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

[AnonymousGuy]

1. Cinebench is the least taxing benchmark they could probably have done.

 

2. Holy shit linpack on this thing is going to trip residential circuit breakers.  I'm hard as a rock right now.  I'm finally going to beat my previous record of 1300W from the wall.

[AnonymousGuy]

1 hour ago, ionbasa said:

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.

All energy becomes heat at the other end.  Flipping transistors requires electron-friction.

[Jito463]

On 9/28/2017 at 11:41 AM, MageTank said:

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.

Before I built my Ryzen system, my previous PSU was a PC P&C 610 WT Silencer that lasted me for around 8 years - across multiple builds - and was still working perfect when I sold it with my old FX-8370 system (I almost regret giving it up, PC P&C made incredible PSUs in their day).  Whether or not a 600 WT PSU would need to be replaced depends greatly on the original quality of it.

[Prysin]

On 9/27/2017 at 12:09 PM, Syntaxvgm said:

yah

I just wanna know if we're gonna have this problem again 

we're already there. People are saying you cannot run these chips without 8+8 pins and beefy as fuck VRMs

[Dabombinable]

7 hours ago, Prysin said:

we're already there. People are saying you cannot run these chips without 8+8 pins and beefy as fuck VRMs

Still don't know why manufacturers won't use the older+superior copper fins (some times gold plated to prevent corrosion/discolouring)+heatpipe design to cool the VRM of the X299 boards. Its not like they are budget products and they really do need something far better than what we see on Z87 through Z270 considering how much the new X series consumes.

[Nicnac]

crash and burn literally

[MageTank]

9 minutes ago, Dabombinable said:

Still don't know why manufacturers won't use the older+superior copper fins (some times gold plated to prevent corrosion/discolouring)+heatpipe design to cool the VRM of the X299 boards. Its not like they are budget products and they really do need something far better than what we see on Z87 through Z270 considering how much the new X series consumes.

People are already not putting airflow over top of the aluminum fins we have. Is it really going to be a good idea to change them to copper, which holds on to heat longer than aluminum? lol.

 

Take into consideration that a lot of reviewers are testing these CPU's on open air test benches with AIO's and zero airflow over the VRM. Most cases tends to have a fan at the top, pulling air out through the top near the VRM area. If not, you can typically rig something up to provide airflow to the VRM area to deal with most of the heat. The material used is not a problem at all, but rather, the lack of airflow (and in the case of the original X299 Apex, lack of fins in the first place). Once companies stop treating these heatsinks as decorations, and more as a functional part of the motherboard, we might finally see this problem disappear altogether. Perhaps we need to go back to integrated fans on the VRM until then. 

[Dabombinable]

2 minutes ago, MageTank said:

People are already not putting airflow over top of the aluminum fins we have. Is it really going to be a good idea to change them to copper, which holds on to heat longer than aluminum? lol.

 

Take into consideration that a lot of reviewers are testing these CPU's on open air test benches with AIO's and zero airflow over the VRM. Most cases tends to have a fan at the top, pulling air out through the top near the VRM area. If not, you can typically rig something up to provide airflow to the VRM area to deal with most of the heat. The material used is not a problem at all, but rather, the lack of airflow (and in the case of the original X299 Apex, lack of fins in the first place). Once companies stop treating these heatsinks as decorations, and more as a functional part of the motherboard, we might finally see this problem disappear altogether. Perhaps we need to go back to integrated fans on the VRM until then. 

Thing is, on the open air test benches the heat can dissipate more easily than if it were in a case with poor airflow, and what aluminium fins are you talking about? Even on Z97 boards the only one I've seen with proper fins are the Sabertooth MKI with the "armour" and fans integrated into it. Most boards just use blocks with a varying amount of aluminium cut out.

[MageTank]

Just now, Dabombinable said:

Thing is, on the open air test benches the heat can dissipate more easily than if it were in a case with poor airflow, and what aluminium fins are you talking about? Even on Z97 boards the only one I've seen with proper fins are the Sabertooth MKI with the "armour" and fans integrated into it. Most boards just use blocks with a varying amount of aluminium cut out.

I disagree with this. Poor airflow over the VRM is still better than no airflow over the VRM at all. Unless you are using an ancient Optiplex case from 2005, most modern cases at least have one top mounted fan near the VRM. Even if they don't, people that cheap out on cases tend to also cheap out on coolers, and go with an air cooler that will at least get some air around the VRM. Since we are talking in the context of X299, it's safe to assume we are not talking about cheap coolers or cheap cases. Find me an expensive case that doesn't have top-oriented airflow.

 

As for the fins I am referring to, take your pick: 

https://www.asus.com/us/Motherboards/ROG-RAMPAGE-VI-APEX/overview/

https://www.newegg.com/Product/Product.aspx?Item=N82E16813119017&ignorebbr=1&nm_mc=AFC-C8Junction&cm_mmc=AFC-C8Junction-PCPartPicker, LLC-_-na-_-na-_-na&cm_sp=&AID=10446076&PID=3938566&SID=

https://www.evga.com/products/product.aspx?pn=131-SX-E295-KR

 

In the case of EVGA, they are coming out with 2 more boards (X299 Dark, and X299 FTW-K), which use a horizontal fin array like the Micro listed above. While they may not be extremely dense fin arrays you see on traditional CPU heatsinks, they are still fins nonetheless, and work better when accompanied with airflow. Switching to copper will do absolutely nothing if people refuse to channel at least a little bit of air through them, and if anything, it would end up worse for those that refuse to do so. @done12many2 has a tiny little fan on his VRM's, and that is more than enough to keep his Apex cool, even under heavy load from a 4.8ghz 7920X. ASUS came a very long way from their original design, which lacked fins entirely, and did not come with a little fan. 

[dexT]

22 minutes ago, Dabombinable said:

Thing is, on the open air test benches the heat can dissipate more easily than if it were in a case with poor airflow, and what aluminium fins are you talking about? Even on Z97 boards the only one I've seen with proper fins are the Sabertooth MKI with the "armour" and fans integrated into it. Most boards just use blocks with a varying amount of aluminium cut out.

My ASUS X99M-WS has them.

[MageTank]

5 minutes ago, dexT said:

My ASUS X99M-WS has them.

Those X99 WS boards are also arguably the sexiest motherboards that I have ever seen. Would love to see ASUS translate that exact design over to X299. 

[dexT]

6 minutes ago, MageTank said:

Those X99 WS boards are also arguably the sexiest motherboards that I have ever seen. Would love to see ASUS translate that exact design over to X299. 

They perform very well too. Almost zero vdroop at extreme settings with no LLC. 

 

It's funny when your mATX board has more EPS connectors than the flagship EATX Rampage board.

[leadeater]

8 hours ago, Dabombinable said:

Still don't know why manufacturers won't use the older+superior copper fins (some times gold plated to prevent corrosion/discolouring)+heatpipe design to cool the VRM of the X299 boards.

 

8 hours ago, MageTank said:

Take into consideration that a lot of reviewers are testing these CPU's on open air test benches with AIO's and zero airflow over the VRM. Most cases tends to have a fan at the top, pulling air out through the top near the VRM area.

The airflow is an issue but it's not actually the real cause of them problem. There are not enough phases on the early boards that were used for reviews which caused each phase to deliver current outside of the optimal efficiency range causing a large amount of heat to be produced.

 

There is a steady drop in efficiency and therefore increase in heat output from the VRMs as the current increases, you could add better cooling to try and fix this or add phases so less heat is actually produced. Adding phases is the more expensive fix but really for how much these boards cost add the phases.

 

If you have 8 phases for the CPU and vcore of 1.1v and the total package power is 300W that is 273A (34A per phase) which is already well outside the recommended current for the mosfets in use, overclocking is just pushing them more in to thermal runaway territory. Simply adding on 2 more phases halves the heat output of the VRMs, I really do not think 10 phase power delivery for the CPU is too much to ask for.

[Dabombinable]

1 hour ago, leadeater said:

 

The airflow is an issue but it's not actually the real cause of them problem. There are not enough phases on the early boards that were used for reviews which caused each phase to deliver current outside of the optimal efficiency range causing a large amount of heat to be produced.

 

There is a steady drop in efficiency and therefore increase in heat output from the VRMs as the current increases, you could add better cooling to try and fix this or add phases so less heat is actually produced. Adding phases is the more expensive fix but really for how much these boards cost add the phases.

 

If you have 8 phases for the CPU and vcore of 1.1v and the total package power is 300W that is 273A (34A per phase) which is already well outside the recommended current for the mosfets in use, overclocking is just pushing them more in to thermal runaway territory. Simply adding on 2 more phases halves the heat output of the VRMs, I really do not think 10 phase power delivery for the CPU is too much to ask for.

Perhaps the old 16 phase setup would work best. My P5Q Deluxe has a 16 phase VRM (and people say its a doubled design not 16 true phases, but I haven't found any evidence either way) and the VRM stays rather cool-the heatsink is the warmest above the Northbridge (which was under an active heatpiped cooler on the P5 Turbo)

Edit: Integrating the NB into the CPU should have made motherboards a bit cheaper+less complex in their design thus making them cheaper. And with what we have now they seem to be ripoffs. My old board was 119 pounds when it was reviewed in 2011 (which would have been towards the end of its life) http://www.trustedreviews.com/reviews/asus-p5q-deluxe

Edit 2: It was around 122 pounds including VAT during 2007: https://www.bit-tech.net/reviews/tech/motherboards/asus-p5q-deluxe-intel-p45-has-arrived/1/

[Prysin]

10 hours ago, Dabombinable said:

Still don't know why manufacturers won't use the older+superior copper fins (some times gold plated to prevent corrosion/discolouring)+heatpipe design to cool the VRM of the X299 boards. Its not like they are budget products and they really do need something far better than what we see on Z87 through Z270 considering how much the new X series consumes.

they need active cooling, regardless what kind of heatspreader you put on them. Reason: Most people uses water to cool those CPUs, thus no airflow for the mobo. Mobo must supply its own.

[Tam3n]

I imagine most potential buyers will use water cooling for this cpu. If you don't use a monoblock, or water block for the VRM, I think you would need a dedicated fan rigged to blow right at the VRM heatsink, if you plan to do moderate overclock that's usable at the task this cpu is designed for... It's going to be interesting to see how the different VRM designs will behave, and what kind of heat sinks will be used.

[ATFink]

On 9/27/2017 at 2:30 AM, LunaP0n3 said:

Not to mention a motherboard has to handle the crazy amount of power going to the chip.

RIP VRMs

[ATFink]

On 9/27/2017 at 4:33 AM, Bananasplit_00 said:

there are 2WK PSUs, i cant say you would need more then that for any single CPU + dual GPU setup which is what we are moveing towards

i9 7980XE + dual Vega 64 all OC'd on LN2 as far as possible. Actually would that be 2kW? It's gotta be close if not more. Assuming >1kW for 7980XE and >500W per Vega 64 that may be possible. Only like 0.01% or less of people that tinker with PCs would ever do that though, so who knows.

[JediFragger]

Fuuuuuuuuuuuuuuuuuuuuuuuck.....................

[dexT]

10 hours ago, Dabombinable said:

Perhaps the old 16 phase setup would work best. My P5Q Deluxe has a 16 phase VRM (and people say its a doubled design not 16 true phases, but I haven't found any evidence either way) and the VRM stays rather cool-the heatsink is the warmest above the Northbridge (which was under an active heatpiped cooler on the P5 Turbo)

That's a sick board, same VRM layout as my Maximus II Formula and my dearly departed X48 Rampage Extreme. I wanted the P5Q3 D but some new-old-stock GA-EP45T-UDR3 just hit ebay and I caved.

 

EDIT: Noooooooooooo just got a message that they sold them all and there was an ebay error

[DXMember]

On ‎2017‎.‎09‎.‎27‎. at 1:40 PM, HKZeroFive said:

None of those systems will require a 1000W power supply. Look up power consumption benchmarks.

that's at stock

the 7980XE at stock doesn't require a 1kW PSU either.

It's when you start overclocking and having multi-GPU setup

[Dabombinable]

3 hours ago, dexT said:

That's a sick board, same VRM layout as my Maximus II Formula and my dearly departed X48 Rampage Extreme. I wanted the P5Q3 D but some new-old-stock GA-EP45T-UDR3 just hit ebay and I caved.

 

EDIT: Noooooooooooo just got a message that they sold them all and there was an ebay error

Unluckily for me, the board only runs up to 4GB of DDR2 1066 in a 2+1+1 configuration (if it decides to at all), and is stuck at 8GB of RAM as its limit (not the stated 16GB). They kind of work best with unlocked CPU (they can even give you grief with a 1600MHz FSB)

[The Owl]

If we counted the numbers of Watts that people said after reading this title, the total consumption would be over 9000.

 

Yeah I know .. shocking joke.

[Potato*Salad]

FX 9590 is suddenly a very efficient CPU..