for... whatever reason.
Having some margin is good - overclocks can crank power requirements through the roof, and you need to have some provision for transients and aging and other stuff.
The only reason PSU OEMs make the recommendations that they do is because they have to assume you are going to get some POS PSU, and the only way they can not be liable is if they so vastly overshoot the requirements that they are probably going to be safe.
For CPU power, you have to look at the max turbo power, not the nominal TDP. In your case, that's 142 W, not 105. If you have a factory overclocked video card, then your particular SKU might well be rated at significantly above 320 W.
Also, the rated power draw for a power supply is the power that you're drawing from the power supply, which is likely to be about 10% less than what the power supply is drawing from the wall. It also excludes the power draw of monitors and other peripherals that don't go through the power supply. I'm not sure how you measured that 500 W, but if it's at the UPS, then it's reporting what the UPS is pulling from the wall, which is quite a bit more than what the computer is pulling from the power supply.
For CPU power, you have to look at the max turbo power, not the nominal TDP. In your case, that's 142 W, not 105. If you have a factory overclocked video card, then your particular SKU might well be rated at significantly above 320 W.
Also, the rated power draw for a power supply is the power that you're drawing from the power supply, which is likely to be about 10% less than what the power supply is drawing from the wall. It also excludes the power draw of monitors and other peripherals that don't go through the power supply. I'm not sure how you measured that 500 W, but if it's at the UPS, then it's reporting what the UPS is pulling from the wall, which is quite a bit more than what the computer is pulling from the power supply.
You say all of that, but the number out of the wall is what it is.
I'd also point out - in paragraph one, you argue my number should be higher, because... math.
Then, in paragraph two, the number should be lower, because... math.
Maybe you are arguing for offsetting penalties? I don't know.
But it is what it is, no matter what we think it should be and the reasons why - it measures what it measures. And it seems to be more or less in line with my thumbrule -- which isn't intended to be a perfectly exact calculation, just a method to get in the ballpark; and it appears to do that very well.
(A hint, I talked about what my power draw out of the UPS itself was, and it was definitely not the same number as what I reported my computer as, and if you read it, it even tells why they aren't the same number)
I will, however, concede that you are right about 142W Max Turbo versus 105W TDP, although it takes some very specific conditions for the chip to hit up to and hold 142W for any length of time.
In general with an AMD Cpu, I wouldn't bother with watercooling the cpu. The stock cooler should be plenty. An air cooler also has less maintenence. Intel coolers for a time had inadequate stock coolers. I am not sure if that is still the case. I have switched to only water cooling the GPU as that tends to get crispy with inadaquette coolers.
In general with an AMD Cpu, I wouldn't bother with watercooling the cpu. The stock cooler should be plenty. An air cooler also has less maintenence. Intel coolers for a time had inadequate stock coolers. I am not sure if that is still the case. I have switched to only water cooling the GPU as that tends to get crispy with inadaquette coolers.
That CPU doesn't come with cooler.
Also even if AMD's stock coolers are good enough to keep the CPU cool, a good third party cooler can make the computer much quieter.
With the latest versions of Boost - yeah, the stock cooler will work.
But the better cooling you can get, the higher the chips can clock and the longer they can hold higher clocks.
So, where it used to be that you just needed to keep a chip cool enough and that was good enough - now, better cooling leads to better and more consistent performance.
I don't know that a full blown custom water loop would be worth the investment over a decent air cooler, but yeah, the days where you would just throw the stock cooler or a cheap CM212 at it and call it good enough ... well, it at least is enough of a deal to make you pause and consider it. Especially if you are spending anything extra and buying a nicer CPU.
In general with an AMD Cpu, I wouldn't bother with watercooling the cpu. The stock cooler should be plenty. An air cooler also has less maintenence. Intel coolers for a time had inadequate stock coolers. I am not sure if that is still the case. I have switched to only water cooling the GPU as that tends to get crispy with inadaquette coolers.
The CPU doesn't come with a stock cooler. It was a choice between buying an air cooler or buying a water cooler. When I started looking, I was actually more interested in an air cooler, but getting a good one actually cost more than what I paid for a water cooler.
Your link goes to it costing $97.58, plus shipping. I paid $50 for a similar cooler with free shipping. I didn't go in thinking I was going to get the best and pay whatever it cost. I wanted a good value for the money.
I've got all of the parts except for the UPS, so I started assembling the computer. I'm stopping for the night now because my back hurts, so it's not done yet.
Installing the water cooler for the CPU is a lot more work than I realized. It's four "unscrew this from the motherboard", then 28 "screw this into that", and then a maze of eight "plug this 4-pin connector into that one" steps. And after I've screwed everything into place, I notice that the radiator is blocking the motherboard's CPU fan header. I might be able to sneak it in underneath, but I'm likely looking at removing the radiator (eight screws) to plug in the the fan header. I'm sure that it's fixable, but it's more work.
That's the reason why I paid to have my current PC assembled. It's just too much effort for me to do that anymore.
Logic, my dear, merely enables one to be wrong with great authority.
I've got all of the parts except for the UPS, so I started assembling the computer. I'm stopping for the night now because my back hurts, so it's not done yet.
Installing the water cooler for the CPU is a lot more work than I realized. It's four "unscrew this from the motherboard", then 28 "screw this into that", and then a maze of eight "plug this 4-pin connector into that one" steps. And after I've screwed everything into place, I notice that the radiator is blocking the motherboard's CPU fan header. I might be able to sneak it in underneath, but I'm likely looking at removing the radiator (eight screws) to plug in the the fan header. I'm sure that it's fixable, but it's more work.
That's the reason why I paid to have my current PC assembled. It's just too much effort for me to do that anymore.
For CPU power, you have to look at the max turbo power, not the nominal TDP. In your case, that's 142 W, not 105. If you have a factory overclocked video card, then your particular SKU might well be rated at significantly above 320 W.
Also, the rated power draw for a power supply is the power that you're drawing from the power supply, which is likely to be about 10% less than what the power supply is drawing from the wall. It also excludes the power draw of monitors and other peripherals that don't go through the power supply. I'm not sure how you measured that 500 W, but if it's at the UPS, then it's reporting what the UPS is pulling from the wall, which is quite a bit more than what the computer is pulling from the power supply.
You say all of that, but the number out of the wall is what it is.
I'd also point out - in paragraph one, you argue my number should be higher, because... math.
Then, in paragraph two, the number should be lower, because... math.
Maybe you are arguing for offsetting penalties? I don't know.
But it is what it is, no matter what we think it should be and the reasons why - it measures what it measures. And it seems to be more or less in line with my thumbrule -- which isn't intended to be a perfectly exact calculation, just a method to get in the ballpark; and it appears to do that very well.
(A hint, I talked about what my power draw out of the UPS itself was, and it was definitely not the same number as what I reported my computer as, and if you read it, it even tells why they aren't the same number)
I will, however, concede that you are right about 142W Max Turbo versus 105W TDP, although it takes some very specific conditions for the chip to hit up to and hold 142W for any length of time.
I argue that the first number (theoretical power envelope) should be higher, and then that the second number (actual power draw from the power supply) should be lower. Those aren't offsetting. Both will tend to make the gap between theoretical and actual power draw larger.
While the power draw at the wall is what it is for electric bill purposes, or even UPS purposes, that's not the same as power supply purposes. If you turn on a microwave or dryer, it will draw a lot of power from the wall, but that's irrelevant to your computer's power supply. The power draw of monitors is likewise irrelevant to your power supply, even if it's important to your UPS.
Ultimately, I'm not sure how you're measuring the second number. If it's the power draw from the power supply, then sure, that's the right number, but that's typically hard to measure. But that's not the same as power draw from the wall. Even if you unplug monitors or whatever, a 90% efficient power supply might pull 500 W from the wall even as it's only delivering 450 W to the rest of the computer.
first number (theoretical power envelope) should be higher, and then that the second number (actual power draw from the power supply) should be lower. Those aren't offsetting. Both will tend to make the gap between theoretical and actual power draw larger.
While the power draw at the wall is what it is for electric bill purposes, or even UPS purposes, that's not the same as power supply purposes. If you turn on a microwave or dryer, it will draw a lot of power from the wall, but that's irrelevant to your computer's power supply. The power draw of monitors is likewise irrelevant to your power supply, even if it's important to your UPS.
Ultimately, I'm not sure how you're measuring the second number. If it's the power draw from the power supply, then sure, that's the right number, but that's typically hard to measure. But that's not the same as power draw from the wall. Even if you unplug monitors or whatever, a 90% efficient power supply might pull 500 W from the wall even as it's only delivering 450 W to the rest of the computer.
I have no idea how washing machines or microwaves come into play here.
I also have no idea what you are talking about. If you can't figure out how I measured at the wall, when I include the link of the exact piece of equipment I used to do so for just the computer, then... I have no idea.
I think I'm going to chalk this up as - I wrote exactly what I meant, and just leave it at that.
I also have no idea what you are talking about. If you can't figure out how I measured at the wall, when I include the link of the exact piece of equipment I used to do so for just the computer, then... I have no idea.
I didn't read your post carefully enough before responding. I apologize.
As an update, I've now got the computer up and working. I ran into two additional problems.
First, the monitor cable isn't long enough to reach from where I want the monitor to where I want the case. The power cable is, but the HDMI cable is not. It's not an especially short cable, as it's about 5-6 feet long. I just want them further apart than that. I'm more pleased that the power cable can handle the ~8 feet of separation that I want than upset that the included monitor cable can't.
That's easy enough to fix: buy a new monitor cable. And this time, I'll make it DisplayPort rather than HDMI, since both the monitor and the motherboard support both. But it is annoying to have to wait several more days to finish getting everything set up.
The other problem is that one of the case fans is extremely loud. It's conceivable that there's something in there that the fan is hitting, but I can't get physical access to the fan from any side to fix it or even check on it. I know that it's that particular fan because when I unplugged the power to the fan from the other end of the cable, suddenly the computer was quiet. There was still normal air-blowing noise, but nothing like the awful racket that the one fan was making. As far as I'm concerned that fan is effectively dead on arrival.
Even so, that's why you get a case with more fans than you need: redundancy. If you only have one case fan and it dies, you have a problem. If you have four case fans and one fails, it's not a big deal. Especially since I don't have a discrete video card in the computer yet and the CPU is liquid cooled, so that I'm not relying on the case fans to handle CPU heat.
Comments
Maybe he's a hardcore Aerosmith fan and he lives their music.
Building a desktop that might commonly pull 400 W from a 750 W power supply and hardly ever pull more than 500 W is hardly living on the edge.
Installing the water cooler for the CPU is a lot more work than I realized. It's four "unscrew this from the motherboard", then 28 "screw this into that", and then a maze of eight "plug this 4-pin connector into that one" steps. And after I've screwed everything into place, I notice that the radiator is blocking the motherboard's CPU fan header. I might be able to sneak it in underneath, but I'm likely looking at removing the radiator (eight screws) to plug in the the fan header. I'm sure that it's fixable, but it's more work.
My current rig can pull 500W from the wall if I have the CPU and GPU cranked at the same time, but most of the time it's down around 400W. I'm running an 750W PSU. That is on a 5900X and a 3080 - the 3080 has some mild factory overclocks.
Checking that against my thumbrule: (GPU + CPU + 100)
320W (3080) + 105W (CPU) + 100W misc = ... 525W. Actual draw = 500W. Looks pretty close. 750W for me is plenty of margin, I probably could get away just as well with a 650W.
The biggest PITA was that I had to upgrade my UPS, since i also have 2 monitors, audio monitor speakers, a lamp and a network switch on it - my old 800VA unit wasn't big enough and kept throwing out on overload, I had to bump up a couple of sizes. Everything on my desk pulls 700W all together if I have the computer running burn-in stuff.
You don't need to go batshit bonkers on PSU ratings, unless you just have a habit of going with the cheapest thing you can find, or just like to throw money at your rig for... whatever reason.
Having some margin is good - overclocks can crank power requirements through the roof, and you need to have some provision for transients and aging and other stuff.
The only reason PSU OEMs make the recommendations that they do is because they have to assume you are going to get some POS PSU, and the only way they can not be liable is if they so vastly overshoot the requirements that they are probably going to be safe.
Also, the rated power draw for a power supply is the power that you're drawing from the power supply, which is likely to be about 10% less than what the power supply is drawing from the wall. It also excludes the power draw of monitors and other peripherals that don't go through the power supply. I'm not sure how you measured that 500 W, but if it's at the UPS, then it's reporting what the UPS is pulling from the wall, which is quite a bit more than what the computer is pulling from the power supply.
I'd also point out - in paragraph one, you argue my number should be higher, because... math.
Then, in paragraph two, the number should be lower, because... math.
Maybe you are arguing for offsetting penalties? I don't know.
But it is what it is, no matter what we think it should be and the reasons why - it measures what it measures. And it seems to be more or less in line with my thumbrule -- which isn't intended to be a perfectly exact calculation, just a method to get in the ballpark; and it appears to do that very well.
http://www.p3international.com/products/p4400.html
(A hint, I talked about what my power draw out of the UPS itself was, and it was definitely not the same number as what I reported my computer as, and if you read it, it even tells why they aren't the same number)
I will, however, concede that you are right about 142W Max Turbo versus 105W TDP, although it takes some very specific conditions for the chip to hit up to and hold 142W for any length of time.
I have switched to only water cooling the GPU as that tends to get crispy with inadaquette coolers.
Also even if AMD's stock coolers are good enough to keep the CPU cool, a good third party cooler can make the computer much quieter.
Cooler Master MasterLiquid ML240L V2 RGB Cooling Fan/Radiator/Water Block
But the better cooling you can get, the higher the chips can clock and the longer they can hold higher clocks.
So, where it used to be that you just needed to keep a chip cool enough and that was good enough - now, better cooling leads to better and more consistent performance.
I don't know that a full blown custom water loop would be worth the investment over a decent air cooler, but yeah, the days where you would just throw the stock cooler or a cheap CM212 at it and call it good enough ... well, it at least is enough of a deal to make you pause and consider it. Especially if you are spending anything extra and buying a nicer CPU.
Logic, my dear, merely enables one to be wrong with great authority.
While the power draw at the wall is what it is for electric bill purposes, or even UPS purposes, that's not the same as power supply purposes. If you turn on a microwave or dryer, it will draw a lot of power from the wall, but that's irrelevant to your computer's power supply. The power draw of monitors is likewise irrelevant to your power supply, even if it's important to your UPS.
Ultimately, I'm not sure how you're measuring the second number. If it's the power draw from the power supply, then sure, that's the right number, but that's typically hard to measure. But that's not the same as power draw from the wall. Even if you unplug monitors or whatever, a 90% efficient power supply might pull 500 W from the wall even as it's only delivering 450 W to the rest of the computer.
I have no idea how washing machines or microwaves come into play here.
I also have no idea what you are talking about. If you can't figure out how I measured at the wall, when I include the link of the exact piece of equipment I used to do so for just the computer, then... I have no idea.
I think I'm going to chalk this up as - I wrote exactly what I meant, and just leave it at that.
First, the monitor cable isn't long enough to reach from where I want the monitor to where I want the case. The power cable is, but the HDMI cable is not. It's not an especially short cable, as it's about 5-6 feet long. I just want them further apart than that. I'm more pleased that the power cable can handle the ~8 feet of separation that I want than upset that the included monitor cable can't.
That's easy enough to fix: buy a new monitor cable. And this time, I'll make it DisplayPort rather than HDMI, since both the monitor and the motherboard support both. But it is annoying to have to wait several more days to finish getting everything set up.
The other problem is that one of the case fans is extremely loud. It's conceivable that there's something in there that the fan is hitting, but I can't get physical access to the fan from any side to fix it or even check on it. I know that it's that particular fan because when I unplugged the power to the fan from the other end of the cable, suddenly the computer was quiet. There was still normal air-blowing noise, but nothing like the awful racket that the one fan was making. As far as I'm concerned that fan is effectively dead on arrival.
Even so, that's why you get a case with more fans than you need: redundancy. If you only have one case fan and it dies, you have a problem. If you have four case fans and one fails, it's not a big deal. Especially since I don't have a discrete video card in the computer yet and the CPU is liquid cooled, so that I'm not relying on the case fans to handle CPU heat.