[UPDATED] Der8auer States Some X299 Low End boards Have Bad VRM Coolers

[Marinatall_Ironside]

6 minutes ago, cj09beira said:

now imagine what those 14/16/18 core cpus will pull   

Yep, and that's why high wattage (1+ kW) power supplies will still exist, no matter what.

 

Certain configurations will require bigger power supplies, and do keep in mind it's never good to run a power supply 75% or more of its total capacity sustained, as it will shorten its effective lifespan, despite it being able to withstand those kinds of loads. Plus it's always a good thing to have the power supply fan not turn on at low to moderate loads.

 

3 minutes ago, Taf the Ghost said:

On the v2 boards?

 

The i9-7980XE should just be called the "Circuit Flipper". We all know it's going to be tripping breakers a bunch.

Some boards like the Asus Rampage VI Extreme should handle those current loads that a CPU like the i9-7980XE will pull once you plug in the 8-pin and the 4-pin CPU power connectors at the same time. I wouldn't be surprised if there are rev.2 versions of those boards like that one where the revision has the dual 8-pin connectors to sustain those loads. I mean look at the Asus Zenith Extreme; that has dual 8-pin CPU connectors to support overclocking on the Ryzen Threadripper chips for the 14-core and the 16-core models, as the amount of current would easily overwhelm the single 8-pin and single 4-pin once overclocked.

[MageTank]

4 minutes ago, Taf the Ghost said:

Without the spec sheets for the Motherboards, it's hard to say for sure, but, for all of the issues with how Der8auer was able to cause the issue, he has found something repeatable across multiple boards. (There's no way to call a $300 USD motherboard "low-end", but they're the cheapest tier.) Which really just means that the Minimum Head Dissipation spec for the VRMs was set too low. If it was just one board from one company, then it was just a design flaw. He managed to force it in 3(?) companies, which means it's a minimum specification issue.

 

This will be fixed when the v2 end up launching later in the year.

 

But, I do love this thread.  At least it's nice to know that if you want to OC your X299 chip, make sure you're ready to spend $2000 USD.

I've seen the videos, and Der8auer has already stated his methodology. He removes the VRM current limit within the bios, manually sets VCCIN, and throws 12k FFT Prime95 at it. The multiple boards he repeated it on, were still technically the "cheapest" X299 boards, even though they were $300ish boards. In this context, low-end is a relative term.  Low-end compared to say, the $500 Aorus Gaming 7 or ASUS X299 Deluxe, which do not have these problems. Low-end in regards to quality, rather than price.

 

I personally believe the board partners were not expecting this level of heat for the VRM's, probably due to the fact that they didn't expect high-core count unsoldered chips to clock this high, or they completely underestimated the wider AVX bit-width. It's hard to say at this point, but it's certainly something they need to take into consideration for future boards. Give the low-end less RGB's and less features, but make sure the boards themselves are capable of supplying worst-case power without throttling. If people want to have their cake and eat it too, then they need to invest in more expensive boards to get that extra sex appeal or features that they want.

[AnonymousGuy]

2 minutes ago, cj09beira said:

if having to delid a 2000 dollar cpu isn't bullshit i don't know what is 

If you're going for max overclocks, you shouldn't be surprised that you have to go over the top to succeed.  Leave it stock then if you can't hang or drop back to a lower overclock.

[cj09beira]

2 minutes ago, AnonymousGuy said:

If you're going for max overclocks, you shouldn't be surprised that you have to go over the top to succeed.  Leave it stock then if you can't hang or drop back to a lower overclock.

for 2grand i expect liquid metal, hand picked silicon for he best oc, and a blow job

[done12many2]

50 minutes ago, MageTank said:

I want everyone to remember exactly why Der8auer uses Prime95. He brought it up in that second VRM video of his. He builds pre-built machines for people, and personally believes to use "the most stressful" means of validating stability. Aside from him being wrong about Prime95 being the most stressful (seriously, Linpack MKL, not even once), his point is relatively simple. If he cannot use this methodology to provide his clients a full, 100% stable overclock under the "worst" conditions, then it is a problem for him. His worry is, if he uses less stressful programs, and someone gets their system and decided to use Prime95 and it throttle or crash, that he sold them an invalid overclock. It's a perfectly fine way of thinking, and I do not blame him for having that rationale. However, everyone's personal stress test methodology is going to be different. You might find Prime95 to be primitive and brutal, and overkill for something nobody will need, but others believe it's the only way to guarantee complete stability at all times.

 

I don't agree with this.  Der8bauer has demonstrated otherwise in previous Skylake-X related videos in which he touted the high clock speeds of Skylake-X chips using Prime95 (no AVX) as his source of validation.  Even going as far as saying that he doesn't understand why people are complaining about Skylake-X (the CPU that could do 5 GHz delidded with an AIO using a non AVX load).

 

CaseKing is also known for selling 5.2 to 5.3 GHz 7700k CPUs.  Apply the reasoning that you outlined above to that and I think you'll see see where I'm going with this.  Maybe it's just okay that they sell them separately described as such as long as they don't include them in their in-house builds?

 

Why all of a sudden start talking about the strictest of requirements on CPUs?  The "worst" conditions seem to change based on the message needing to be delivered.  

 

Here's a sample of one of there 5.1 GHz chips currently available for sale.  The non-AVX version of Prime?  hmm.

 

 

 

[AnonymousGuy]

Just now, cj09beira said:

for 2grand i expect liquid metal, hand picked silicon for he best oc, and a blow job

Tangent from the thread but I've been wanting Intel to sell Diamond Platinum Jizz-You-Pant-Extreme-Edition chips that are pre-binned based on the equivalent of ASIC value.  It's 100% possible to do using the data that is already available (yay vertical integration of your fab and assembly test processes), it's just not being capitalized on (probably because the market is really small despite the massive profit margin you'd get).

 

It wouldn't surprise me if the 7980XEs are already going to at least be somewhat pre-binned to fit in the TDP.

[cj09beira]

1 minute ago, AnonymousGuy said:

Tangent from the thread but I've been wanting Intel to sell Diamond Platinum Jizz-You-Pant-Extreme-Edition chips that are pre-binned based on the equivalent of ASIC value.  It's 100% possible to do using the data that is already available (yay vertical integration of your fab and assembly test processes), it's just not being capitalized on (probably because the market is really small despite the massive profit margin you'd get).

 

It wouldn't surprise me if the 7980XEs are already going to at least be somewhat pre-binned to fit in the TDP.

the market who be too small to account for the disruption in the assembly line, 

they might be

[MageTank]

13 minutes ago, done12many2 said:

 

I don't agree with this.  Der8bauer has demonstrated otherwise in previous Skylake-X related videos in which he touted the high clock speeds of Skylake-X chips using Prime95 (no AVX) as his source of validation.  Even going as far as saying that he doesn't understand why people are complaining about Skylake-X (the CPU that could do 5 GHz delidded with an AIO using a non AVX load).

 

CaseKing is also known for selling 5.2 to 5.3 GHz 7700k CPUs.  Apply the reasoning that you outlined above to that and I think you'll see see where I'm going with this.  Maybe it's just okay that they sell them separately described as such as long as they don't include them in their in-house builds?

 

Why all of a sudden start talking about the strictest of requirements on CPUs?  The "worst" conditions seem to change based on the message needing to be delivered.  

 

 

You would have to ask him that. I am simply stating what he said his reasoning was, and that I understood the rationale behind it. I test for worst case myself, and I use worst-case as my 24/7 profile because I know under less stressful situations, I should be absolutely fine. Whether or not you believe him is one thing, but his reasoning isn't wrong in that regard.

 

I've also stated several times over, that he made mistakes in his videos. The 80C VRM temperature that he called "dangerous" is far from true, when these things can hit 125C and still be safe. His poor choice in a title was also condemned by most people (though I did still give him the benefit of the doubt) and as I've said before, his methodology is not a law that must be followed by all, nor is his opinion over whether or not these VRM's are bad, the be all, end all on the subject. He brought to light that under a very specific, very niche level of load, that certain "lower end" boards have inadequate cooling, resulting in a throttle for that specific workload. While some see this exposure as "the sky is falling" and that it's a scare tactic, people like myself see it as "more information for those that may be interested, is good". 

 

You've done a great job at countering his thesis with your board. Rather than try to attack his intent, counter it with your own results.

[done12many2]

24 minutes ago, MageTank said:

Rather than try to attack his intent, counter it with your own results.

 

Truthfully, I like his stuff and nothing's really changed with that as a result of any of this.  He's brought some great leading information to the community and I'll continue to support him in doing so.  Unless he keeps up with the flashy title shit.  

[VagabondWraith]

I still don't get his intention behind the video. I mean you throw Prime 95 on any Intel CPU and you're going to run into issues. Why X299 all of a sudden? I mean a sensationalist headline purposefully made to draw attention for a majority of people who are not running Prime 95 with Furmark with 1 case fan to go with it.  

 

His intentions are unsound.

[uplink777]

@VagabondWraith I think so too. Like, the guy has TONS of experience, I acknowledge that. And what he says almost makes complete sense. But still. Just as I wrote. It's like trying to get my 6850K to 4.7 GHz at 1.48x vcore, and using MSi X99A Raider, or Asus X99-E. It's a no go, at both cases, no matter how "good the design of the motherboard is."

[VagabondWraith]

4 minutes ago, uplink777 said:

@VagabondWraith I think so too. Like, the guy has TONS of experience, I acknowledge that. And what he says almost makes complete sense. But still. Just as I wrote. It's like trying to get my 6850K to 4.7 GHz at 1.48x vcore, and using MSi X99A Raider, or Asus X99-E. It's a no go, at both cases, no matter how "good the design of the motherboard is."

I mean, his video doesn't apply to me because I'll never use Prime 95 in my system, ever. Much like I'll never use Furmark to burn my GPU to a crisp. But if you go and create the most unrealistic scenarios for anything out there, you're gonna have problems. 

 

I wonder what will happen if I take all my case fans out and let my components fend for themselves... they're gonna burn up right? No shit!  That's how stupid his video is. If people are going to run Prime 95, then good on them. But I find it hard to believe people buying into HEDT don't understand the risks when they're not taking proper precautions and making sure that all their components are up to snuff.

 

The title of his video tells me he has vile intentions and was meant to rile people up.

[done12many2]

2 minutes ago, VagabondWraith said:

The title of his video tells me he has vile intentions and was meant to rile people up.

 

This I would definitely agree with.  Considering the extremes it takes to get to the point of seeing this on lower-end boards hardly constitutes a VRM disaster.  "VRM concerns under worst case scenarios" doesn't sound as cool and won't necessarily get the views.  

[VagabondWraith]

3 minutes ago, done12many2 said:

 

This I would definitely agree with.  Considering the extremes it takes to get to the point of seeing this on lower-end boards hardly constitutes a VRM disaster.  "VRM concerns under worst case scenarios" doesn't sound as cool and won't necessarily get the views.  

Don't get me wrong, I think Der8auer is s fantastic overclocker and respect his work, but I'm a bit salty about his video title and content therein.

-----

I'm gonna try for 5GHz this weekend. Going to get my office temps down to ~ 13-14c then go for it.  Also bought a kill-a-watt that'll be here tomorrow.  

[Dabombinable]

I'm beginning to think that a board like my P5Q Deluxe would have its VRM go up in flames with some of the CPU power figures being thrown around. Even with a very well cooled 16 phase VRM.

[DarkSmith2]

Well, i think Tom Logan got it wrong again. Its not only about Prime95. In fact Der8auer shows that Cinebenchs Load is harder then Prime95's. He was also pointing out that people dont even realise that they are throttling. You infact can run Firestrike or any other application/benchmark and get worse results without knowing whats wrong. Especially if you run a Custom Loop and CPU temps are acceptable but your VRM temps are not.

 

Tom doesnt even changed all the settings necessary to run into the same problems as he himself told us in his video. His CPU was throttling due to overheat protection at 94°C period and it managed to go up to 105°C+ while he didnt even notice that this was the reason prime95 lost load on his tests after talking 1week to Der8auer about it.

 

What we know infact is. SkylakeX is much to power hungry, hot and expansive. We also know now that the x299 Mainboards are overprized for the flaws they come with. "Cheaper kind of boards" doesnt mean anything if you pay 300€ for it, period. If you pay 300€ you can expect decent VRM cooling.

 

 

[leadeater]

1 hour ago, DarkSmith2 said:

Especially if you run a Custom Loop and CPU temps are acceptable but your VRM temps are not.

I've switched to EKWB's monoblock CPU+MB+VRM so should be fine if I was running X299? Also how many people run custom loops and don't do the MB/VRM?

 

I'm assuming EKWB will do one of these for X299 at some point.

https://www.ekwb.com/shop/ek-fb-asus-r4be-monoblock-acetal-nickel

https://www.ekwb.com/shop/ek-fb-asus-r5e-monoblock-acetal-nickel

[Dabombinable]

22 minutes ago, leadeater said:

I've switched to EKWB's monoblock CPU+MB+VRM so should be fine if I was running X299? Also how many people run custom loops and don't do the MB/VRM?

 

I'm assuming EKWB will do one of these for X299 at some point.

https://www.ekwb.com/shop/ek-fb-asus-r4be-monoblock-acetal-nickel

https://www.ekwb.com/shop/ek-fb-asus-r5e-monoblock-acetal-nickel

Part of the problem with cooling motherboards is that the blocks in the past at least were very hard to come by, even though older boards would benefit more than newer ones due to non-integrated NB.

[Lord Nicoll]

1 hour ago, leadeater said:

I've switched to EKWB's monoblock CPU+MB+VRM so should be fine if I was running X299? Also how many people run custom loops and don't do the MB/VRM?

 

I'm assuming EKWB will do one of these for X299 at some point.

https://www.ekwb.com/shop/ek-fb-asus-r4be-monoblock-acetal-nickel

https://www.ekwb.com/shop/ek-fb-asus-r5e-monoblock-acetal-nickel

I have a Maximus VIII Formula on the peasant platform, while I don't currently use the liquid part, that's really because one of my rad blocks one of the fitting holes. My VRM gets pretty toasty on 800MHz switching, not efficient but great stability. 

[Deli]

57 minutes ago, leadeater said:

I've switched to EKWB's monoblock CPU+MB+VRM so should be fine if I was running X299? Also how many people run custom loops and don't do the MB/VRM?

 

I'm assuming EKWB will do one of these for X299 at some point.

https://www.ekwb.com/shop/ek-fb-asus-r4be-monoblock-acetal-nickel

https://www.ekwb.com/shop/ek-fb-asus-r5e-monoblock-acetal-nickel

I think plenty. Me included.

[leadeater]

44 minutes ago, Dabombinable said:

Part of the problem with cooling motherboards is that the blocks in the past at least were very hard to come by, even though older boards would benefit more than newer ones due to non-integrated NB.

Yep, I buy my motherboards based on what blocks are available. Problem with that is you basically have to buy the highest model Asus/Gigabyte etc sells, I was going to anyway but it's a real problem. Would be nice if motherboard manufactures kept to a more uniform component height and mounting layout so almost every board could use the same water block.

[DarkSmith2]

1 hour ago, leadeater said:

I've switched to EKWB's monoblock CPU+MB+VRM so should be fine if I was running X299? Also how many people run custom loops and don't do the MB/VRM?

 

I'm assuming EKWB will do one of these for X299 at some point.

https://www.ekwb.com/shop/ek-fb-asus-r4be-monoblock-acetal-nickel

https://www.ekwb.com/shop/ek-fb-asus-r5e-monoblock-acetal-nickel

Well i guess the average Custom Loop user havnt had a reason to use a Monoblock if not for aesthetics, atleast not in the last couple of years. But as we've heard from Tom Logan, who asked Asus, what would be if he manage to get better cooling to the VRMs and the answer was that it'll just work fine.

 

Tom and Roman planning to test different thermal pads and stuff to bring those temps down. Because the termal pads included seem to be really bad..

 

For EK ive heard they havnt planned any Monoblocks for x299 just yet. They havnt sold them well in the past...

 

Also from my understanding of the comments below both videos the Motherboard design of these "cheap" boards doesnt allow you to clock much higher then "stable" 4,5GHz no matter what you are doing anyway... because the voltage you can pull isnt enough to go much higher... and the VRMs also arent really the best.

So if you plan to buy x299 and want to overclock really high you just cant buy any Motherboard out for that yet. Even if you would have a really nice Custom loop.

[leadeater]

2 minutes ago, DarkSmith2 said:

Well i guess the average Custom Loop user havnt had a reason to use a Monoblock if not for aesthetics, atleast not in the last couple of years. But as we've heard from Tom Logan, who asked Asus, what would be if he manage to get better cooling to the VRMs and the answer was that it'll just work fine.

 

Tom and Roman planning to test different thermal pads and stuff to bring those temps down. Because the termal pads included seem to be really bad..

 

For EK ive heard they havnt planned any Monoblocks for x299 just yet. They havnt sold them well in the past...

 

Also from my understanding of the comments below both videos the Motherboard design of these "cheap" boards doesnt allow you to clock much higher then stable 4,5GHz no matter what you are doing anyway... because the voltage you can pull isnt enough to go much higher... and the VRMs also arent really the best.

So if you plan to buy x299 and want to overclock really high you just cant buy any Motherboard out for that yet. Even if you would have a really nice Custom loop.

Yea I went with monoblock for aesthetics but also because it makes hard tubing a hell of a lot easier, less fittings and bends etc. I actually converted from separate CPU and MB/VRM blocks to the monoblock, I'm still running R4BE (X79) and don't really have any plans to upgrade soon.

[Lord Nicoll]

Just now, leadeater said:

Yea I went with monoblock for aesthetics but also because it makes hard tubing a hell of a lot easier, less fittings and bends etc. I actually converted from separate CPU and MB/VRM blocks to the monoblock, I'm still running R4BE (X79) and don't really have any plans to upgrade soon.

I think the monoblocks are ugly as hell, I much prefer the tight bends and fittings from VRM/CPU block combos, monoblocks are just masses that completely break the flow of a board.

[leadeater]

Just now, Lord Nicoll said:

I think the monoblocks are ugly as hell, I much prefer the tight bends and fittings from VRM/CPU block combos, monoblocks are just masses that completely break the flow of a board.

Does come down to what you like, the monoblock on the Black Edition does look nice and I use the acetal one not the clear cos that really is bloody ugly. I actually think it only looks nice due to the Black Edition full black design with no decals on it anywhere.

 

Unfortunately I only have a before but it's pretty easy to tell what it would look like using the monoblock.