RADEON VII - Liquid Metal Tuning
Vložit
- čas přidán 9. 02. 2019
- Follow der8auer & Marry:
/ der8auer
/ marryfoxreal
---------------------------------------------------------
Support me on Patreon:
/ der8auer
---------------------------------------------------------
Find my products at Caseking:
Delid Die Mate 2: bit.ly/2Rhv4y7
Delid Die Mate X: bit.ly/2EYLwwG
Skylake-X Direct Die Frame: bit.ly/2GW6yyC
9th Gen OC Frame: bit.ly/2UVSubi
Debug-LED: bit.ly/2QVUEt0
Pretested CPUs: bit.ly/2Aqhf6y
---------------------------------------------------------
My Equipment:
USB- Microscope: amzn.to/2Vi4dky
My Camera: amzn.to/2BN4h2O
(Affiliate Links: If you buy something, I will get a part of the profit)
---------------------------------------------------------
Music / Credits:
Fyze - Mystery
Fyze - Skypath
fyzemusique.bandcamp.com/ - Věda a technologie
Liquid metal on the 5700 XT would be great to see!
@der8auer Could you do a thermal test with your graphite thermal pad, the one that Steve show off at CES?
Also, new drivers out now!
@kaizer5lock You can't compare Der8auer to Intel!! He's too honest
@@noel8313 that's not the point lol
@@liquidxlucidity7976 ssshhhh, no one is listening anyway
In this review he said that the small benefit wasn't worth the hassle, so I'd say he's pretty unbiased/ fair.
Wow! This card is so easy to disassemble! I look forward to swapping mines as soon as I get it
Thanks for the English video.
Some guy posted on reddit that he is getting 2050MHz core and 1200MHz memory which increased his synthetic benchmark score by 21%, apparently the overclocking only has issues when in full screen mode.
I'm beginning to think that's your nail polish not marry's :P
Tech Jesus gave it to him at the last CES meet. This bromance never ends
I was going to say the same thing!
I will never hire him for manicure saloons kappa!
Yes I am happy see to more radeon vii videos
soon as i seen it being put back together, i knew it wouldn't work like you'd expect. Removing that pad and throwing liquid on there is leaving a gap/pressure issue since it's designed for a specific thickness. add a plastic washer that meters in the same thickness of the pad you cut.
Really nice video :)
“Let’s see what we can get out of this.....Baby.” Cracked me up, subscribed
🤣
Would be cool to see one of these cards hit 2400MHz on LN2. Given you could cram enough vCore into it and it doesn't bug out at full pot temps that might even be possible.
I have no desire to buy this uninnovative card but I am thankful that you made this video because I am getting to the point where I want to try to possibly mod some of my gpus or at least one of them for practice.
And I did not know about the the graphite pad. And have yet to use liquid metal.
Now that I'm thinking about it I did open up my GTX 760 and reapply some paste to that and it did help a little bit. I guess cuz it was old it benefited.
Great videos. Cheers. 😉😁
"Uninnovative"? First ever 7nm GPU, first gaming card to hit 1 TB/s stock (RTX drastically slower), first gaming card to use 4 stacks of HBM...
What was Pascal? (answer would be lame, my Titan Xp is nothing special other than its a Star Wars edition, ooooh so innovative). The feature RTX was named for only works in 2 games, and 1 is technically a demo (but it is super awesome, so you should definitely go run the Quake ray traced mod). However, just saying, Radeon RAY was a thing a ways back, and ray tracing has been a thing for an extremely long time. I don't think the tensor cores actually do anything useful that CUDA couldn't (CUDA can ray trace, its done in a lot of applications...), its worse than PhysX (which was an awesome idea, and actually useful, but it also went the way of the dodo bird and only had like 20 games it was poorly implemented in).
Vega VII is much more innovative than any recent card I can think of (besides the Fiji cards from 2015 that were the first ever to use HBM)
@@jakegarrett8109 7nm doesn't do anything on an existing architecture exponentially. 7nm does nothing in and of itself either. Just smaller. Wowee. A large bus is nothing new from AMD. They have done that before and so has NVIDIA. Adding more VRAM is nothing new either. Both have done that too. What have they really brought to the table to change the gaming industry? Nothing. At least the RTX features are something to be there to push graphics forward. And without the RTX features the Nvidia cards still are better than AMD. So yeah what is innovative my friend? Blah, blah, blah. Keep making excuses for AMD. It does nothing to help gamers. Maybe if you like certain calculations it may be good but I don't care about those as a gamer.
thanks for doing this, I was really considering doing the LM route with my VII. seems to make so little difference as to not be worth it .
I get 2ghz at 1.1v on a morpheus 2 cooler, something that helps stable oc is lifting the left side of the voltage curve and/or the middle of it to reduce voltage fluctuations.
Thank you for making English version of this video. Did you test with new driver 19.2.1?
nail polish & vaseline, you have come a long way indeed xD
That's a nice red color. Do you have the green for the nvidia GPUs, too?
guten tag :)
Another awesome video! Had my VII arrive from OCUK on Friday. I've had conformation from EKWB blocks are in design. Think I'll run my blocked Vega 64 til then lol!
"EKWB Configurator (Marian) (EKWB Support)
Feb 8, 16:00 CET
Dear Richard,
We would like to inform you that we now have plans to design a block for Radeon VII card. I will may be available by end of Q1.
regards
Marian"
I'm praying that sapphire is paying attention. And also hoping game developers will implement cmaa .
Good note on the thermal pad thickness. Does this mean for water cooling you need to use a large amount of thermal paste? I have heard from other overclockers that this gpu needs to be "gooped" with paste
Great little test run bud.
Do You think the results might be better on a MOBO that supports PCI-E 4.0 and above? I would really like to see the card running full tilt to the CPU. ^__^
@der8auer Just found out info about Carbonaut .Is it possible to share results using this thermalpad?
From my experience it looks like some of the cards may have air bubble under preapplied carbon TIM or slightly bent vapor chamber. In such situation usage of TIM is a better solution mainly because of potential hotspots.
Maybe you could try doing the washer mod, to increase mounting pressure? Just add a couple washers to increase torque on the retention bracket screws.
I'm so confused. When you go over the AMD Temperature graph after applying thermal paste the Junction Temp seems to go up to 107 and 109°C, but in your chart you say it peaked at 101°C, why is that?
Ah. Don't be confused by that. Those lines were not the actual test runs. That's just B-Roll footage
You get better thermal conductivity with graphite powder polished into both heat source and heat sink but they both have to be extremely flat. But I wonder if its possible to make an alloy with Gallium and Graphite powder and get even better thermal conductivity between source and sink.... ;)
I wonder if the resin coating used to level the die causes any negative effects towards cooling, I do know it's there to improve cooling due to some cards from Vega not having a flush die and resulting in no cooling but really I do wonder what other effects it can have on the hard. I'm sure they spent a good deal of time working on the resin but honestly resin isn't known for it's heat transfer so to some degree it surely has a impact.
I also wonder if this would prevent etching from the liquid metal on the silicon die as it shouldn't react to the coating.
I wonder if trimming the shroud (near the Radeon symbol) would increase the efficiency of the fans by letting some of the hot air have an escape.
How does the card perform with your own thermal pad (the highly flexible carbon one you showed off to Gamers Nexus)?
By chance did you try adding a spacers to the back of the tension mechanism to get a bit more pressure on the GPU die? maybe that will make a difference?
Does the nail polish melt and spill out when the temp in that area goes that high? Also is the nail polish non-conductive?
IF you get the Radeon VII direct from AMD are you more likely to get a better binned gpu core for under-volting/OC?
Any word there in Germany from Watercool making a Heatkiller block for Radeon VII?
Just ran the same test on mine I got 6933 rx Vega 64 1700/1000 intel 4970 at 4.4. Will be interesting to see how far these Vega 7 goes with proper drivers
How well does the nail polish hold up to the temperature so close to the die? What sort of lifespan does it have?
Hello der8auer. Please, if today I apply Thermal Grizzly Conductonaut Liquid Metal to my CPU, how much time must pass to give it a new application?
Thank you.
www.gamersnexus.net/guides/3359-liquid-metal-aging-one-year-test-how-often-to-replace-liquid-metal
I hope we can get some better overclocking with more mature drivers. Mine comes Tuesday and i'd like to play around with Wattman.
You still have it?
Is there some where you can buy the graphite thermal pad?
Hello from Ukraine, thnks for you video. Excellent!AMD perfect!
Is there a way to turn off the lights via software? or the only way is to disconnect that cable from the inside? Thanks for the video.
Not an Apple Fan -with the new drivers and undervolting - got ~2020 GHz...Would be interesting to see your OC skills applied once AMD release 'decent' drivers.
How would the Radeon VII do with folding? I've been unable to find any info about it.
it should be very competitive in any compute workload.
Hello i need help sorry it doesn’t concern this video . I need to replace pads on my gpu what should i take 6 k/mk pads or higher what is the default of gpu manufactures? My gpu from gigabyte and they dont say
Can I use liquid metal to repaste my 1080ti?
Wait, if you took off the graphite thermal pad and replaced it with LM and did not see very much drop in temp even with fans at 100%.
Does that possibly mean that the card cooling is already saturated or that the thermal pad is just really good at what it does?
Probably both. I think the die is too small for this power rating. The power density is over the roof and maybe the die is unable to transfer that much heat.
wait you can put nail polish onto the resistors?
Any plans to do die and heatsink lapping, or using washers to change the mounting pressure? Also undervolting.
He already mentioned that once he undervolted the card, it became unstable and would consistently crash.
But another guy did it so it's possible. Also the drivers won't be broken forever.
I lapped mine. Max junction temp hasn't really improved, but average temp during testing is lower, and max fan speeds aren't ramping up as high with the same settings.
Going to try a larger undervolt soon. Right now i'm testing the settings that tested stable before lapping.
omg a reply with some substance...I don't even know what to say! 😄
Good info! So I guess fan speed is tied to avg temp. Too bad max junction didn't improve. Did you see the thing where the guy lowered that by putting washers where it screws to the backplate? I think he got 10°C off his max junction temp. It seems at stock the pressure is more toward the edges of the die but that mod helps the centre of the die make better contact.
Good luck with the undervolting. I wonder how far the average card can go with full stability and how much it improves thermals and frequency.
@@bardacuda82 fan speed is on the junction temp. The max fan rpm isn't down very much, but it's consistently down every test. I think it's hitting that max temp very early, faster than the fan ramps up, and then the temp never gets quite that high again.
I did add some nylon washers as well. Didn't seem to help with my card. I will say that my lap job wasn't perfect. There was still some discolored areas where it wasn't perfectly flat, but it was much better than it started. Was a bitch trying to sand with so little space to work with.
What I'd like to know is if I can use a graphite thermal pad for vrm and mosfet cooling cause that'd be a great option because it'd be something you'd literally never have to replace and know you'll get solid performance out of it over whatever stock pads those components already use.
I'd be cautious about using it anywhere that has contact to the pc because graphite definitely isnt the best insulator of electricity so you might have problems .. Nit an expert on that but srill
@@alexisleftist778 I suppose that would make it trickey. :/ I guess Fujipoly ultra extreme is still the best performing solution then, albeit expensive.
Graphite pads are electrically conductive, you don't want them on components that carry electricity.
I'm getting 4K Optimized results in the high 9800s with my Strix 1080 Ti with a light daily OC (+25/+250, 110%, 90C). Aren't those Radeon VII numbers a bit low?
real-world game fps numbers >>> synthetic benchmarks. The 2080, 1080TI, and Radeon 7 are functionally identical in performance in most games, even though each has an edge on specific non-gaming workloads
People have said that on VEGA cards AMD overvolted most chips insanely high, would it be possible to undervolt this card to make it quieter and less power hungry without losing clock speed?
The price isn't an issue. the issue with the Radeon 7 is the drive optimization. AMD should have waited another month or two and nailed down the drivers before releasing the card.
Hey you have contacts within Asus! Get some programmers to crack radeon pro drivers so they work on the radeon 7!
Hey Roman, wie sieht es eigentlich aus mit einem Power Mod für Nvidea Rtx 2080 ti ?
are you not going to try your thermal pad on it?
Please try lapping the VII cooler's cold plate. Reddit user discovered it's concave and had great results.
Try removing the fans and replace it with 120mm ones, made good difference on my strix vega 64. I had huge change in temps going to liquid metal on vega 64. Anyway i read the nail polish need to have acrylic to be useful and only a few nail polishes have this. Whats your comment on that?
5°C in the junction temp is a small improvement, but would be good for AIBs to look into implementation of liquid metal usage on custom cards. Great video
What happens when you overclock it?
What about adding thermal pad around die/HBM stack + Liquid metal on to the die/HBM + washers under the screws for more mounting pressure? That heatsink/heatpipe combo looks garbage.
what nail polish did you use?
The fact you do two videos in two languages blows my mind.
In the video you didn't show the fan curve at 100%. Did you do a run with the fan on max just to see if it makes a difference?
yea, no difference
there was one guy that says that the pad isn't a pad but a phase change material. i dunno if he says it's true. his name on youtube is igor's lab. he's also german like you.
I hope this this card gets support in mojave osx. It could be a great egpu card to boost up macbook Pro.
Honestly seems to be a limitation of the cooler design
But as the fan speed was increased it didn't increase the performance. If it's limited by the cooler, then the heat pipes must be insufficient.
I'm so confused, I still don't know if I want one.
Like he said, wait and see if there are any improvements. I think the potential here is worth waiting for!
The real question is finding one
Would have done this anyway if I could find a Radeon VII!
look in the trash.
0.5% performance improvement. Very interesting. Seems that it's still thermally and power constrained. God Vega runs hot...
I wonder if this would work on Vega 56 & Vega 64?
Lap the heatsink.
Very little benefit in sanding / polishing these days.
doing that would destroy the Vapor Chamber as they use thin copper to make them because it reduces the amount of metal between the GPU Die and the cooler
I think that this card has a ton of overclocking potential and will see it when Asus and Saphire get to tinker with it.
What kind of nailpolish exactly?
cut the shroud under the radeon logo. Also do a core uniformity test using a glass block and some grease.
5:23 Total Guy You Couldnt Use the frickin Clear coat?!?! gotta go with night out on the town Red Dude lol Clear coat!!!!! my ocd just exploded
What is the "junction temperature" and why is it so high?
a little late, but the junction temp is the temp of the hotest sensor on the die. the die doesnt have a constant or equal temp. mostly because of imperfections in the die or the surface of the cooler.
Does this also add UEFI support?
Try increasing the mounting pressure, some people on reddit tried adding washers on the 4 screws on GPU to increase it. They say theres quite a difference
It may work only for graphite pad, since it's depends of a pressure.
has anyone confirmed if thats actually worth doing?
funnily enough what that does is lighten the overall tightness on the screws on the rear bracket, which increases the mounting pressure on the package. The same effect can be achieved by slightly (and consistently on all 4 screws) loosening them by a full turn
Are they going to release Nano version?
"Multiple days"
It's a pretty card
keep the english videos comming roman!
I wish AMD would get with it and push out stable, working drivers.
EDIT: I just saw that they’d pushed out the 19.2.1 drivers.
Yep, although the launch day drivers are much improved compared to this video. I can beat everything he does with the stock cooler so I'm hoping it will improve further.
the Radeon 19.2.1 drivers got pushed out and seem to be far more stable than the press release drivers. From other channels it seems that the Radeon VII consistently outperforms the RTX 2070, but trades blows with the RTX 2080 often coming in just a few percent lower. It still shows good FPS and frame times.
The comparisons I’ve seen of the R VII v. 1080ti and 2080ti leaving lagging well behind by about 30% on average. No huge surprise there, but those are cards (when and IF you can find a decent used 1080ti for sale) that have a price tag of double the MSRP for the R VII reference card.
For $700USD new from a legitimate seller it doesn’t seem a terrible deal. IF you can find one in stock.
The asshats trying to scalp these on eBay for 150% - 200%+ of retail MSRP need to die slow, painful, humiliating deaths.
@@ljhaller Yeah. I'm going to put my own video up on it just to help with the misinformation surrounding this card. I have had 8719 in 4K superposition after overclocking. I expect performance will improve in games too. Really needs to be fully revisited in a few months when stock is available and drivers are matured.
@@ljhaller I am waiting for a price drop and a waterblock from watercool.de or EK etc.
@Andrew Joy I don’t see much room for them to drop the pricing and still turn a profit. The HBM memory in an of itself is around $300 - $350. Then you live got all the other bit and the PCB which is around another $150 - $200. Add in manufacturing labor, shipping, and then the mark up they sell them to retailers at. Doesn’t leave a whole lot of meat on the bone.
Did you try washers?
did you remove the nail polish?
So if there's only a small benefit, to use LM, instead of the graphite pad, it is really F***ing good
are liquid metal pastes based on gallium? I know that gallium makes a very bad reaction with aluminum and COPPER. But it is excellent on nickel-plated objects.
Gallium and indium as the proper alloy has such a low eutectic point it stays liquid in room temperature while pure gallium doesn't. Gallium amalgamates with aluminium, but it does not with copper. Some diffusion may occur, but that doesn't deconstruct the copper cooler base material.
@@peterkiss1204 Thanks for the explanation
I wish someone sold precut foam for laptop cpu's to use as a barrier for Liquid Metal.
what about using a water block?
Why nail polish and not conformal coating ?
I'm sure it is convenient to purchase and also it is red do it is easy to see where it is applied. Just speculation tho
Again liquid tape is harder to apply very viscous, smells aweful, takes ages to dry, hard to remove. Nail polish has the same properties cheaper, easy to apply and removes with acetone easily.
Earthquake in Germany @0:57 ?
If it is so easy to protect the SMT just by nail polish - why dont they do it from the factory?
Make it $500 or $550 for the Radeon VII and I'd be interested. Or third-party custom versions... or with a full-cover waterblock. :) New tech is always finicky but it's exciting to see where this will go. 7nm Ellesmere / Polaris replacement would be awesome.
Make it $1500 for stock and I would be interested :P
5:08 - What is he using?
I was playing Apex Legend and i set my fan speed within my drivers to 60% but msi after burner was showing it set to 70% so i had to set speed to 100% before it fixed itself. dropping it down to 60% then worked.
My worry is when i see reviewers using this 3rd party app they can make a card look hotter at a set noise fan speed by showing msi after burner showing 60% fan when in fact its set to 50%. i am now unable to replicate it, sliding the slider to 100% fixed the problem.
u should lock the fan speed to a set speed when comparin
How do I buy this card? It went out of stock in 30 minutes and I'm getting the Vega 64 launch shortage PTSD.
it went out of stock because stock is almost non-existing most places, not because of popularity, the card is hot garbage. tons in stock here in germany, no-one wants them, at $499 it would have been better and with custom cooling solutions or water. but meh, 300 watts for less than 2080 perf which draws like half. i miss ati ...
I think the more appropriate question is "why would you buy this card?"
AMD site has them in
According to Igor Wallossek this is not a graphite paste, it's a polymer solution.
It's Hitachi HM03
@@der8auer Hitachi TC-HM03 www.hitachi-chem.co.jp/english/products/cc/026.html They also have other products like this.
can you use clear nail polish? :p i mean, its okay if you like the red stuff on there but...
ikr i didnt even know it was safe to use nail polish
Lohnt sich sowie so nicht mit dem LM, weil es mit der Kupferoberfläche des Kühlkörpers eine Legierung eingeht und schön hart wird. Danach kann man es nur noch abschleifen. Also entweder vorher vernickeln oder es gleich lassen. Viele wissen auch nicht das man Liquidmetall niemals mit Wasser entfernen sollte, danach wird es nämlich ätzend. Also wenn etwas auf den Tisch oder die Kleidung kommen sollte. Niemals mit Wasser abwaschen :). Erst alles abtupfen. Ansonsten habt ihr zumindestens in der Kleidung ein Loch.
I think you need to put more tension on it... By Putting small washers on it