Hot things get hot! cont. 7980XE, 4.8GHz, and hell's fire.

[leadeater]

4 hours ago, MageTank said:

Either way, X299's VRM "disaster" is smoke and mirrors, and the only people complaining about thermals, are the people that think small AIO's are worthy of cooling 10c+ CPU's while still overclocking them.

Motherboard makers release badly designed products, Intel's fault heh. It's not like AIO cooling is even part of the discussion too if people bring it up, AIO doesn't cool the VRMs so what does an AIO not being enough to cool these chips have to do with that at all anyway?

 

One of the most important factors for VRM cooling is operating efficiency, this plays a very big part in what is required to keep them cool and how much power draw from the PSU is required.

 

Here is an efficiency graph and shows why more VRM phases are better, I suspect most people don't actually know why or think it's something else which isn't exactly correct.

Quote

Efficiency graph of a MOSFET used on the Gigabyte Z97X-SOC Force motherboard

https://www.ekwb.com/blog/what-is-mosfet-and-why-should-you-keep-it-cool/

 

As you can clearly see as you draw more current the efficiency goes down and this is why more VRM phases is better, each VRM phase has to deliver less current as more are added reducing the overall circuit power and heat output.

 

A CPU at 1.2V drawing 300W amps would require 250 amps, an 8 phase design would mean 31.25 amps per phase at roughly 87% efficiency and 6W of heat output per phase totaling 48W of heat. Going from 8 to 10 phases would drop the heat output to about 40W. This might not seem like a lot but remember this is the same amount of heat many many mainstream desktop processors are putting out.

 

Lowering your vcore voltage while drawing the same amount of power will increase your VRM temps.

[Lays]

2 minutes ago, leadeater said:

Motherboard makers release badly designed products, Intel's fault heh. It's not like AIO cooling is even part of the discussion too if people bring it up, AIO doesn't cool the VRMs so what does an AIO not being enough to cool these chips have to do with that at all anyway?

 

Ask a power supply maker to comment on how much it takes to cool VRMs, cooling VRMs isn't exactly new and bleeding edge even at these power draws. Ask a Class A power amplifier maker what it takes to cool a 250W per channel amp, Class A is roughly 30% efficient for output power so doing the math is easy.

 

That is how much it takes to keep 583W (per channel) cool passively, so a single heatsink section as depicted in the above picture can cool 194W without a fan. Notice just how freaking huge that is . Not that a power amp is directly applicable but represents very well how to keep hot things cool.

 

One of the most important factors for VRM cooling is operating efficiency, this plays a very big part in what is required to keep them cool and how much power draw from the PSU is required.

 

Here is an efficiency graph and shows why more VRM phases are better, I suspect most people don't actually know why or think it's something else which isn't exactly correct.

https://www.ekwb.com/blog/what-is-mosfet-and-why-should-you-keep-it-cool/

 

As you can clearly see as you draw more current the efficiency goes down and this is why more VRM phases is better, each VRM phase has to deliver less current as more are added reducing the overall circuit power and heat output.

 

A CPU at 1.2V drawing 300W amps would require 250 amps, an 8 phase design would mean 31.25 amps per phase at roughly 87% efficiency and 6W of heat output per phase totaling 48W of heat. Going from 8 to 10 phases would drop the heat output to about 40W. This might not seem like a lot but remember this is the same amount of heat many many mainstream desktop processors are putting out.

 

Lowering your vcore voltage while drawing the same amount of power will increase your VRM temps.

How much Kool-aid did you drink today bro

[leadeater]

2 minutes ago, Lays said:

How much Kool-aid did you drink today bro

Too much ofc 

 

But hey if someone finds that useful then cool.

[MageTank]

3 minutes ago, leadeater said:

Motherboard makers release badly designed products, Intel's fault heh. It's not like AIO cooling is even part of the discussion too if people bring it up, AIO doesn't cool the VRMs so what does an AIO not being enough to cool these chips have to do with that at all anyway?

 

Ask a power supply maker to comment on how much it takes to cool VRMs, cooling VRMs isn't exactly new and bleeding edge even at these power draws. Ask a Class A power amplifier maker what it takes to cool a 250W per channel amp, Class A is roughly 30% efficient for output power so doing the math is easy.

 

That is how much it takes to keep 583W (per channel) cool passively, so a single heatsink section as depicted in the above picture can cool 194W without a fan. Notice just how freaking huge that is . Not that a power amp is directly applicable but represents very well how to keep hot things cool.

 

One of the most important factors for VRM cooling is operating efficiency, this plays a very big part in what is required to keep them cool and how much power draw from the PSU is required.

 

Here is an efficiency graph and shows why more VRM phases are better, I suspect most people don't actually know why or think it's something else which isn't exactly correct.

https://www.ekwb.com/blog/what-is-mosfet-and-why-should-you-keep-it-cool/

 

As you can clearly see as you draw more current the efficiency goes down and this is why more VRM phases is better, each VRM phase has to deliver less current as more are added reducing the overall circuit power and heat output.

 

A CPU at 1.2V drawing 300W amps would require 250 amps, an 8 phase design would mean 31.25 amps per phase at roughly 87% efficiency and 6W of heat output per phase totaling 48W of heat. Going from 8 to 10 phases would drop the heat output to about 40W. This might not seem like a lot but remember this is the same amount of heat many many mainstream desktop processors are putting out.

 

Lowering your vcore voltage while drawing the same amount of power will increase your VRM temps.

I was trying to process all of this information, but then I overheated  

[leadeater]

1 minute ago, MageTank said:

I was trying to process all of this information, but then I overheated  

Heh judging by these two last comments I need to shorten it lol. Oh well important part is only the last half I'll just wipe the rest.

[Lays]

Just now, leadeater said:

Heh judging by these two last comments I need to shorten it lol. Oh well important part is only the last half I'll just wipe the rest.

I get what you mean tho, about the whole efficiency range etc.

[MageTank]

Just now, leadeater said:

Heh judging by these two last comments I need to shorten it lol. Oh well important part is only the last half I'll just wipe the rest.

I skimmed through it. I understand the vast majority of it, my post was just poorly worded. By the VRM "disaster" being smoke and mirrors, I meant the video of it being potentially disastrous for the end consumer. We have to remember that a lot of safety mechanisms were turned off, including current limitations (current being the far bigger culprit, not voltage, as you are fully aware) in order to produce these "dangerous" levels. I also firmly stated several times over on this very forum that active VRM cooling was going to become a thing again. This platform as a whole, is not intended for people that want to scrape by with cheap AIO's and budget air coolers. 

 

I also recall scolding Intel on their decision to not solder the CPU's, and I was also very hard on the board manufacturers who decided to spend more money on aesthetics, and less on functional VRM cooling. I mean, ASUS added active fans to the TR boards. That would have been a lovely idea to have for the X299 boards as well. Would make sense for them to bring back that VRM monoblock board as well for X299, but we will have to wait and see on that one. Never really made sense to have it on the small consumer chips in the first place, but I can certainly see it's appeal now. 

[leadeater]

7 minutes ago, MageTank said:

I skimmed through it. I understand the vast majority of it, my post was just poorly worded. By the VRM "disaster" being smoke and mirrors, I meant the video of it being potentially disastrous for the end consumer. We have to remember that a lot of safety mechanisms were turned off, including current limitations (current being the far bigger culprit, not voltage, as you are fully aware) in order to produce these "dangerous" levels. I also firmly stated several times over on this very forum that active VRM cooling was going to become a thing again. This platform as a whole, is not intended for people that want to scrape by with cheap AIO's and budget air coolers. 

 

I also recall scolding Intel on their decision to not solder the CPU's, and I was also very hard on the board manufacturers who decided to spend more money on aesthetics, and less on functional VRM cooling. I mean, ASUS added active fans to the TR boards. That would have been a lovely idea to have for the X299 boards as well. Would make sense for them to bring back that VRM monoblock board as well for X299, but we will have to wait and see on that one. Never really made sense to have it on the small consumer chips in the first place, but I can certainly see it's appeal now. 

Yea I agree with your statement, was really just giving information below that about how much heat is being generated by the VRMs. The only disaster was motherboard makers failing at math for how much cooling would be required.

 

Any idea how many phases were on the review motherboard for X299? Actually finding it hard to source that info, short of just guessing by counting chokes.

 

Edit:

Looks like most were 8 phases and all the new boards coming out are 13 phases.

[Lays]

1 minute ago, leadeater said:

Yea I agree with your statement, was really just giving information below that about how much heat is being generated by the VRMs. The only disaster was motherboard makers failing at math for how much cooling would be required.

 

Any idea how many phases were on the review motherboard for X299? Actually finding it hard to source that info, short of just guessing by counting chokes.

Pretty sure it was 7+1.  I don't think it was motherboard manufacturers failing at math, I think it was them trying to cheap out on things and put RGB's everywhere. These companies have been making boards for YEARS now that didn't have issues, remember X58 with huge heatsinks, boatloads of VRM? They're perfectly capable of developing beastmode boards.

[leadeater]

2 minutes ago, Lays said:

Pretty sure it was 7+1.  I don't think it was motherboard manufacturers failing at math, I think it was them trying to cheap out on things and put RGB's everywhere. These companies have been making boards for YEARS now that didn't have issues, remember X58 with huge heatsinks, boatloads of VRM? They're perfectly capable of developing beastmode boards.

Looks like everything coming out now is 13 so has basically halved the heat output of the power delivery.

[MageTank]

17 minutes ago, leadeater said:

Yea I agree with your statement, was really just giving information below that about how much heat is being generated by the VRMs. The only disaster was motherboard makers failing at math for how much cooling would be required.

 

Any idea how many phases were on the review motherboard for X299? Actually finding it hard to source that info, short of just guessing by counting chokes.

The one that was overheating in Der8auers video? That was 7+1. The Apex that was tested with the 7980XE in this thread is 8 phases (ASP 1405 + IR3555). 

 

EDIT: For anyone curious about any X299 VRM information with explicit details to the exact components, give this thread a read: http://www.overclock.net/t/1632665/intel-x299-socket-2066-vrm-thread

 

A ton of research (including from industry insiders) in that thread. 

[Bananasplit_00]

13 hours ago, MageTank said:

As I have said before, it is still using traditional compound (not solder) and most of this is owed to the larger die. Also as I said, the 12c chip is going to be king for people that want a mix of heavy multi-core performance AND higher clock speeds. It has the best of both worlds. 

 

For those of you that look at 90C on a custom loop and say "that's too hot", remind yourself of how many cores these CPU's have, and cross reference that with the heat produced by traditional consumer i7's, and tell me you honestly expected the 18 core to run anywhere near as cool as your non-soldered quad core. Better yet, get a 7700k (non-delidded), overclock it on a custom loop, and run Linpack MKL for me. I can't help but feel you will still hit 90C on a custom loop, lol.

 

Now just imagine what these bad boys can do with a delid and some liquid metal alongside these custom loops. With a low enough ambient, and strong enough pump, you might be able to make 5ghz happen on all 18 cores assuming the board doesn't flake out after the constant pressure over time. Safe to say the X299 Apex and X299 OC Formula are going to be required to hold that clock. I highly doubt it can be held on any of the sub $400 boards without some serious active cooling. Now you just have to find someone with the stones to actually delid a $2000+ CPU. 

there is an X299 OC Formula? awesome...

 

*edit*

never mind its not yellow...  god damn it ASRock stick to what your boards are suposed to be

[Sauron]

If this is done on an AiO cooler (which I STRONGLY doubt) then it's truly impressive and would show some massive improvement over everything we've seen from intel thus far.

 

However, if it's a custom loop there's no telling how many radiators the guy is using... and at that point a 4.8 GHz overclock may not sound as good when put into perspective with the price of the cooling solution. Still, it's good to know it's possible at least.

15 hours ago, MageTank said:

Same reason the entire LTT forum complains about the potential pitfalls of the enthusiast platform without ever owning/intending to buy said platform. Because they can 

 

Seriously though, a lot of people complain about X299 and threadripper without even having valid complaints, or an intent to even buy the product. Get's extremely old x.x

Well, if to criticize a product we had to buy it every time it would get a little complicated  if I can criticize, say, a pentium without having any intention of buying it (despite having more than enough money to do so), why shouldn't I criticize a part simply because I can't afford it? Besides, if we're talking about x299 in general, I might have been interested in the lower end options (which I could afford) - if they weren't terrible.

[done12many2]

35 minutes ago, Sauron said:

If this is done on an AiO cooler (which I STRONGLY doubt) then it's truly impressive and would show some massive improvement over everything we've seen from intel thus far.

 

However, if it's a custom loop there's no telling how many radiators the guy is using... and at that point a 4.8 GHz overclock may not sound as good when put into perspective with the price of the cooling solution. Still, it's good to know it's possible at least.

 

There's a point of diminishing return on any water loop regardless of how big you build it.  That is to say that 7980XE cooled by a dedicated 360mm rad may not cool any better than a loop with 5 x 360mm rads.  Sure the latter would allow you to run the fans at a much lower speed while retaining the same thermal capacity, but it does not mean that more radiators allow for a higher overclock.

 

The heat transfer between the die and the IHS and then IHS to water is the choke point and 12 radiator ain't going to change that. 

 

 

Quote

However, if it's a custom loop there's no telling how many radiators the guy is using... and at that point a 4.8 GHz overclock may not sound as good when put into perspective with the price of the cooling solution. Still, it's good to know it's possible at least.

Well, if to criticize a product we had to buy it every time it would get a little complicated  if I can criticize, say, a pentium without having any intention of buying it (despite having more than enough money to do so), why shouldn't I criticize a part simply because I can't afford it? Besides, if we're talking about x299 in general, I might have been interested in the lower end options (which I could afford) - if they weren't terrible.

 

The point that @MageTank is making is if one lacks interest, why complain?  If you aren't buying a Bugatti Veyron, why complain about it's gas mileage?  

 

The answer on LTT is often, "because I can!"

[Sauron]

3 minutes ago, done12many2 said:

 

There's a point of diminishing return on any water loop regardless of how big you build it.  That is to say that 7980XE cooled by a dedicated 360mm rad may not cool any better than a loop with 5 x 360mm rads.  Sure the latter would allow you to run the fans a much lower speed while retaining the same thermal capacity, but it does not mean that more radiators allow for a higher overclock.

 

The heat transfer between the die and the IHS and then IHS to water is the choke point and 12 radiator ain't going to change that. 

 

Regardless, a 360mm radiator dedicated to the cpu is quite  an expense on top of the cpu itself.

[done12many2]

Just now, Sauron said:

Regardless, a 360mm radiator dedicated to the cpu is quite  an expense on top of the cpu itself.

 

I would agree, but common sense kicks in here.   As @MageTank and others have stated, people who buy these chips aren't usually the kinds of folks that stick a $100 AIO on it.  Of the very few idiots that do, Darwin's is in effect.  

[Mr.Meerkat]

16 hours ago, MageTank said:

My small little post was about the people on this forum that constantly complain about X299's VRM being "explosive" or that it's a bad platform, without having any real experience with that product segment, or any intent to buy it.

Does this include the large number of people crying about the VRMs on B350 motherboards?  

[Sauron]

1 minute ago, done12many2 said:

 

I would agree, but common sense kicks in here.   As @MageTank and others have stated, people who buy these chips aren't usually the kinds of folks that stick a $100 AIO on it.  Of the very few idiots that do, Darwin's is in effect.  

Of course, but having decent temperatures at 4.8GHz with a dedicated 360mm rad is far less impressive than using an AiO.

[PCGuy_5960]

1 minute ago, Sauron said:

Of course, but having decent temperatures at 4.8GHz with a dedicated 360mm rad is far less impressive than using an AiO.

But much more realistic, people who buy the 7980XE aren't going to be using an AiO.

[Mr.Meerkat]

Just now, PCGuy_5960 said:

But much more realistic, people who buy the 7980XE aren't going to be using an AiO.

INB4 OEM builder pairs a 7980XE with an AiO...although with a 360/480mm rad  

[FakeNSA]

16 hours ago, DocSwag said:

The real question we need to know.... Is how much is the power draw?!?!?!?!?

To reach the same clock speed on two CPUs that are same arch but different core counts you should only need a similar amount of voltage (assuming temps are similar). What IS going to be SIGNIFICANTLY higher is current draw. This thing at those clock speeds is probably going to draw close to double of what a 4.6ghz 7900x is doing. AKA enough to make your pc a space heater.

@nicklmg @LinusTech @Slick Pizza warmer pc part 2? Please?

[done12many2]

3 minutes ago, Sauron said:

Of course, but having decent temperatures at 4.8GHz with a dedicated 360mm rad is far less impressive than using an AiO.

 

To be honest.  Temps from either Intel or AMD on the big core parts do not look great.  1950x CPUs are getting very hot on AIOs at stock speeds. 7980XE CPUs are going to get very hot on AIOs at stock speeds.

 

The big point is that enthusiast platforms usually call for enthusiast level supporting parts.  

 

Try to run ANY 16c or higher part made by ANY CPU manufacturer on an AIO at much higher than 4 GHz and it's going to get hot.  

[DildorTheDecent]

Just now, done12many2 said:

 

1950x CPUs are getting very hot on AIOs at stock speeds.

>AMD

>Hot

 

kek. Watch yourself. 

[done12many2]

17 minutes ago, DildorTheDecent said:

>AMD

>Hot

 

kek. Watch yourself. 

 

Speaking of hot.  I was shocked when I heard that the 1950x starts to throttle at 68c on the die.  

 

Just as big boy solutions are required for Intel Extreme, you better have big boy solutions for AMD enthusiast level too.  Gone are the days of spending $1k or more on a CPU and cooling it with leftover air parts because you blew your wad.  Wait, did idiots ever do that?  Sure they did.  

 

 

[DildorTheDecent]

3 minutes ago, done12many2 said:

 

Speaking of hot.  I was shocked when I heard that the 1950x starts to throttle at 68c on the die.  

 

Just as big boy solutions are required for Intel Extreme, you better have big boy solutions for AMD enthusiast level too.  Gone are the days of spending $1k or more on a CPU and cooling it with leftover air parts because you blew your wad.  Wait, did idiots ever do that?  Sure that did.  

But hang on? Isn't AMD supposed to be the cheap holy grail of PC gaming? Surely that means I can reuse my 60mm box fan since the Athlon XP days? Why is my TR running at the 90s? Muh soldered IHS tho. 

 

For real though. Intel had lava dunked toasters as well. It's just more fun to rip AMD for it though.