• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

The RX Vega 64 Owners Thread

Associate
Joined
11 Oct 2017
Posts
51
Ok, that block is reasonably good, i'm wondering if you need to reapply the Tim, early vega dies had uneven height problems with the core and hbm, i'd at least loosen and retighten the mounting plate.
When I had high high hot spot temps for me it was due to the tim having failed.
With that block you should see core temp 50-60c at 1.05v and hotspot 70-80c

Those temps you posted what app or game was running at the time.
Was using jump force in 4k, Just cause i know it keeps the gpu at 99% usage, Also i have to mention this is with the gpu being undervolted and slightly overclocked, Things are worse without it, Guess i'll try what you both suggested and see if it changes.
 
Associate
Joined
1 Mar 2019
Posts
234
New cards yes they are not made any more. Custom AIB only.
Sapphire Nitro + has waterblock made by Bykski (the one used above), and Strix has also made by EK & Bykski.

I can tell you from personal experience, the Bykski is superb block and only cost me £86 including postage from China.

And they are going to be my primary waterblock company from now on.

Hi,

Where did you buy the block from? Also has it been ok since you've fitted it? Have the same card and thinking of buying this to add to my fractal s36 aio cooler .
 
Associate
Joined
25 Nov 2015
Posts
313
HBM voltage is set to 900 in wattman. I was under the impression from reading around that HBM voltage in wattman actually don't control HBM voltage :p. Perhaps that is incorrect, I don't know. Gaming has been very good with good frame intervals and no hickups unless the game itself is done poorly. I don't have specific numbers but Division 2 open beta was running exceptionally in DX12(surprised me considering how many says otherwise on the forums), Witcher 3 was running very well, same was Apex. "Running well" means low frame time intervals for solid consistency and above 75ish fps, preferable above 100 fps, depending on the title with shooters leaning towards the latter.


Yes its Samsung memory.
The hbm voltage doesn't control the hbm voltage - as you say it's set - but it sets the voltage floor for the GPU and HBM ( I don't totally understand it).
At 900mv you're probably throttling the power to the memory, which is why you don't get improved fps over 1025mhz on the hbm.
Just try it at 1000mv and see if raising the hbm frequency then increases framerates.
Getting a steady 1620mhz clock @ 1025mv sounds like a good card.
I'd be well happy with that.
Try 1080/1000 or 1100/1000 for the memory and see if that pushes your bechmarks higher.
Find one that's stable and doesn't create artifacts - job done :) Enjoy
 
Soldato
Joined
8 Jan 2003
Posts
3,692
Location
Scotland
Had my Vega 64 for about a week and loving it, great card. Got my undervolted to 1632MHz @ 1000mV and it runs solidly at that. Been playing Sea of Thieves all week with those settings and not had a single crash or issue. Hotspot temp is around 75c with the GPU at 64c.
 
Associate
Joined
1 Mar 2019
Posts
234
Just been stress testing my Sapphire Vega 64 for a bit as fitted the system in a much better case.

Do these look normal? Looks like the fans arent ramping up that high. This is undervolted and overclocked to 1630ish

30li0ld.jpg
 
Associate
Joined
11 Oct 2017
Posts
51
Ok, that block is reasonably good, i'm wondering if you need to reapply the Tim, early vega dies had uneven height problems with the core and hbm, i'd at least loosen and retighten the mounting plate.
When I had high high hot spot temps for me it was due to the tim having failed.
With that block you should see core temp 50-60c at 1.05v and hotspot 70-80c

Those temps you posted what app or game was running at the time.

Ok so i've opened the thing up and checked the paste... Owhhff there was this small yellow dot of dryish paste, However replaced the paste and temps are still ending up 70c with thermal thottling... Not sure what else to do.
 
Soldato
Joined
27 Mar 2010
Posts
3,069
Ok so i've opened the thing up and checked the paste... Owhhff there was this small yellow dot of dryish paste, However replaced the paste and temps are still ending up 70c with thermal thottling... Not sure what else to do.

What kind of imprint was on the die and block face when you took the block off? Did it seem like there was good pressure and a good amount of tim on the hbm and die?
Did you reapply with arctic silver again or you try an alternative thermal paste? Does the die have resin inbetween the die and hbm or is there a channel?

I'm thinking a few things here, to check
The block surface isn't true or that the gpu package has an uneven height, also possible that the block isn't touching the vrm mosfets properly?
There's not enough thermal compound or you need to change to a different thermal paste, or that the block pump/line/rad has a flowrate issue, or that the pump voltage isn't high enough, have you checked the temp of the coolant in the radiator during testing??
How good is your psu, what make and model and are you you connecting it via 2x separate 8pin power cables?


Here was my reference v64 when I took the heatsink off the thermal paste was dry and causing me problems.
IMG-20180518-074504.jpg
IMG-20180518-074517.jpg
 
Associate
Joined
11 Oct 2017
Posts
51
What kind of imprint was on the die and block face when you took the block off? Did it seem like there was good pressure and a good amount of tim on the hbm and die?
Did you reapply with arctic silver again or you try an alternative thermal paste? Does the die have resin inbetween the die and hbm or is there a channel?

I'm thinking a few things here, to check
The block surface isn't true or that the gpu package has an uneven height, also possible that the block isn't touching the vrm mosfets properly?
There's not enough thermal compound or you need to change to a different thermal paste, or that the block pump/line/rad has a flowrate issue, or that the pump voltage isn't high enough, have you checked the temp of the coolant in the radiator during testing??
How good is your psu, what make and model and are you you connecting it via 2x separate 8pin power cables?


Here was my reference v64 when I took the heatsink off the thermal paste was dry and causing me problems.
IMG-20180518-074504.jpg
IMG-20180518-074517.jpg

Na everything looked/felt right apart from the slight yellow spot and some dry paste, That i assume was typical paste faliure, It didn't look nearly as bad as your photos i should mention, Used the same type of paste but a fresh new one this time
The psu i'm using is a corsair rm750x PLUS gold, And yes i'm using 2 8pin plugs into the gpu.

At this rate i assuming the pump is ****** or there's something messing with the flow, Cause it wasn't always this bad.
The drivers show that at max speed i'm getting is 2351prm on that pump, When a burst of heat goes though it i hear a small click kind of like an old hardrive, And the thing has always had bubbly sounds coming out of it, But then again googling it that seems normal, Same bubble sounds from my h100i when at max speed. Oh also no i haven't since there's no way to check the coolents temps that i know of.

Think a lack of coolent could cause the issues even ?

At this rate i'll end up selling this damned thing and getting a 1080ti.
 
Associate
Joined
11 Mar 2019
Posts
2
Hi! I have the MSI Wave Vega 64 Liquid Cooled. The fan started to make noises i could not stans so i decided to swap it. I ordered a EK Pwn adapter and a new fan. While waiting i removed the old fan an use a fan connected to the motherboard since i need the adapter to plug new fan to the Gpu pcb. But without a fan connected directly to the Gpu i get no monitor signal. Is that intended? I was certain it would work and that the Gpu itself with work as long as the temps are good. But no signal, that is brutal. Anyone with experience in this?
 
Soldato
Joined
8 Jan 2003
Posts
3,692
Location
Scotland
Hi! I have the MSI Wave Vega 64 Liquid Cooled. The fan started to make noises i could not stans so i decided to swap it. I ordered a EK Pwn adapter and a new fan. While waiting i removed the old fan an use a fan connected to the motherboard since i need the adapter to plug new fan to the Gpu pcb. But without a fan connected directly to the Gpu i get no monitor signal. Is that intended? I was certain it would work and that the Gpu itself with work as long as the temps are good. But no signal, that is brutal. Anyone with experience in this?

The GPU is probably detecting there's no fan connected so is protecting itself by not powering up. Pretty sure once you connect a fan to the GPU fan header it'll work.
 
Last edited:
Soldato
Joined
10 Oct 2012
Posts
4,415
Location
Denmark
The hbm voltage doesn't control the hbm voltage - as you say it's set - but it sets the voltage floor for the GPU and HBM ( I don't totally understand it).
At 900mv you're probably throttling the power to the memory, which is why you don't get improved fps over 1025mhz on the hbm.
Just try it at 1000mv and see if raising the hbm frequency then increases framerates.
Getting a steady 1620mhz clock @ 1025mv sounds like a good card.
I'd be well happy with that.
Try 1080/1000 or 1100/1000 for the memory and see if that pushes your bechmarks higher.
Find one that's stable and doesn't create artifacts - job done :) Enjoy

I'll give your suggestion a try later on.. Just installed the latest 19.3.1 drivers and while performance is great so is the darn power draw!!! even undervolted im hitting 500 watt at the wall vs 400-450 before. Same wattman settings and all
 
Back
Top Bottom