How much watts would be perfect for my build?

RogueNineOne · January 26, 2020

CPU - i9 9900K
Cooler - Noctua NH D15S
Motherboard - Gigabyte Z390 Aorus Master
RAM - G.Skill Trident Z RGB 4x8 (for some weird reason the stores here only have 8GB sticks)
GPU - Asus GeForce RTX 2080 Ti 11 GB ROG Strix Gaming OC
Case - Cooler Master MasterCase H500P Mesh ATX Mid Tower Case
SSD - Samsung 970 Evo Plus 500 GB M.2-2280 NVME
HDD - Western Digital BLACK SERIES 2 TB 3.5" 7200RPM

Display - Viewsonic XG2405

Considering this PC would run for almost 20 hrs everyday, how much watts should I go for? I was initially thinking of Corsair HX1200i but now I'm having doubts and considering the HX850i....would it suffice or should I go for the 1000W?

ddennis002 · January 26, 2020

10 minutes ago, RogueNineOne said:

CPU - i9 9900K
Cooler - Noctua NH D15S
Motherboard - Gigabyte Z390 Aorus Master
RAM - G.Skill Trident Z RGB 4x8 (for some weird reason the stores here only have 8GB sticks)
GPU - Asus GeForce RTX 2080 Ti 11 GB ROG Strix Gaming OC
Case - Cooler Master MasterCase H500P Mesh ATX Mid Tower Case
SSD - Samsung 970 Evo Plus 500 GB M.2-2280 NVME
HDD - Western Digital BLACK SERIES 2 TB 3.5" 7200RPM

Display - Viewsonic XG2405

Considering this PC would run for almost 20 hrs everyday, how much watts should I go for? I was initially thinking of Corsair HX1200i but now I'm having doubts and considering the HX850i....would it suffice or should I go for the 1000W?

The 850 would be fine for sure but, what do you plan on doing on the system? Depending on where your power draw is going to be most of the time if power efficiency is something you might want to consider most PSU are the most efficent at 40-50% load. You system full load will probably draw around 600W assuming you OC the gpu and CPU. So you would want to be in that 40-50% load range. IF you don't care about $30-40 extra on you power bill a year then 850w would be more then enough to handle your system under full load conditions.

seon123 · January 26, 2020

About 550-650W should be fine. What model you should buy depends on what's available and the pricing. Whether or not you should care about the efficiency depends on the electricity price, how you will use your PC, and the PSU pricing. Obviously running it idle consumes less power than mining on it.

RogueNineOne · January 26, 2020

16 minutes ago, ddennis002 said:

The 850 would be fine for sure but, what do you plan on doing on the system? Depending on where your power draw is going to be most of the time if power efficiency is something you might want to consider most PSU are the most efficent at 40-50% load. You system full load will probably draw around 600W assuming you OC the gpu and CPU. So you would want to be in that 40-50% load range. IF you don't care about $30-40 extra on you power bill a year then 850w would be more then enough to handle your system under full load conditions.

This rig will be used for both gaming and video editing. Also, just out of curiosity....if I get a 1000W PSU or maybe even a 1200W one, what's the downside other than the extra electricity bill? I mean does PSU performance drop if the difference between required wattage and supplied wattage is significantly high?

ddennis002 · January 26, 2020

6 minutes ago, RogueNineOne said:

I mean does PSU performance drop if the difference between required wattage and supplied wattage is significantly high?

Thats where effiency charts come in, your psu will only draw from the wall what is required by the system at that particular instant. So if the hardware is consuming 600w full load that is what is required and being supplied from the psu and from the wall. A larger psu can keep heat down because its load is lower and this in turn will keep the fan noise down or off if the psu supports 0% fan speed.

Spotty · January 26, 2020

3 hours ago, RogueNineOne said:

Considering this PC would run for almost 20 hrs everyday, how much watts should I go for? I was initially thinking of Corsair HX1200i but now I'm having doubts and considering the HX850i....would it suffice or should I go for the 1000W?

With only a single 2080Ti your power consumption would be 500W or so under heavy load. Maybe 550-600W if you overclock the CPU & GPU and stress the GPU while also preforming a CPU intensive task that utilises AVX where the 9900K can peak at close to 250W when overclocked. Though while gaming you're more realistically going to be drawing around 400W or so.

If you're looking at the HXi models they start at 750W, which will be more than enough for your system with a single graphics card. When I checked a few days ago I think there was only $7 or so difference between the HX750i and HX850i, so either one of those won't break the bank. The HX1000i on the other hand is about $40-$50 more expensive, and the HX1200i was something like $80 or $90 more expensive [than the HX750i]. You won't be getting any benefit for having a higher wattage unit if your system doesn't need it so you'd effectively be wasting your money on those higher wattage units. Though prices can vary and I have in the past seen higher wattage units selling close to the same as the lower wattage models due to overstock caused by the mining bust (some stores stocked heaps of high wattage units in anticipation for miners but when prices crashed they were stuck with these high wattage units that 99.9% of people don't need)

So, depending on price probably the HX750i or if it's the same price the HX850i would also be fine. There's no reason to spend more money on the higher wattage 1000W or 1200W units.

2 hours ago, ddennis002 said:

Depending on where your power draw is going to be most of the time if power efficiency is something you might want to consider most PSU are the most efficent at 40-50% load. You system full load will probably draw around 600W assuming you OC the gpu and CPU. So you would want to be in that 40-50% load range.

I can see how you would believe that if you only look at misleading marketing graphs like the one that you posted...

2 hours ago, ddennis002 said:

See how the graph goes from 10% to 20% then 50% then 100%? It's very misleading. They use 10%, 20%, 50%, and 100% values as that is what is tested for the 80+ efficiency certification, however if they want to show off those values they should do so in a table, not in a graph, and most definitely never in a misleading graph with a misleading scale like that. I don't blame you for that graph, I blame Corsair for it since it originally came from their website and product manual, however thankfully Corsair had the sense to remove it from the website ages ago.

Here is the same graph once I've "adjusted" it back to a proper scale with 10%, 20%, 50% and 100% roughly where they should be, though this isn't entirely accurate either since there's no data points for 30%, 40%, 60%, 70%, 80%, 90%

Instead of looking at marketing graphs you really should look at proper efficiency testing that tests at multiple load levels, like Cybenetics efficiency testing which tests at 10% increments up to 110%.

^{(Source: Cybenetics)}

And also the efficiency graph for the HX750i

^{(Source: Cybenetics)}

3 hours ago, ddennis002 said:

You system full load will probably draw around 600W assuming you OC the gpu and CPU. So you would want to be in that 40-50% load range. IF you don't care about $30-40 extra on you power bill a year then 850w would be more then enough to handle your system under full load conditions.

I'm going to go with the load value of 600W since that is what you specified...
At 115V AC mains with a load of 600W there is only 0.2% difference in efficiency between the HX1200i and HX750i (91.054% vs 90.847%) ^{[Source: Cybenetics]}. That's 1.2W difference at 600W load. Over the course of 1 year, running the system at full load 24/7, that's a difference of approximately 10kWH, or around about $2-$3 varying slightly depending on electricity prices. Keep in mind you have to pay an extra $90 for the higher wattage HX1200i over the HX750i based on current prices on Newegg. So if you run the system at full load 600W 24/7 it would take you 30+ years before you start seeing a return on the higher upfront cost of the higher wattage unit.

3 hours ago, ddennis002 said:

A larger psu can keep heat down because its load is lower and this in turn will keep the fan noise down or off if the psu supports 0% fan speed.

That's not how that works. The higher wattage unit does not run cooler because it's at a lower percentage of its maximum rated load.

Basic theory: Two PSUs with 91% efficiency at 500W load. Since both PSUs have the same efficiency they are both drawing 550W AC from the mains and providing 500W to the system, with the remaining 50W being wasted as heat due to efficiency losses. Both PSUs have 50W of heat to dissipate regardless of what their maximum rated load is.

The differences start to come in to play are when in order to keep higher wattage units cool under their maximum rated load they might use a different design, use larger heatsinks, or use a fan that spins at a higher RPM or a more aggressive fan profile.

The Corsair HXi series PSUs use very good fans with a sophisticated fan controller. According to Cybenetics testing at 600W load the HX750i spins its fan at 840RPM and noise measured at 20dBa (virtually inaudible and under the sound floor of most rooms). The HX1200i at 600W load spins its fan at 750RPM with noise measured at 19dBa (virtually inaudible and under the sound floor of most rooms). Besides, at 600W load the PSU is going to be the quietest thing in the system with GPU and CPU fans needing to run at much higher RPMs to keep them cool under load.
There's not going to be any difference between the two in regards to fan noise.

TLDR; The 750W and 1200W are both the same in terms of noise and efficiency. Buying a higher wattage than you need does not make your system more efficient and it does not make it quieter.

RogueNineOne · January 26, 2020

8 hours ago, Spotty said:

With only a single 2080Ti your power consumption would be 500W or so under heavy load. Maybe 550-600W if you overclock the CPU & GPU and stress the GPU while also preforming a CPU intensive task that utilises AVX where the 9900K can peak at close to 250W when overclocked. Though while gaming you're more realistically going to be drawing around 400W or so.

If you're looking at the HXi models they start at 750W, which will be more than enough for your system with a single graphics card. When I checked a few days ago I think there was only $7 or so difference between the HX750i and HX850i, so either one of those won't break the bank. The HX1000i on the other hand is about $40-$50 more expensive, and the HX1200i was something like $80 or $90 more expensive [than the HX750i]. You won't be getting any benefit for having a higher wattage unit if your system doesn't need it so you'd effectively be wasting your money on those higher wattage units. Though prices can vary and I have in the past seen higher wattage units selling close to the same as the lower wattage models due to overstock caused by the mining bust (some stores stocked heaps of high wattage units in anticipation for miners but when prices crashed they were stuck with these high wattage units that 99.9% of people don't need)

So, depending on price probably the HX750i or if it's the same price the HX850i would also be fine. There's no reason to spend more money on the higher wattage 1000W or 1200W units.

I can see how you would believe that if you only look at misleading marketing graphs like the one that you posted...

See how the graph goes from 10% to 20% then 50% then 100%? It's very misleading. They use 10%, 20%, 50%, and 100% values as that is what is tested for the 80+ efficiency certification, however if they want to show off those values they should do so in a table, not in a graph, and most definitely never in a misleading graph with a misleading scale like that. I don't blame you for that graph, I blame Corsair for it since it originally came from their website and product manual, however thankfully Corsair had the sense to remove it from the website ages ago.

Here is the same graph once I've "adjusted" it back to a proper scale with 10%, 20%, 50% and 100% roughly where they should be, though this isn't entirely accurate either since there's no data points for 30%, 40%, 60%, 70%, 80%, 90%

Instead of looking at marketing graphs you really should look at proper efficiency testing that tests at multiple load levels, like Cybenetics efficiency testing which tests at 10% increments up to 110%.

^{(Source: Cybenetics)}

And also the efficiency graph for the HX750i

^{(Source: Cybenetics)}

I'm going to go with the load value of 600W since that is what you specified...
At 115V AC mains with a load of 600W there is only 0.2% difference in efficiency between the HX1200i and HX750i (91.054% vs 90.847%) ^{[Source: Cybenetics]}. That's 1.2W difference at 600W load. Over the course of 1 year, running the system at full load 24/7, that's a difference of approximately 10kWH, or around about $2-$3 varying slightly depending on electricity prices. Keep in mind you have to pay an extra $90 for the higher wattage HX1200i over the HX750i based on current prices on Newegg. So if you run the system at full load 600W 24/7 it would take you 30+ years before you start seeing a return on the higher upfront cost of the higher wattage unit.

That's not how that works. The higher wattage unit does not run cooler because it's at a lower percentage of its maximum rated load.

Basic theory: Two PSUs with 91% efficiency at 500W load. Since both PSUs have the same efficiency they are both drawing 550W AC from the mains and providing 500W to the system, with the remaining 50W being wasted as heat due to efficiency losses. Both PSUs have 50W of heat to dissipate regardless of what their maximum rated load is.

The differences start to come in to play are when in order to keep higher wattage units cool under their maximum rated load they might use a different design, use larger heatsinks, or use a fan that spins at a higher RPM or a more aggressive fan profile.

The Corsair HXi series PSUs use very good fans with a sophisticated fan controller. According to Cybenetics testing at 600W load the HX750i spins its fan at 840RPM and noise measured at 20dBa (virtually inaudible and under the sound floor of most rooms). The HX1200i at 600W load spins its fan at 750RPM with noise measured at 19dBa (virtually inaudible and under the sound floor of most rooms). Besides, at 600W load the PSU is going to be the quietest thing in the system with GPU and CPU fans needing to run at much higher RPMs to keep them cool under load.
There's not going to be any difference between the two in regards to fan noise.

TLDR; The 750W and 1200W are both the same in terms of noise and efficiency. Buying a higher wattage than you need does not make your system more efficient and it does not make it quieter.