Raijntek Morpheus II + Vega = hotspot problem

[don dolarson]

This cooler is a nightmare to get decent "hotspot" temperatures. It's the only one temperature which won't let me play games or even use my Vega 56 graphic card more intensively, because of the "hotspot" temperatures going up to the hell high 105-110C in just couple of seconds, which result in a throttle & downclock to 850-900MHz on the core (EVGA GTX 970 Superclocked performance) and very often lost connection to the monitor (black screen) and a need to turn the PC off by holding the power button. I'd never buy it... if reviewers of couple (4-6) of reviews mention about this issue, which appears to be a common thing for many (not all) Vega users who tried Morpheus II on their Vega.

My Vega is "molded", without different heights between the core and memory & Morpheus II is the standard version, black heatsink, if that matters. It's mounted properly, I've checked everything couple of times

Every temperature else looks to be very good in idle, (24C in my room) 28-32C on every sensor except HBM VRM vel GPU VR MVDD Temperature which is 42C, and while running Unigine Heaven stress test, they all are in the 55-60C range, except the evil "hotspot" which reach 105-110C in just couple of seconds after starting the benchmark. I've already refitted the Morpheus II heatsink 7 times and every time was "hotspot" the only one problem. I've read many advice on internet, especially these about making more pressure to the heatsink by screwing on all 4 screws a little bit more than when they stops, but don't you think I've already tried it? It's maximum as it's now and won't get any more. I've already fucked heads of 2 screws after all these tries and need to order new ones before I try 8th time on me. I've tried with the Morpheus II "cross" and with the "cross" coming from my reference cooler. The one coming from the reference cooler work a bit better, but far away from good. I've also tried with and without spacers but the result of all tries is the same - hotspot is reaching 105-110C in either 2 or 20 seconds. The higher "hotspot" temperatures, the longer time in 100-110C range then more throttle, ending up in 850-900MHz on core (constantly) when it stays at 104-106C and fans speed reaching their maximum. No matter my own fan curve. The hotspot temperature is a part of the fan speed and replaces my curve. Maybe because of the critical temperature.

What else could be wrong here? Maybe something I missed, haven't tried yet or something I just don't know? I'm starting to wonder if everything is OK with mounting brackets, if they're straight, and/or the whole heatsink itself is an issue and need a replacement? How to check all that before I contact Raijntek and try to get a replacement? Everything looks to be OK to my eye. I can't imagine that I put so much time and can't get this shit to work after 7 tries. It's been a 3 weeks nightmare for me. I'm completely tired of all these tries with taking the GPU out of PC, unscrew the heatsink, clean it and the GPU die out of thermal paste, apply new thermal paste, screw on, mount again, check, read on internet... everything that takes at least 2 hours every time I do it. So sad I can't run any other game than Solitaire since 3 weeks when I first tried to mount this cooler I've never had a single issue with any cooler I've had in my life, and I've had CPU coolers from Noctua, Be-Quiet and now I've just bought the Scythe Fuma II and everything works like a charm. I've also had GPU coolers back in time, namely Gelid Icy Vision and Arctic Accelero. No single problem with any of them either.


Anyone could help me to manage this? Please.

BTW. Is there a way to change the "no connection" and a black screen when the GPU fails? Such a thing never happened to me when I've had NVIDIA GTX 970 before. If something got wrong with it, only the driver stopped/restarted, but with Vega I need to turn off the PC with the power button and start it again? Not even reset works as it ends up with a restarted PC with "no connection" box flying on the monitor...

 

[eidairaman1]

We need complete system specs. Ranting and raving won't help your matters either.

 

[don dolarson]

ASUS STRIX X470 Gaming-F, Ryzen 1700X no OC with Scythe Fuma II (got it yesterday, before was Be-Quiet Pure Rock), Vega 56 reference version with molded GPU & HBM die (Samsung), Morpheus II cooler from this thread attached to the card instead of the reference cooler along with 2x Be-Quiet 120mm 2000RPM high-speed fans, 32GB of RAM Corsair Vengeance PRO LED 3000MHz, Corsair RM 750W v2 2018 Edition, Samsung 860 EVO, Intel 545s and 2x Toshiba P300. Everything sits in a Fractal Design Define R6 with 2x140mm front Be-Quiet 1600RPM intake and 1x rear exhaust fan.

Speeding both intake fans to maximum won't help. I've also tried to put the reference backplate onto the graphics card top, without screwing it to the card as it's impossible without the stock cooling plate which need to be adjusted and fitted before. Did this backplate thing only to see if it help with the "hotspot" temperature but haven't seen any differences. I shot the graphics card with a laser temperature meter and the hottest point was the GPU die area on it's top and the highest noted temperature was 63C.

 

[eidairaman1]

Might need a pad instead of paste.

Did you get this card new or used?

 

[TheoneandonlyMrK]

Might need a pad instead of paste.

Did you get this card new or used?

The extra fraction of pressure plus better lateral transfer of a graphene pad might work better, that or liquid metal did it for me.
Damn hot spots, not seen one in a while.:,p

 

[eidairaman1]

The extra fraction of pressure plus better lateral transfer of a graphene pad might work better, that or liquid metal did it for me.
Damn hot spots, not seen one in a while.:,p

The problem was exhibited on a sample /Radeon VII board that had a resin/epoxy coated gpu die with hbm, the hsf wasnt applying pressure needed to make proper contact, could be same deal here.
I will post video.

 

[delshay]

I will looking into this as I have a Vega 56 on it's way. What I learned from modding my Nano will be transferred to the Vega 56 hopefully with improvement's.

 

[kapone32]

Do you have heatpads on the VRM that are controlling the power connectors? On reference cards they should sit underneath the 8+6 or 8+8 pin headers on your card. This was my issue with a card i had. It would throttle until I put some heatsinks on it and now temps are great.

 

[Partaypooper]

I have a Morpheus mounted on my Vega 56 and ran into the same issues when installing. I must have remounted the thing more than 10 times. Actually ran out of thermal paste and had to run get some more. I eventually got it by reusing the "X" brace from the original cooler and the 4 screws from the Morpheus kit. Put the springs underneath the "X" brace with the screws through them.

Of course I could have just gotten lucky with the mounting pressure after 10 tries. After this my hot spot temps dropped back to where they were with the OEM cooler. Been mounted like this for over a year with no changes.

 

[eidairaman1]

I have a Morpheus mounted on my Vega 56 and ran into the same issues when installing. I must have remounted the thing more than 10 times. Actually ran out of thermal paste and had to run get some more. I eventually got it by reusing the "X" brace from the original cooler and the 4 screws from the Morpheus kit. Put the springs underneath the "X" brace with the screws through them.

Of course I could have just gotten lucky with the mounting pressure after 10 tries. After this my hot spot temps dropped back to where they were with the OEM cooler. Been mounted like this for over a year with no changes.

Thx for the solution, i guess the included bracket is cheap...

 

[TheoneandonlyMrK]

Thx for the solution, i guess the included bracket is cheap...

You know when I read his post it jolted the memory that I retained the hotspot issues until I refitted that tensioner bracket(on first conduconaught run)
At the time I posted a thread on it but remember a few hours after thinking about ,if Ek said to refit it and I had not, I wasn't sure , still not sure but thought I had a brain fart that I had just posted on (not fitting it back on in the first place) and scuttled off quite like.
I thought they had just shown screws fitted, not the tensioner but it definitely helps a hell of a lot, I'd work to retain it on any GPU and with any heatsink, water or air now, it's critical the pressure part now in my mind.

 

[don dolarson]

Might need a pad instead of paste.

Did you get this card new or used?

I haven't tried with a pad. What pad do you recommend, sir? I've Gelid GP-Extreme Thermal Pad 1mm 12W/mK on hand. Will it be good enough? Will the 1mm thickness be ok? No problem to buy other.
I'm quite skeptical about the liquid metal and shortcut/damage it could done to the card (and maybe the Morpheus II heatsink base too?) but if the pad won't help, liquid metal will be my last chance to get lower hotspot temperatures and have fun with this card, I guess. Right now I'm using Kryonaut Grizzly thermal paste and already used more than half of the 5,55g tube for all these 7 tries...

I bought an 1,5 year old used Vega 56 card. It worked flawless with the stock reference cooler. I was getting higher GPU/HBM/VRM temperatures overall, they were in the 70-75C range and hotspot was just 5-10C above them all. I've flashed Vega 64 to the 2nd BIOS and the reference cooler could manage it too, with a slightly higher temperatures than with the Vega 56 standard BIOS, and haven't seen any throttle. The reference turbin fan was making about the same noise as the 2x Be-Quiet 120mm 2000RPM attached to the Morpheus II now, my fan curve in MSI Afterburner is ignored, probably because the hotspot runs sooo high... they're spinning at their maximum as far as there's a a load to the graphic card and the hotspot reach +100C...

Do you have heatpads on the VRM that are controlling the power connectors? On reference cards they should sit underneath the 8+6 or 8+8 pin headers on your card. This was my issue with a card i had. It would throttle until I put some heatsinks on it and now temps are great.

I haven't had small heatsinks on them the first 5 tries, but then I put them on the 6th time I reattached the Morpheus and not seeing any differences. Here is a picture of how I've placed them all.
Red color represent heatsinks and green line thermal pad between them to achieve more cooling area. I've read about chokes(?) not generating much heat, and a solution about putting heatsinks on them anyway, and a thermal pad between them and heatsinks on VRMs next to them, and this is exactly what I've did with 6th try. I see 6-8C less in HBM VRM vel GPU VR MVDD Temperature in idle and about 10-12C in stress test thank to that, but it won't help to the hotspot temperature as it's the same as before when all small heatsinks haven't had any contact with each other. I've also cut ends of the biggest heatsink with a dremel, as you can see it on the attached picture. It's sitting on the VRM, next to the longest green line.

I have a Morpheus mounted on my Vega 56 and ran into the same issues when installing. I must have remounted the thing more than 10 times. Actually ran out of thermal paste and had to run get some more. I eventually got it by reusing the "X" brace from the original cooler and the 4 screws from the Morpheus kit. Put the springs underneath the "X" brace with the screws through them.

Of course I could have just gotten lucky with the mounting pressure after 10 tries. After this my hot spot temps dropped back to where they were with the OEM cooler. Been mounted like this for over a year with no changes.

I've tried with the Morpheus "X" brace and with the reference one, as told in the first post. Both fails, but the reference cooler one is slightly better. Hotspot temperature reach 100-110C in about 20 seconds with the reference one in 3 of 4 tries. It's mounted now but hotspot temperature this time reach +100C in just 2 seconds, while with the Morpheus "X" brace it takes always 2 seconds, and I see +100C almost immediately after Unigine Heaven has been started. The difference here is that I haven't put springs from Morpheus screws underneath the reference "X" brace (as it is with the reference screws and the reference "X" brace), but screw it through the "X" brace as they are. I tried to separate springs from screws coming with Morpheus kit, but couldn't do it. Maybe I need to use more force, but I don't wanted to crack anything and end with no screws at all since screws from reference "X" brace seems to be a bit too short. I've did exactly as this guy in 15:55.

 

[eidairaman1]

Do not use liquid metal on anything unless if you like replacing parts often

 

[Partaypooper]

I do remember having to use a concerning amount of force to get the stock "X" to screw down tightly. Also for what it's worth I used all of the small heatsinks on the VRM for my card.

 

[don dolarson]

Do not use liquid metal on anything unless if you like replacing parts often

Thank you for it. It means that liquid metal isn't for me. I just want to have a working and silent dead graphic card, without throttling, and as fast as the hotspot problem is solved, I'll never even touch it anymore. It's been a nightmare so far, and still is. Since it's done, the Morpheus gonna be there to the end of life of this card. Then I will pick it up and attach to another card, if it's gonna be possible on the new card, or sell it. What's about the thermal pad?

I do remember having to use a concerning amount of force to get the stock "X" to screw down tightly. Also for what it's worth I used all of the small heatsinks on the VRM for my card.

Would you clarify what you mean with "use a concerning amount of force"? I've reattached Morpheus 7 times and I believe that I was using different force every time I've done it. I don't really know if it's still to little (or maybe too much?), but believe me, screws are at their maximum now. Screwed them all in a "cross" pattern until each of them stopped, and then I've added about 1/4 turn more to each of them.

I need to buy new GPU screws and maybe new spacers too. Spacers are 0,5mm thick. I've tried to contact Raijntek in getting them all straight from the manufacturer, but it's impossible. They want rather that you call them to Taiwan Does anyone know if these screws will fit? I guess no one do and I will need to take measure of the length, as the M2 type seems to be the right type of screws.

 

[Bones]

Most any decent nut and bolt supply store should have what you need related to the screws. If you have to, take the card/cooler/whatever in and let them match it up. If you have at least one of the old screws they can match that up on the spot.
If you can't get the exact screw length go with whatever is the very next up in length and you can adjust/shim for that with nylon washers and so on to take up the excess length.

 

[don dolarson]

Thank you all. Since I live in a bunker, I need to take a measure of it and buy similar screws online. Next step is to buy Gelid GP-Extreme Thermal Pad 0,5mm 12W/mK and try with it, if it won't help then I'll try with 1mm pad which I've on hand. I hope that this will solve my problem, but if not, I'm thinking of cutting the cooling plate of reference cooler to fit Morpheus and try with it. I don't believe this step will help to the hotspot temps, but I'll not have any other choice left...

 

[don dolarson]

I've changed my mind little bit. I spent couple of hours on more reading about this hotspot problem and many of people who had problems with it used too density thermal paste. The problem has been gone since they reattached Morpheus (or they good from the first run) with applied either Noctua NT-H1 or Arctic Cooling MX-2/MX-4 paste. I don't know if Grizzly Kryonaut has too high density (it worked flawless on my previous CPU heatsink Be-Quiet Pure Rock and now with Scythe Fuma II), but it's wort of trying it first, since paste is quite cheaper than thermal pads in my region and I'm running out of Grizzly paste. I've just ordered all these 3 thermal pastes and will try with any of them.

 

[don dolarson]

The whole Vega graphic card lineup is an evil lineup to deal with, or something is really really wrong with the card or with the cooler itself. It's a hopeless card, so far... ZERO ENJOY.

I've refitted it 10 times now, and still can't get this shit to work. I've changed thermal paste from Grizzly Kryonaut to Noctua NT-H1 (applied a little bit more paste as per some tips) and the hotspot temp went down from 105-110C to 95-98C at its highest. Sometimes, while playing CS GO in 1440p hotspot went down to 75C for 2-4 seconds and then back to 95-98C in just 1 second. I haven't seen so much throttle though, but clocks varied from about 1210 to 1360MHz. The same was in Unigine Heaven. All other temperatures were at 60-65C. Then... I've underclocked P7 to 1100mv and P6 to 1050mv, raised power limit to +50% and... hotspot temperature went up to 110C in just 1 second and stayed at that. I've also tried with the Gelid GP-Extreme Thermal Pad 1mm 12W/mK and temperatures looked exactly the same as for the Noctua NT-H1 thermal paste, with an exception for 4-5C higher temperatures other than *hotspot in stress*, in both idle and stress.

I don't know what to do next I'm sooo pissed of because of it. This card is an expensive story and waste of time. I've got it for 200$, but could rather buy a new RTX 2060 with an aftermarket cooler instead of the Vega crap, Morpheus II crap + fans, couple of thermal pastes, thermal pad, new screws with springs and plastic mylas imported from Raijntek in Germany. I've already spent about 340$ dealing with it.

I've bought Arctic Cooling MX-2 and MX-4 pastes for testing as the price of them were under 1$ per gram and I think I put whole 8g MX-4 tube on the freaking core to sink it in and say bye to the card and go buy NVIDIA instead. It's starting to go on my nerves. It's the last AMD GPU I've ever bought...

 

[phanbuey]

sell that... get a openbox 2060 super. Never worry about this again.

 

[don dolarson]

I think it will be the best way to skip this nightmare, before something goes wrong and the card is completely dead. I'm happy that I haven't touched the stock cooler and still can get it on the card and sell it with stock cooler, and get the 2070/Super instead of 2060/Super, new or used with warranty, or even 1080/Ti would be a good choice. I really need a card with 8GB memory as I'm playing in 1440p.

I'd never get me into AMD and Vega if I knew about such problems before. The sad thing is that no one reviewer of Vega card or Morpheus II cooler told about hotspot problems, really NO ONE, and I've read about 5-6 reviews before I bought them both. I can't believe that no one of reviewers doesn't had a hotspot problem as reddit is full of people having the same problem as me.

I'm gonna try last time with MX-4 and 2g paste on the core, then if not, I go buy NVIDIA and say goodbye AMD forever. I hate nightmares.

 

[INSTG8R]

See the problem is the hotspot is on the “back” of the card under the chip so improving is near impossible. I Kryonauted my Vega 64 Nitro+ a couple of times going back and forth from ,MX-4. I have a new tube of Kryonaut for.my new 5700XT as the hotspot temps aren’t much better but there is very little improvement because of its location. I just ran an aggressive fan curve on my Vega same as I’m doing now with my 5700XT.

 

[don dolarson]

I just don't understand why these AMD GPUs working and are designed sooo bad. They could put 10 more sensors into it and go bankrupt as no one would even think of buying them. On the really loud, less effective stock cooler I haven't had this problem at all, and the thermal paste was so old and bad after almost 2 years of use that it was almost glued to the core and I've spent about 15-20 mins for properly cleaning before I first put Morpheus on it, but this Morpheus thing isn't any quiter because of the hotspot temp... when it's high, it turn up fan speed, no matter if Core or HBM temps are low. Hotspot was about 10C higher than other temperatures on the stock cooler and I've flashed 2nd BIOS to Vega 64 BIOS and it run fine with it too, no problem with temperatures at all. I feel like I got into a shit choosing an AMD GPU It's kinda funny how bad they're. It's then not strange that most of people goes NVIDIA when AMD can't do properly working graphic cards. It's my 1st AMD GPU since 6xxx series back in 2010 which I've repasted too, but this new AMD thing is just nothing to have. I've read about people having the same hotspot problem on their 3xx cards as well as on the RX 5700/XT cards. I got a lesson and homework to do, though in my humble opinion AMD drivers are better, more functional and more user friendly than NVIDIAs.

Thank you all for your time and help here. I'll try my last time and as told, I'll put 2g of MX-4 paste onto the core and try again. If not, welcome back NVIDIA. It's been enough. Were no gaming in 1,5 month...

 

[INSTG8R]

Well the hotspot temp should be a secondary concern anyway. Yes it’s alarming; I wish I couldn‘t see it. Keeping the actual GPIU/HBM cooled and will have a knock on effect on the hotspot. But 20C over the GPU temps is perfectly ”normal” My 5700XT is no different. i kept my Vega around 90C I believe 105c has been tossed around as the throttle point but dont quote me o; that. Actually I run a custom curve on my 5700XT which has exhibited some odd behaviour that it’s running the hotspot temp not the GPU. which honestly I don’t have a problem with despite reporting it as a bug nut it does make my fans occasionally “whooshy”

 

[don dolarson]

I've just disassembled Morpheus for 11th time and took a shot of thermal pad which I put instead of thermal paste last time I reattach it for 2 days ago. It sit there almost perfectly covering entire GPU chip. I run Heaven benchmark just before disassembling it and seen that changing core voltage for P7 & P6 to even less, as this time I've tried with 1020/990mv had almost 0 effect on hotspot temps, but increasing/decreasing power limit had. +10 resulted in 105C and -10 resulted in 80C. Power draw while +10 was about 170W and with -10 was 130W. Everything looks go quite linear on everything while playing with OC/UC... performance, power draw, temperatures.