Intel Core i9-7980XE Hits 845W at 4.9GHz

[Dylanc1500]

2 hours ago, ScratchCat said:

Although this puts skylake-X in a bad light , we have gone from a 10 core $1799 chip which could reach 4.3 GHz on a good day to an 18 core $2000 chip which reaches greater speeds in a single generation. If this isn't testament to how much Intel was sitting around / Ryzen  exceeded expectations I don't know what is.

Im interested to see what the everyday stable overclock is , 4.5 should be achievable. Although this chip is consuming ~15X the power of a Core i5 at stock!

So imagine the 28 core die at these speeds. Actually, imagine the power draw. Now that would make for a show.

[ScratchCat]

18 minutes ago, Dylanc1500 said:

So imagine the 28 core die at these speeds. Actually, imagine the power draw. Now that would make for a show.

Intel should send some to the European fusion research project. Would advance the project by years

[Belgarathian]

There is no way that a Corsair H105 can sustain 850W of heat dissipation. Would have finished Cinebench in ~5s. 

[Hybris5112]

I wonder if 2 360 push/pull rads would be enough to keep things cool with this kind of chip.

 

Either way good to see Intel take their biggest advantage and slap AMD with it for a change even when they still cripple themselves with bad TIM.

[Syntaxvgm]

10 hours ago, HKZeroFive said:

...but you said it yourself that the FX-8350 is a 125W CPU with no overclock... when the TDP is 125W.

 

Except for that part where I continue on as if that's the case, and do not factor this in.

 

10 hours ago, Syntaxvgm said:

Don't factor that in, but it's worth knowing that it's often that the TDP is lower than the peak power consumption under maximum stress.  

 

TDP isn't 1:1 with power consumption necessarily, and often for intel it's lower than the actual power consumption peak, and more to represent typical power use. I added this because I thought it was a point you might bring up, TDP != power use, but I find it's lower a lot. In either case, for overclock power consumption, it is not even worth considering. 

 

I went on to use total system power draw for each of these scenarios because that's a real number that was actually measured in use. It also conveniently means we only have to add GPU power consumption for each of these scenarios since everything else is factored in for their test benches, which have nothing crazy going on. 

 

10 hours ago, HKZeroFive said:

...but you said it yourself that the FX-8350 is a 125W CPU with no overclock... when the TDP is 125W.

But I digress.

You did read what I said right?

So yes, have power-hungry components, overclock everything, run it through stress tests and it'll surpass that magical 650W figure. I already agreed with that. I thought that was already done with.

 

No, not at all. If you read what I said, it's not even acceptable for stock, no overlook. The toms hardware article that you linked number of 610 watts was gaming and with a cpu that uses much less power! An 8350 will exceed 650 watts without an overclock in typical gaming, not even synthetic benchmarks. How fucking hard is this?

 

10 hours ago, HKZeroFive said:

 

I'll repeat what I said to @Bananasplit_00:

 

The whole point of your argument is based upon the fact that people will put together the most power-hungry of components, then go the extra step further to overclock them to further increase their power consumption and then go one further by running them through synthetic (read: unrealistic) torture tests that are a worst case simulation. It's a "what if" argument.

 

No, no it's not. I did both arguments, and non-overclocked an 8350 and a 295x2 cannot be run on a 650w power supply. WIth a more practical less power hungry cpu, you have a 610 watt peak, and that's stupid. I said 8350 the entire time, and you're the on that linked a 4770k, which should not be run from a 650w psu either. 

 

10 hours ago, HKZeroFive said:

 

 

Don't manipulate the facts to suit your own agenda. I'm not sure what else I can say.

I'm not manipulating facts, these are facts. I'm presenting them as I find them, since you keep wanting to argue this stupid argument. You cannot run a 295x2 from a 650w psu. It draws 507 watts when in full use. My original scenario was pairing it with an 8350 and a normal overclock. People always overclocked those things, they sucked unless you did. BUt your scenario for some reason was a 4770k and that disproved me somehow, and no 610 watt draw when gaming on a 650w psu is not doable, because gaming isn't peak draw, and even if it was that's beyond stupid. 

These are facts, I didn't manipulate them,  they are what they are. My agenda is to drill them in your head. 

[Chaos_Sorcerer]

uhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh

 

so are these going away now?

 

[79wjd]

7 hours ago, Dylanc1500 said:

So imagine the 28 core die at these speeds. Actually, imagine the power draw. Now that would make for a show.

I hear Michael Bay used one in the making of Transformers 6. 

[mr moose]

21 hours ago, MageTank said:

TDP is worthless once OCing is involved.

TDP figures need to be banned from discussions unless you have a formal qualification relating to their use.  No one uses them correctly, not even the supposed tech authorities that report on this stuff.

 

20 hours ago, Syntaxvgm said:

 

what if I told you TDP is not actually the power consumption limit, but just a ballpark number that's used for cooling? It's actually often far off from the real draw numbers. Don't factor that in, but it's worth knowing that it's often that the TDP is lower than the peak power consumption under maximum stress. 

 

 

Thank you, I was getting Irate at TDP numbers being bandied around as if they relate to power draw.  I think we need some sort of techquicky video on TDP.  And more importantly why people should ignore it. 

 

 

With regard to the article,  if they managed to keep 18cores at 4.9Ghz down to 63C with a H105, then I'm guessing their run time was short or there were some other extenuating factors.  I think while it is prudent to understand they are glossing over some details for the sake of a review,  lets not dismiss the fact that we are probably not far from seeing this more often and in more stable configurations.  

[tbake0155]

3 hours ago, mr moose said:

TDP figures need to be banned from discussions unless you have a formal qualification relating to their use.  No one uses them correctly, not even the supposed tech authorities that report on this stuff.

 

 

Thank you, I was getting Irate at TDP numbers being bandied around as if they relate to power draw.  I think we need some sort of techquicky video on TDP.  And more importantly why people should ignore it. 

 

 

With regard to the article,  if they managed to keep 18cores at 4.9Ghz down to 63C with a H105, then I'm guessing their run time was short or there were some other extenuating factors.  I think while it is prudent to understand they are glossing over some details for the sake of a review,  lets not dismiss the fact that we are probably not far from seeing this more often and in more stable configurations.  

While it is true that TDP is not the power used by the CPU, it is also true that all of the power that goes into the CPU is eventually released as heat. 

 

So although TDP != power draw, we can simplify and say that TDP == power draw.  That's why it is used as a simplification for the masses, who generally don't care about having an expert's understanding of TDP. 

 

[Hunter259]

What is up with people wanting to run their PSU's within 100 watts of their rating? Do you guys just enjoy stress testing or something? Two overclocked 980ti's and a highly overclocked 2600k drew around 800 watts by itself in my testing. Even with a platinum unit i'm not highly for pushing it like that for extended periods of time.

[HKZeroFive]

18 hours ago, WereCat said:

My 1080ti alone consumes 400W....

The one in graph is FE at stock. Once you OC VRAM and core the power will skyrocket. Same with CPU OC, especially on 18-core.

400W is absurdly high for a GTX 1080Ti by itself. Considering around 400W is normal for total system power consumption figures (including an overclocked GTX 1080Ti and a HEDT CPU), most people would be fine with a decent 550W unit... I have a mate running a i7-5820K with an overclocked GTX 1080Ti on a Seasonic 550G.

20 hours ago, Bananasplit_00 said:

PCPP is basically a PSU calculator, and i said they are not perfect. 3 SSDs and 3HDDs, RAID 0 950s or something with some quick SSD storage and RAIDed HDD bulk and backup, overkill? yes, but what im getting at is that there ARE and HAVE BEEN usees for 1KW PSUs for ages.

And PSU calculators overestimate by a large amount and they might as well be random number generators. You want numbers, you look up power consumption benchmarks for an accurate figure of total power draw.

 

Again, your argument centres on the fact that this small demographic of people also happen to have an absolute boatload of drives when a) SSDs such as the 950 PRO don't exactly consume that much power (and why would you RAID 0 them when it's impractical?) b) you probably won't need that much storage in the first place, even with a backup.

20 hours ago, Bananasplit_00 said:

this is about there being uses for 1KW PSUs, not about that you can make a HEDT system that draws 450W

980TIs Total board power is capped at 275W unless you unlimit them, take two of those and you get 550W, and then an overclocked 7900x which is 450W and you get 1000W... i brought up the 480 because i said there has been reasons to have 1KW PSUs for ages.

Again, when I said "there is now a valid reason to get a 1KW unit", I was referring to the current PC market where most systems with two high-end GPUs can run fine on a 850W unit, not the one seven years ago where the name Fermi/Thermi was synonymous with the words "fire hazard".

 

I do suppose that Skylake-X has now brought up a new meaning of "power-hungry" CPUs, due to the inclusion of AVX-512 without a proper AVX offset implementation and the constant abuse of MCE by motherboard manufacturers.

20 hours ago, Bananasplit_00 said:

it still performs like the 1080TI, thats why. if you are upgradeing from whatever HEDT platform you were on you can still keep that card.

No. The R9 295X2, even in a game that scales well with Crossfire, does not perform as well as the GTX 1080Ti... in fact, it performs significantly worse (look up TechPowerUp's benchmarks).

 

And like I said, if you're going to drop $1000 on a CPU alone, you're not going to be running a R9 295X2. You'll most likely (read: definitely) be upgrading it to something better.

20 hours ago, Bananasplit_00 said:

im claiming people build systems that will draw need 1KW PSUs or bigger, because they are. a pair of 980TIs paired with a 7900x isnt to me an unrealistic system at all. 

Alright, Skylake-X aside, since power consumption on those chips alone is ridiculous, you would be hard pressed to find any dual GPU system that requires a 1000W unit. Before Skylake-X, any system with dual GTX 980Tis would only require a 850W unit max... even a 750W one would have been fine.

20 hours ago, Bananasplit_00 said:

because you claimed there was no other reason to get a 1KW PSU beyond mining. 

 

oh and this is all compleatly disregarding anyone that does LN2 overclocking, do that and 2KW aint enough anymore.

So yes, beyond a niche market, nobody needs a 1000W unit. Seems to fit into my claim of "most rigs don't require 1000+W power supplies...".

12 hours ago, Syntaxvgm said:

Except for that part where I continue on as if that's the case, and do not factor this in.

 

TDP isn't 1:1 with power consumption necessarily, and often for intel it's lower than the actual power consumption peak, and more to represent typical power use. I added this because I thought it was a point you might bring up, TDP != power use, but I find it's lower a lot. In either case, for overclock power consumption, it is not even worth considering.

 

I went on to use total system power draw for each of these scenarios because that's a real number that was actually measured in use. It also conveniently means we only have to add GPU power consumption for each of these scenarios since everything else is factored in for their test benches, which have nothing crazy going on.

I agree so far.

12 hours ago, Syntaxvgm said:

No, not at all. If you read what I said, it's not even acceptable for stock, no overlook. The toms hardware article that you linked number of 610 watts was gaming and with a cpu that uses much less power! An 8350 will exceed 650 watts without an overclock in typical gaming, not even synthetic benchmarks. How fucking hard is this?

Calm down. Why are we constantly trying to bring in the more power-hungry FX-8350, much less an overclocked one, into the equation when the i7-4770K provided in the Tom's Hardware article consumes less power and performs better, meaning that it's more realistically suited to be used in a system of this calibre unless you're a diehard AMD fanboy?

12 hours ago, Syntaxvgm said:

No, no it's not. I did both arguments, and non-overclocked an 8350 and a 295x2 cannot be run on a 650w power supply. WIth a more practical less power hungry cpu, you have a 610 watt peak, and that's stupid. I said 8350 the entire time, and you're the on that linked a 4770k, which should not be run from a 650w psu either. 

So since you insisted that I didn't read the article properly the first time around, I went back and had another look to make sure that I wasn't misremembering certain statistics.

Quote

We measure just under 430W in while gaming, which is in line with company's specifications and a lot less than the >500W figure we've seen thrown around... The stress test is next. AMD's Radeon R9295X2 stays just under 450W there as well, which goes to show that if you're not collecting data quickly enough, your results will be way off.

Here's your fatal flaw; the 610W peak power draw number that you've been referencing this entire time is the number displayed on the primary side of the digital power meter that the website is using.

Quote

The Chroma Digital Power Meter never shows more than 610W, and it usually stays well below that number.

And the power draw on the primary side does not equate to the power draw of the total system. To make it easier to follow, let's use two R9 295X2s as an example. Using a 1000W beQuiet! unit, the same digital power meter doesn't show more than 1100W on the primary side, meaning that the system's total power consumption with two R9 295X2s (essentially four R9 290Xs) is under 1000W. That's rather impressive.

 

For the sake of it, I'll quote Tom's even more:

Quote

If the 430W average power consumption measurement from our gaming test proves correct, then an available maximum of 600W should be enough for the graphics card, a decent motherboard, a Core i7-4770K at stock speed, 8GB of RAM and an SSD... The PSU does prove ample.

So the article still managed to get a full fat system that included a i7-4770K and a R9 295X2 to run on a 600W unit without any problems. Why? Because the total system power draw was less than 600W. If it was over, it would trigger the PSU's protections and it would shut off.

12 hours ago, Syntaxvgm said:

I'm not manipulating facts, these are facts. I'm presenting them as I find them, since you keep wanting to argue this stupid argument. You cannot run a 295x2 from a 650w psu. It draws 507 watts when in full use. My original scenario was pairing it with an 8350 and a normal overclock. People always overclocked those things, they sucked unless you did. BUt your scenario for some reason was a 4770k and that disproved me somehow, and no 610 watt draw when gaming on a 650w psu is not doable, because gaming isn't peak draw, and even if it was that's beyond stupid. 

 

These are facts, I didn't manipulate them, they are what they are. My agenda is to drill them in your head. 

I'm not trying to argue that a R9 295X2 should be run on a 650W system, but rather the notion that it's doable under normal gaming conditions. Of course, getting an overclocked power-hungry chip FX-8350 is not going to do anything but further add on the load. But like I said, why would you pair a R9 295X2 with a FX-8350 in the first place if there are better, less power-hungry options available?

 

Bottom line is that the facts you've presented are flawed because they've been understandably misinterpreted.

[Hunter259]

4 minutes ago, HKZeroFive said:

-snip

Alright, Skylake-X aside, since power consumption on those chips alone is ridiculous, you would be hard pressed to find any dual GPU system that requires a 1000W unit. Before Skylake-X, any system with dual GTX 980Tis would only require a 850W unit max... even a 750W one would have been fine.

So yes, beyond a niche market, nobody needs a 1000W unit. Seems to fit into my claim of "most rigs don't require 1000+W power supplies...".

-snip

No one here is talking about stock vs stock but you. Yes you can run dual 980ti's and a 6 core on an 850w. But overclock it and you will be hitting around 800+w which is what the majority here are saying. When I ran dual OC'ed 980tis (115-120ish% power limits) and my 2600k I was pulling 800ish watts off that. At stock I would be sitting around 650-700. So yes you can TECHNICALLY do it but most enthusiasts aren't running things at stock. And this is completely ignoring stock custom cards. With those you might be able to get back up to 750.

[WereCat]

11 minutes ago, HKZeroFive said:

400W is absurdly high for a GTX 1080Ti by itself. Considering around 400W is normal for total system power consumption figures (including an overclocked GTX 1080Ti and a HEDT CPU), most people would be fine with a decent 550W unit... I have a mate running a i7-5820K with an overclocked GTX 1080Ti on a Seasonic 550G.

 

1

You are underestimating how much more power the card draws just from VRAM OC.

I was hitting the 127% (355W) power limit on my FTW3 before reaching 1.062V on core with overclocked VRAM (+700MHz / +1400MHz effective) causing the core clock fluctuate heavily (by +50MHz to +100MHz) depending on load which effectively negated my core OC.

Unlocking power limit allowed me to go all the way up to 1.093V and higher and achieving +50MHz better core OC than I could do before with the difference that now the core clock won't even flinch under the load which greatly improved minimum FPS. The card alone now goes well into the 400W territory.

[HKZeroFive]

10 minutes ago, Hunter259 said:

No one here is talking about stock vs stock but you. Yes you can run dual 980ti's and a 6 core on an 850w. But overclock it and you will be hitting around 800+w which is what the majority here are saying. When I ran dual OC'ed 980tis (115-120ish% power limits) and my 2600k I was pulling 800ish watts off that. At stock I would be sitting around 650-700. So yes you can TECHNICALLY do it but most enthusiasts aren't running things at stock. And this is completely ignoring stock custom cards. With those you might be able to get back up to 750.

I know. If I tried to argue about stock clocked cards, I would be screaming much lower numbers

 

With a custom BIOS, I managed to get 1502MHz core on my own GTX 980Ti, and together with my i7-4790K at 4.5GHz (admittedly, a rather average overclock) they typically draw around 400W in my own test runs. I'd imagine that adding a similarly clocked card would remain within the limitations of a 850W power supply unit.

[MageTank]

2 hours ago, WereCat said:

You are underestimating how much more power the card draws just from VRAM OC.

I was hitting the 127% (355W) power limit on my FTW3 before reaching 1.062V on core with overclocked VRAM (+700MHz / +1400MHz effective) causing the core clock fluctuate heavily (by +50MHz to +100MHz) depending on load which effectively negated my core OC.

Unlocking power limit allowed me to go all the way up to 1.093V and higher and achieving +50MHz better core OC than I could do before with the difference that now the core clock won't even flinch under the load which greatly improved minimum FPS. The card alone now goes well into the 400W territory.

You should get the XOC bios on it, and see what power draw really looks like without any power limit, lol. Can push voltage up to 1193 i think, 100mv higher than normal. I have the XOC bios running on my 1080 Ti FTW3 Hybrid, and it certainly helps get the clocks up assuming you can keep it all cool. Sadly, I don't see the practicality of running voltage that high, for 100mhz higher clock speeds. 

 

As for the rest of you arguing over power draw and PSU requirements, need I remind any of you how efficiency curves work? People dictating to each other that 1000w+ PSU's are worthless, seem to have no idea why these PSU's exist in the first place. If everyone would just step back for a second, ignore the worthless TDP numbers, and actually think about why having a PSU that is a little bigger than what you need is helpful, this thread would be a few pages shorter. 

[WereCat]

9 minutes ago, MageTank said:

You should get the XOC bios on it, and see what power draw really looks like without any power limit, lol. Can push voltage up to 1193 i think, 100mv higher than normal. I have the XOC bios running on my 1080 Ti FTW3 Hybrid, and it certainly helps get the clocks up assuming you can keep it all cool. Sadly, I don't see the practicality of running voltage that high, for 100mhz higher clock speeds. 

 

As for the rest of you arguing over power draw and PSU requirements, need I remind any of you how efficiency curves work? People dictating to each other that 1000w+ PSU's are worthless, seem to have no idea why these PSU's exist in the first place. If everyone would just step back for a second, ignore the worthless TDP numbers, and actually think about why having a PSU that is a little bigger than what you need is helpful, this thread would be a few pages shorter. 

I do use the XOC BIOS. I am on air cooling though so using high voltage is a no no for me.

Besides, I can barely get any higher clocks regardless if I am at 1.093V or higher because of temperatures.

[Dylanc1500]

13 minutes ago, MageTank said:

You should get the XOC bios on it, and see what power draw really looks like without any power limit, lol. Can push voltage up to 1193 i think, 100mv higher than normal. I have the XOC bios running on my 1080 Ti FTW3 Hybrid, and it certainly helps get the clocks up assuming you can keep it all cool. Sadly, I don't see the practicality of running voltage that high, for 100mhz higher clock speeds. 

 

As for the rest of you arguing over power draw and PSU requirements, need I remind any of you how efficiency curves work? People dictating to each other that 1000w+ PSU's are worthless, seem to have no idea why these PSU's exist in the first place. If everyone would just step back for a second, ignore the worthless TDP numbers, and actually think about why having a PSU that is a little bigger than what you need is helpful, this thread would be a few pages shorter. 

This is why I am running a 3000w power supply, way overkill. I typically will draw around 1500w.

[MageTank]

4 minutes ago, WereCat said:

I do use the XOC BIOS. I am on air cooling though so using high voltage is a no no for me.

You are not really missing much. Highest I can get to on this AIO is 2200mhz, which hits about 50C under superposition, but scaling stops hard after 2125-2150 depending on benchmark. So I normally leave mine at 2100 if I need the extra frames. When I am running easier titles, I just load the stock XOC profile which locks it at 2000mhz, which is plenty fast enough for 99% of the titles I play at 1440p/165hz. My point being, after using XOC on a single 1080 Ti and 5ghz 7700k, I can push a 650w supernova pretty hard. I completely understand people wanting to use 1000w units with a similar setup to achieve 50% curve under constant load. It's not only easier on the PSU from a longevity stance, but it gives them room to expand in the future, or to continue with what they have as the capacitors age. People that can get away with pushing say, 500w on a 550/600w PSU, will likely have to upgrade PSU's long before those that got overkill units in the first place. The only time it's an inherently bad idea to get an overkill PSU, is if you cannot fall within it's peak efficiency curve at any time. Then, you are better off getting something suited to your levels of load. 

 

Hopefully the guys arguing over TDP and PSU requirements figure that out eventually. 

4 minutes ago, Dylanc1500 said:

This is I am running a 3000w power supply, way overkill. I typically will draw around 1500w.

Wow, my house won't even let me pull 2000 from a single wall, let alone 3000, lol. 

[WereCat]

13 minutes ago, MageTank said:

You are not really missing much. Highest I can get to on this AIO is 2200mhz, which hits about 50C under superposition, but scaling stops hard after 2125-2150 depending on benchmark. So I normally leave mine at 2100 if I need the extra frames. When I am running easier titles, I just load the stock XOC profile which locks it at 2000mhz, which is plenty fast enough for 99% of the titles I play at 1440p/165hz. My point being, after using XOC on a single 1080 Ti and 5ghz 7700k, I can push a 650w supernova pretty hard. I completely understand people wanting to use 1000w units with a similar setup to achieve 50% curve under constant load. It's not only easier on the PSU from a longevity stance, but it gives them room to expand in the future, or to continue with what they have as the capacitors age. People that can get away with pushing say, 500w on a 550/600w PSU, will likely have to upgrade PSU's long before those that got overkill units in the first place. The only time it's an inherently bad idea to get an overkill PSU, is if you cannot fall within it's peak efficiency curve at any time. Then, you are better off getting something suited to your levels of load. 

 

2

I can push about 2067MHz-2088MHz core max with what I've got which is plenty for me (also 165Hz 1440p). I am happy that the clocks no longer fluctuate as they did without XOC which had a huge impact on minimum framerates.

Now I can finally enjoy the benefits of higher VRAM clocks as before the VRAM OC affected my core clocks because of power limit.

 

I actually had 550W SuperFlower unit until I started to trip the overcurrent protection when I pushed the card so I upgraded to 750W G3.

[Dylanc1500]

46 minutes ago, MageTank said:

Wow, my house won't even let me pull 2000 from a single wall, let alone 3000, lol. 

I ran a dedicated 220v circuit to my office as I have a 5000VA UPS with things connected that are pretty critical for work. Although I have never seen my workstation actually pull more than 1500w from the plug.

[leadeater]

7 hours ago, tbake0155 said:

So although TDP != power draw, we can simplify and say that TDP == power draw.  That's why it is used as a simplification for the masses, who generally don't care about having an expert's understanding of TDP. 

That would be mostly safe for the standard desktop range of CPUs but for HEDT it is not. When something is so different, TDP vs power draw, you can not safely use it as an indicator of power usage.

 

Also it's not used as a simplification for the masses it's used as a technical specification for cooler manufacturers, it became a marketing tool later but the actual usage of it never changed.

 

If you want to know how much power something uses read a review, no understanding required. They tell you the actual figures you need to know.

[mr moose]

8 hours ago, tbake0155 said:

While it is true that TDP is not the power used by the CPU, it is also true that all of the power that goes into the CPU is eventually released as heat. 

 

So although TDP != power draw, we can simplify and say that TDP == power draw.  That's why it is used as a simplification for the masses, who generally don't care about having an expert's understanding of TDP. 

 

nope.  Apart from the fact that not all power drawn by a CPU comes out as heat (some has to drive outputs).  TDP is not a figure that is meant to represent power draw in any meaningful way.  If you where to do that you would be severely under powering your CPU, especially in the case of overclocking (or even in some cases just using the factory boost under load).

 

This is a topic I have debated one too many times on this forum.  It is time for Linus to do a techquicky.  For majority of users (even enthusiasts) it is a number that can largely be ignored.

[Max_Settings]

On 9/27/2017 at 4:08 AM, HKZeroFive said:

However, with the Corsair H105, the chip managed to stay at a rather cool 63C during benchmarking. Impressive.

Bull shit. That's about what my H105 keeps my 4790K at while overclocked. There is no way that's right.

[Ithanul]

8 hours ago, Hunter259 said:

No one here is talking about stock vs stock but you. Yes you can run dual 980ti's and a 6 core on an 850w. But overclock it and you will be hitting around 800+w which is what the majority here are saying. When I ran dual OC'ed 980tis (115-120ish% power limits) and my 2600k I was pulling 800ish watts off that. At stock I would be sitting around 650-700. So yes you can TECHNICALLY do it but most enthusiasts aren't running things at stock. And this is completely ignoring stock custom cards. With those you might be able to get back up to 750.

O god, yep, I have 980Tis...OC, and do a modded BIOS, those cards become power hogs. 

 

On 9/27/2017 at 1:29 PM, Tam3n said:

Well Chinebench takes like what, 4 seconds to complete with that setup, so the cpu & vrm doesn't have the time to get very hot. Continued 100 % load on an AIO no snow balls chance in hell. Might not be doable even on a high end custom loop and monoblock.

 

Anything draws high watts when oc'ed balls to the walls. Hell, I've put 800 W through the pcie cables to my R9 290 @ 1.6 V (VRM Vout).

Agree, I highly doubt the chip can go 24/7 on that setup.

I run my 5960X around 4.3-4.4GHz 24/7 doing BOINC.  I can tell you right now....those chips get mighty toasty and are firing breathing dragons at those clocks.  It easily gets over 65-75C and that with two GPUs on idle and with one 360mm and one 200mm radiator in the loop.

18 cores going at 4.9GHz....I would not even do that with water cooling.

 

On the debate of power, I tripped a 900 Watt USP with my main rig, and that when it had a stock 4770K and two OC 980Tis.  That with the CPU not doing anything.  So, a 295x2 I can sure bet can pull some serious watts when OC and being seriously pushed.

[ionbasa]

12 hours ago, tbake0155 said:

While it is true that TDP is not the power used by the CPU, it is also true that all of the power that goes into the CPU is eventually released as heat. 

 

So although TDP != power draw, we can simplify and say that TDP == power draw.  That's why it is used as a simplification for the masses, who generally don't care about having an expert's understanding of TDP. 

 

Not necessarily. Power in =/= Heat out. There's an efficiency factor attached to it. Think of the TDP as the average heat output at standard conditions (Conduction + Radiative, convection is mostly irrelevant until the heat transfers to a fin array). If anything TDP is more of a 'ballpark number' to help heat sink manufacturers. I say ballpark because TDP does vary based on a number of conditions such as: clock speed, AVX instruction sets, voltages, OC, and more.

 

So while you may be pushing 700W to the CPU, you may only get 150W out as heat. That leaves 550W 'unaccounted' for, but the output power of the device is transferred elsewhere in some useful form, such as: flipping bits, driving gates, transistors, etc. Those devices use energy to do work.  So thats where the rest of the power discrepancies is from.