RX Vega Owners Thread

Never messed with water blocks before. Did you try resetting the BIOS? Sometimes that can help.
 
Have any of you guys tried flashing your Vega 56 cards into a 64? How risky is it? I'm tempted to do it, it seems worth it.
 
sabrewolf732 said:
Fuck guys.


Installed my waterblock today and computer isn't booting, motherboard says vga is the issue.

Power is plugged in, tried two power supplies. Reseated card. Any ideas before I pull the block?
Click to expand...

I would check if you damaged any of the capacitors if you don't see any damage put the stock cooler on it and see :/.
 
SickBeast said:
Have any of you guys tried flashing your Vega 56 cards into a 64? How risky is it? I'm tempted to do it, it seems worth it.
Click to expand...
Pretty easy with unknown risk due to n=1, though suspect low risk as you can only flash one of the two bios. One thing that you will want to note from the buildzoid video that razor1 linked is that the liquid bios has a hard shutdown at 75C, so don't use it if you're on the stock cooler. Oh and run benchmarks before and after unlike me just to make sure it did something :)

razor1 said:
buildziod has a good video on it

Click to expand...

Note that he did release this video just prior to TPU's statement that it isn't actually unlocking shaders.
 
Last edited:
sabrewolf732 said:
Fuck guys.


Installed my waterblock today and computer isn't booting, motherboard says vga is the issue.

Power is plugged in, tried two power supplies. Reseated card. Any ideas before I pull the block?
Click to expand...


What thermal paste did you use, and was your chip epoxied between the Core and HBM2?
 
Shit. How would I have damaged the interposer? Going to pull block and assess today. ontariotl you're ekwb as well yes? Do you have epoxy?
 
And you remember specifically cleaning out tim from between the HBM and GPU?
 
Yeah, I thought about it and I was like ehhh might as well do it right and remove all the old Tim. D'oh.
 
Great...... so no consistency based off of brands seeing as XFX is now 50/50. Would have to come down to batch numbers then, no time for that.

Seems to be that Samsung HBM chips have epoxy, and SK Hynix don't.
 
Last edited:
Have you guys seen EK's fullcover block testing?

Radeon_RX_Vega_Chart.jpg
 
sabrewolf732 said:
Shit. How would I have damaged the interposer? Going to pull block and assess today. ontariotl you're ekwb as well yes? Do you have epoxy?
Click to expand...


Yes mine is epoxied and it was a Gigabyte card. I was worried I would have the non-epoxied one as it would be trickier.

You didn't happen to have any other type of Thermal paste like MX-4 did you? I haven't used the EK stuff as I don't know who the supplier is for EK and whether is it conductive.

If you didn't have the epoxied version, the recommendation is not to remove the TIM between the die's. If Saber said he did, it looks like he might have damaged the interposer.

Saber, take off the block and examine it again. Before removing the TIM, take a pic and share so we can see what is going on. What did you use to clean between the RAM and Core?

Then follow the guide from Tom's hardware on properly cleaning the die again. http://www.tomshardware.com/news/amd-vega-package-problem,35281.html

I'm going to test for conductivity with the EK paste. Ok, I just tested it and there is no conductivity so that is good news.
 
Last edited:
How'd you test? Just stick a multi-meter on continuity and push the probes in?
 
ontariotl said:
I've also read that all Vega 64 packages were epoxied. However looks like Saberwolf proved that theory wrong. So the same could be for 56 as well.
Click to expand...
The initial reports on the different packages, including the two linked above, indicated that either package was possible on the 64.
 
Has anyone used the new GPUZ 2.4 with Vega 56 and 64? I am seeing HBM temperature higher than GPU temp (hot spot) by about 3-5 degrees on both cards. Is that normal for air cooled cards? For example, my core would sit at 67, gpu hotspot would be at 75 and HBM temp would be at 79.

What is GPUZ actually reporting in terms of HBM temperature? What is the safe max for HBM temperature?
 
Using EK waterblock on Vega64 I see (42 gpu, 61 hotspot, 50 HBM) during load.
 
Eymar said:
Using EK waterblock on Vega64 I see (42 gpu, 61 hotspot, 50 HBM) during load.
Click to expand...
What brand? I have XFX for both 64 and 56 and I get the same behavior from both cards. Both cards are black air cooled and I did not mess with the HSF.
 
Mine is Gigabyte 64 Black air cooled version and currently has EK full cover waterblock.
 
_chesebert_ said:
Has anyone used the new GPUZ 2.4 with Vega 56 and 64? I am seeing HBM temperature higher than GPU temp (hot spot) by about 3-5 degrees on both cards. Is that normal for air cooled cards? For example, my core would sit at 67, gpu hotspot would be at 75 and HBM temp would be at 79.

What is GPUZ actually reporting in terms of HBM temperature? What is the safe max for HBM temperature?
Click to expand...
If it helps this is what I got.

GPU-z.gif
 
sabrewolf732 said:
No dice with oem cooler. Shit.
Click to expand...
sorry to hear. Hell you got me scared to even redo mine, the paste and pads, which I wouldn't do for another 2-3 months anyway. Something about this series is very foreign to me, not like any AMD card I have had before. Usually set the clock and it goes there, not Vega. Never mess with voltage, well did once to see if I could clock higher than Cagey on my 290 :) , did. But Vega voltage tampering is a necessity apparently to get a steady cool clock rate. It doesn't help most sensor programs don't read Vega accurately on all sensors. I really want to see VRM temps to see if that is having an impact.
 
SickBeast said:
Have any of you guys tried flashing your Vega 56 cards into a 64? How risky is it? I'm tempted to do it, it seems worth it.
Click to expand...

I am running a PowerColor 56 flashed with the Sapphire 64 Liquid Bios. Here is a shot of my die (I have Samsung HBM)
Vega die.jpg


I have a full cover EK block and was able to get the PowerColor backplate to work just fine with the EK block. Here is a shot of it installed:

Vega.jpg


I have just started testing, but so far I think I hit the silicon lottery once again. I see sustained clocks of 1760~1779 core with HBM @ 1Ghz for now, with Power Limit @ +50%. I can run -50% and still see sustained clocks of 1680~1695. I am still using the 17.8.2 driver since everyone that has already done their firestrike testing on that driver and I am playing catch up. I am using IC Diamond on my Core and HBM, since I was scared to use Liquid Metal like I did with my 290s. Temperatures seem great, mid 30s with a spike of ~40C after a 30 minute stress test.

I popped my CPU block and removed the old Liquid Pro and replaced it with Liquid Ultra. I am currently Priming away @ 5.02Ghz with some lovely temps of 59~63C :D. Once this 12 hour run completes I am going to start tweaking my Vega a bit. I have to admit I am not a big fan of WattMan..It is much more stable with Vega then it was with my 290.
 
_chesebert_ said:
Could someone post their GPUZ screenshot with the stock cooler?
Click to expand...

I wish I could, but my stock cooler was used for all of 15 minutes to make sure my card wasn't DoA and then I ripped it apart. Here is a photo of VEGA with his little brother, my last platinum clocking 290. I am aware of how dirty my case was lol. I have a positive pressure setup, 2 x 200mm front intake and 3 x 120mm AP-15s up top with just the single AP-15 pictured as the only exhaust. I sealed the case up as much as possible, which is kinda ironic since it a heavily modified HAF 932.

vega 290 together.jpg
 
ccityinstaller said:
I wish I could, but my stock cooler was used for all of 15 minutes to make sure my card wasn't DoA and then I ripped it apart. Here is a photo of VEGA with his little brother, my last platinum clocking 290. I am aware of how dirty my case was lol. I have a positive pressure setup, 2 x 200mm front intake and 3 x 120mm AP-15s up top with just the single AP-15 pictured as the only exhaust. I sealed the case up as much as possible, which is kinda ironic since it a heavily modified HAF 932.

View attachment 36232
Click to expand...
You probably should run a separate PCIE cable to your VEGA card . See FAQ https://seasonic.com/faq/

Unless you know for sure you are running less than 225 watts on that single PCIE
 
You must log in or register to reply here.
Back
Top