Personal computing discussed

Moderators: Flying Fox, morphine

 
Shobai
Gerbil
Topic Author
Posts: 94
Joined: Sat Sep 03, 2005 1:18 am

der8auer's at it again, this time with X299 VRMs

Wed Jun 28, 2017 6:23 pm

der8auer, everyone's favourite German overclocker and delidder, has posted a video about what he's discovered after playing with several X299 motherboards. Some interesting findings, for sure.
Last edited by Shobai on Thu Jun 29, 2017 7:43 pm, edited 1 time in total.
 
Vhalidictes
Gold subscriber
Gerbil Jedi
Posts: 1652
Joined: Fri Jan 07, 2005 2:32 pm
Location: Paragon City, RI

Re: der8auer's at it again

Wed Jun 28, 2017 6:36 pm

Yikes. I sure hope those boards are pre-production.
 
Krogoth
Silver subscriber
Gerbil Elder
Posts: 5352
Joined: Tue Apr 15, 2003 3:20 pm
Location: somewhere on Core Prime
Contact:

Re: der8auer's at it again

Wed Jun 28, 2017 7:55 pm

Yep, X299 and Socket 2066 was intended to be launched in Q4 2017-Q1 2018, but pressure from Ryzen and Threadripper forced Intel to launch the platform earlier then excepted and it shows. It is also the only reason that current "Kaby-Lake-X" chips even exist. Intel doesn't have enough Skylake-X silicon for this premature launch and wanted to avoid making into a complete paper launch. Intel was forced to convert regular Kaby Lake silicon into Socket 2066 package as a last-minute measure.

Kaby-Lake-X was originally meant to be the codename for next-generation HEDT chips.
Ivy Bridge i5-3570K@4.0Ghz, Gigabyte Z77X-UD3H, 2x4GiB of PC3-12800, Sapphire RX Vega 64, Corsair CX-600 and Fractal Refined R4 (W). Kentsfield Q6600@3Ghz, HD 4850 2x2GiB PC2-6400, Gigabyte EP45-DS4P, OCZ Modstream 700W, and PC-7B.
 
DrDominodog51
Gerbil Elite
Posts: 666
Joined: Sat Jan 17, 2015 12:23 pm
Location: Silicon Valley

Re: der8auer's at it again

Wed Jun 28, 2017 8:47 pm

Dammit. You beat me to posting this. :P
Celeron G3920 @ 4.7 GHz, Maximus VIII Impact, 8 GB of 4000 MHz 19-21-21-41 Samsung E-Die
 
Shobai
Gerbil
Topic Author
Posts: 94
Joined: Sat Sep 03, 2005 1:18 am

Re: der8auer's at it again

Wed Jun 28, 2017 9:28 pm

DrDominodog51 wrote:
Dammit. You beat me to posting this. :P


Sorry!
 
Shobai
Gerbil
Topic Author
Posts: 94
Joined: Sat Sep 03, 2005 1:18 am

Re: der8auer's at it again

Thu Jun 29, 2017 7:43 pm

Yesterday's shortbread has Guru3d reviewing the Gigabyte X299 Aorus Gaming 3 motherboard, with no mention of VRM temps as far as I could see [admittedly, I only had time for a quick peek].
 
DrCR
Gerbil Team Leader
Posts: 229
Joined: Tue May 10, 2005 7:18 am

Re: der8auer's at it again, this time with X299 VRMs

Thu Jun 29, 2017 8:06 pm

Wow, thanks for the link. I didn't realize he had a youtube channel -- I've got to give him props for bothering to also release English versions of his videos.
 
just brew it!
Gold subscriber
Administrator
Posts: 49673
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: der8auer's at it again, this time with X299 VRMs

Fri Jun 30, 2017 6:09 am

Those are some scary-high VRM temperatures. Most MOSFETs are rated to operate up to a junction temperature of 150C; if the back of the PCB is getting up around 100C, the MOSFET junctions are likely pushing the limit.

At least it sounds like the VRM temperature issue can be addressed with a little well-placed airflow; OCers need to be aware of this.

Hitting the power carrying capacity of an 8-pin power connector is more problematic, as there's no DIY solution to this which doesn't involve modifying the board itself.
Nostalgia isn't what it used to be.
 
Kougar
Silver subscriber
Minister of Gerbil Affairs
Posts: 2027
Joined: Tue Dec 02, 2008 2:12 am
Location: Texas

Re: der8auer's at it again, this time with X299 VRMs

Fri Jun 30, 2017 11:20 am

He's faulting the design of the block of metal aka "VRM cooler" more than anything else. He claims just removing the chunk of metal and pointing a fan at the bare VRMs solves the issue... which tells you just how useless and even detrimental these overstylized "VRM coolers" have become.

just brew it! wrote:
Hitting the power carrying capacity of an 8-pin power connector is more problematic, as there's no DIY solution to this which doesn't involve modifying the board itself.


Hence why he specifically mentioned the X299 boards that come with 8pin+4pin connectors for CPU power delivery. To be honest I didn't even realize some boards did, I will have to make special note of them. I've slagged the EPS12V connector on a 980X using a Rampage board just while gaming, so once was more than enough for me.
 
Mr Bill
Gold subscriber
Graphmaster Gerbil
Posts: 1454
Joined: Mon Jan 21, 2002 7:00 pm
Location: Colorado Western Slope
Contact:

Re: der8auer's at it again, this time with X299 VRMs

Tue Oct 10, 2017 5:26 pm

an ounce or so of 99.999% silver should do the trick!
X6 1100T BE | Gigabyte GA-990FXA-UD3 AM3+ | XFX HD 7870 | 16 GB DDR3 | Samsung 830/850 Pro SSD's | Logitech cherry MX-brown G710+ | Logitech G303 Daedalus Apex mouse | SeaSonic SS-660XP 80+ Pt | BenQ 24' 1900x1200 IPS | APC Back-UPS NS-1350 | WinXP64 Pro
 
Kougar
Silver subscriber
Minister of Gerbil Affairs
Posts: 2027
Joined: Tue Dec 02, 2008 2:12 am
Location: Texas

Re: der8auer's at it again, this time with X299 VRMs

Wed Oct 11, 2017 5:11 pm

Mr Bill wrote:
an ounce or so of 99.999% silver should do the trick!


Don't think even gold will help when he was pulling 1,000 watts from a 7980XE :P Props for him keeping all 18 cores & HT active at 6.1Ghz, nevermind on a chip with a FIVR which limits how cold it can go. Link Kind of amazing X299 VRMs can survive that, let alone the board itself. Also, sort of amusing liquid metal is apparently useless for subzero overclocking.
 
Mr Bill
Gold subscriber
Graphmaster Gerbil
Posts: 1454
Joined: Mon Jan 21, 2002 7:00 pm
Location: Colorado Western Slope
Contact:

Re: der8auer's at it again, this time with X299 VRMs

Thu Oct 12, 2017 10:00 pm

Kougar wrote:
Mr Bill wrote:
an ounce or so of 99.999% silver should do the trick!


Don't think even gold will help when he was pulling 1,000 watts from a 7980XE :P Props for him keeping all 18 cores & HT active at 6.1Ghz, nevermind on a chip with a FIVR which limits how cold it can go. Link Kind of amazing X299 VRMs can survive that, let alone the board itself. Also, sort of a musing liquid metal is apparently useless for subzero overclocking.
Yeah, liquid metals freeze just below room temperature. Even mercury can only make it down to -38C before freezing.

Edit: Watched the video. 1000W eh? I wonder which was more efficient under LN2, Threadripper or the 7980XE? :D You know there are cooling blocks that can meter the refrigerant to maintain a specific temperature. I did not watch the whole thing. Did he mention how he managed to keep temperature above -100C with a -195C liquid?
X6 1100T BE | Gigabyte GA-990FXA-UD3 AM3+ | XFX HD 7870 | 16 GB DDR3 | Samsung 830/850 Pro SSD's | Logitech cherry MX-brown G710+ | Logitech G303 Daedalus Apex mouse | SeaSonic SS-660XP 80+ Pt | BenQ 24' 1900x1200 IPS | APC Back-UPS NS-1350 | WinXP64 Pro
 
DrDominodog51
Gerbil Elite
Posts: 666
Joined: Sat Jan 17, 2015 12:23 pm
Location: Silicon Valley

Re: der8auer's at it again, this time with X299 VRMs

Thu Oct 12, 2017 10:33 pm

Mr Bill wrote:
Did he mention how he managed to keep temperature above -100C with a -195C liquid?

You only put enough LN2 into the pot to keep it around that temperature. LN2 pots tend to have fairly large copper bases that can make it a bit easier to control the temperature.
Celeron G3920 @ 4.7 GHz, Maximus VIII Impact, 8 GB of 4000 MHz 19-21-21-41 Samsung E-Die
 
Mr Bill
Gold subscriber
Graphmaster Gerbil
Posts: 1454
Joined: Mon Jan 21, 2002 7:00 pm
Location: Colorado Western Slope
Contact:

Re: der8auer's at it again, this time with X299 VRMs

Thu Oct 12, 2017 10:42 pm

DrDominodog51 wrote:
Mr Bill wrote:
Did he mention how he managed to keep temperature above -100C with a -195C liquid?

You only put enough LN2 into the pot to keep it around that temperature. LN2 pots tend to have fairly large copper bases that can make it a bit easier to control the temperature.
OK, I was wondering if they might go with mixtures. We used to use pentane and lN2 to make a bath that was stable at -131 as long as chunks of pentane were visible along with liquid. There are more combinations here on wikipedia.

I could not find any links about the Threadripper overclock that included power consumption.
X6 1100T BE | Gigabyte GA-990FXA-UD3 AM3+ | XFX HD 7870 | 16 GB DDR3 | Samsung 830/850 Pro SSD's | Logitech cherry MX-brown G710+ | Logitech G303 Daedalus Apex mouse | SeaSonic SS-660XP 80+ Pt | BenQ 24' 1900x1200 IPS | APC Back-UPS NS-1350 | WinXP64 Pro
 
Kougar
Silver subscriber
Minister of Gerbil Affairs
Posts: 2027
Joined: Tue Dec 02, 2008 2:12 am
Location: Texas

Re: der8auer's at it again, this time with X299 VRMs

Fri Oct 13, 2017 10:33 pm

Mr Bill wrote:
Yeah, liquid metals freeze just below room temperature. Even mercury can only make it down to -38C before freezing.


I would have thought solidified frozen liquid metal would've been a better medium for thermal transfer at that point, it's not any different than soldering the IHS do the chip right? I don't understand the distinction so it got my attention.

He mentioned in the vid he was forced to add a little bit of LN2 every so many seconds to keep the temp as close to the cold bug threshold without crossing it. Did you see the part where he discussed that delidding and filling the entire underside of the IHS with thermal compound gave him the best and most consistent OC results? Have to buy TIM in Costco sizes, which he apparently does going by the size of the jar on his desk. :lol:
 
Mr Bill
Gold subscriber
Graphmaster Gerbil
Posts: 1454
Joined: Mon Jan 21, 2002 7:00 pm
Location: Colorado Western Slope
Contact:

Re: der8auer's at it again, this time with X299 VRMs

Mon Oct 16, 2017 4:51 pm

Kougar wrote:
Mr Bill wrote:
Yeah, liquid metals freeze just below room temperature. Even mercury can only make it down to -38C before freezing.


I would have thought solidified frozen liquid metal would've been a better medium for thermal transfer at that point, it's not any different than soldering the IHS do the chip right? I don't understand the distinction so it got my attention.

He mentioned in the vid he was forced to add a little bit of LN2 every so many seconds to keep the temp as close to the cold bug threshold without crossing it. Did you see the part where he discussed that delidding and filling the entire underside of the IHS with thermal compound gave him the best and most consistent OC results? Have to buy TIM in Costco sizes, which he apparently does going by the size of the jar on his desk. :lol:
Misunderstood your intent. I thought you were talking about liquid metal that would be liquid as a fluid heat exchange medium. oops
However, liquid metal easier to apply and let freeze, wheras soldering 'might' cook something on the die and kill it. Although, again, there are solders that are applied with a solvent as a liquid and then become a metal weld. :D
X6 1100T BE | Gigabyte GA-990FXA-UD3 AM3+ | XFX HD 7870 | 16 GB DDR3 | Samsung 830/850 Pro SSD's | Logitech cherry MX-brown G710+ | Logitech G303 Daedalus Apex mouse | SeaSonic SS-660XP 80+ Pt | BenQ 24' 1900x1200 IPS | APC Back-UPS NS-1350 | WinXP64 Pro
 
Kougar
Silver subscriber
Minister of Gerbil Affairs
Posts: 2027
Joined: Tue Dec 02, 2008 2:12 am
Location: Texas

Re: der8auer's at it again, this time with X299 VRMs

Wed Oct 18, 2017 5:23 am

My point was der8auer stated liquid metal performs worse at subzero cooling than TIM. I'm not talking about soldering the chip, my comparison is that frozen "liquid metal" would be a solid and not much different from a typical soldered chip at those temps, so I am perplexed as to why it would perform worse than paste TIM. :D
 
Mr Bill
Gold subscriber
Graphmaster Gerbil
Posts: 1454
Joined: Mon Jan 21, 2002 7:00 pm
Location: Colorado Western Slope
Contact:

Re: der8auer's at it again, this time with X299 VRMs

Wed Oct 18, 2017 10:27 am

Kougar wrote:
My point was der8auer stated liquid metal performs worse at subzero cooling than TIM. I'm not talking about soldering the chip, my comparison is that frozen "liquid metal" would be a solid and not much different from a typical soldered chip at those temps, so I am perplexed as to why it would perform worse than paste TIM. :D
Hrumph, well then, I think the main point (after watching them put it on in the video), is that this (kryopaste) can be spread all over the entire chip package (due to being an effective electrical insulator) and thus allow the whole CPU package and socket to be cooled. That would be necessary when dissipating 700W in that small a volume.

Edit: ("liquid" solder) should have been (Kryopaste) or whatever they were calling it.
Last edited by Mr Bill on Sun Oct 22, 2017 11:11 pm, edited 1 time in total.
X6 1100T BE | Gigabyte GA-990FXA-UD3 AM3+ | XFX HD 7870 | 16 GB DDR3 | Samsung 830/850 Pro SSD's | Logitech cherry MX-brown G710+ | Logitech G303 Daedalus Apex mouse | SeaSonic SS-660XP 80+ Pt | BenQ 24' 1900x1200 IPS | APC Back-UPS NS-1350 | WinXP64 Pro
 
jihadjoe
Gerbil Elite
Posts: 613
Joined: Mon Dec 06, 2010 11:34 am

Re: der8auer's at it again, this time with X299 VRMs

Wed Oct 18, 2017 10:37 am

Kougar wrote:
My point was der8auer stated liquid metal performs worse at subzero cooling than TIM. I'm not talking about soldering the chip, my comparison is that frozen "liquid metal" would be a solid and not much different from a typical soldered chip at those temps, so I am perplexed as to why it would perform worse than paste TIM. :D

I imagine it'd depend on how the liquid metal crystallizes as it freezes. Depending on how that happens it might contract into clumps and leave huge gaps of silicon without any interface to the heatspreader.
 
Kougar
Silver subscriber
Minister of Gerbil Affairs
Posts: 2027
Joined: Tue Dec 02, 2008 2:12 am
Location: Texas

Re: der8auer's at it again, this time with X299 VRMs

Sun Oct 22, 2017 2:00 pm

Mr. Bill, yeah that'd be another point in the liquid metal's favor. Der8auer already filled the entire IHS with paste, looked akin to someone plastering a wall. :lol:

jihadjoe wrote:
I imagine it'd depend on how the liquid metal crystallizes as it freezes. Depending on how that happens it might contract into clumps and leave huge gaps of silicon without any interface to the heatspreader.


Okay, that makes a lot of sense!
 
DrDominodog51
Gerbil Elite
Posts: 666
Joined: Sat Jan 17, 2015 12:23 pm
Location: Silicon Valley

Re: der8auer's at it again, this time with X299 VRMs

Sun Oct 22, 2017 2:23 pm

At LN2 temperatures, thermal paste cracking is a serious problem. I would imagine that liquid metal would crack quite often.

Regardless, liquid metal behaves as an insulator at low temperatures. I don't know why, but I know it does.
Celeron G3920 @ 4.7 GHz, Maximus VIII Impact, 8 GB of 4000 MHz 19-21-21-41 Samsung E-Die
 
just brew it!
Gold subscriber
Administrator
Posts: 49673
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: der8auer's at it again, this time with X299 VRMs

Sun Oct 22, 2017 2:44 pm

jihadjoe wrote:
Kougar wrote:
My point was der8auer stated liquid metal performs worse at subzero cooling than TIM. I'm not talking about soldering the chip, my comparison is that frozen "liquid metal" would be a solid and not much different from a typical soldered chip at those temps, so I am perplexed as to why it would perform worse than paste TIM. :D

I imagine it'd depend on how the liquid metal crystallizes as it freezes. Depending on how that happens it might contract into clumps and leave huge gaps of silicon without any interface to the heatspreader.

Gallium (which I assume is the primary component of "liquid metal") expands when it freezes. I suppose this could cause other problems, if you get it into small crevices -- as it expands, it will tend to fracture whatever it has gotten inside of.
Nostalgia isn't what it used to be.
 
Mr Bill
Gold subscriber
Graphmaster Gerbil
Posts: 1454
Joined: Mon Jan 21, 2002 7:00 pm
Location: Colorado Western Slope
Contact:

Re: der8auer's at it again, this time with X299 VRMs

Sun Oct 22, 2017 11:13 pm

Kougar wrote:
Mr. Bill, yeah that'd be another point in the liquid metal's favor. Der8auer already filled the entire IHS with paste, looked akin to someone plastering a wall. :lol:
Just realized I meant to say the kryopaste could be slathered all over the socket due to being an insulator but liquid metal would be a conductor and short out the socket if you slathered it all over.
X6 1100T BE | Gigabyte GA-990FXA-UD3 AM3+ | XFX HD 7870 | 16 GB DDR3 | Samsung 830/850 Pro SSD's | Logitech cherry MX-brown G710+ | Logitech G303 Daedalus Apex mouse | SeaSonic SS-660XP 80+ Pt | BenQ 24' 1900x1200 IPS | APC Back-UPS NS-1350 | WinXP64 Pro

Who is online

Users browsing this forum: No registered users and 1 guest