The RX Vega 64 Owners Thread

Illuminist · 1 Feb 2018 at 18:40

AMDMatt said:
I've no idea. What i do know is I've spent enough of my life running CPU stress tests and Memtest tuning Ryzen and Threadripper/memory to optimum overclocks at low voltage.

I don't want to waste more time figuring out why this app throws up errors when running OpenCL tests on my GPUs at stock clocks.

I have HPET disabled, wonder if it could be that...

I have the same thing with my vega many many thousands of errors in one test

LtMatt · 1 Feb 2018 at 18:49

FYI folks, I've just reproduced a bug which causes a crash in Heroes Of The Storm on Vega 64 when running more than one display. Workaround for now is to run the game using DX9 mode via launcher command line or to disconnect 2nd/3rd displays.

gupsterg · 1 Feb 2018 at 19:23

@AMDMatt

Driver v18.1.1 Beta like Alpha is still "up the swan" for VRM info access via AMD ADL or I2C.

Illuminist said:
I have the same thing with my vega many many thousands of errors in one test

Can't say I've done anything differing to my OS/build then the norm out there within our communities. Perhaps we'll get:-

a) more peoples experience share.
b) insight from author.

LtMatt · 1 Feb 2018 at 19:26

gupsterg said:
@AMDMatt

Driver v18.1.1 Beta like Alpha is still "up the swan" for VRM info access via AMD ADL or I2C.

I have no updates to share publicly on that for now.

gupsterg · 1 Feb 2018 at 21:15

AMDMatt said:
I have no updates to share publicly on that for now.

NP

.

Any verification on aspect of OpenCL being limited on VRAM usage? (AIDA64 info)

Any info on expanding Max Memory Object Allocation size?

GoogalyMoogaly · 1 Feb 2018 at 21:44

AMDMatt said:
FYI folks, I've just reproduced a bug which causes a crash in Heroes Of The Storm on Vega 64 when running more than one display. Workaround for now is to run the game using DX9 mode via launcher command line or to disconnect 2nd/3rd displays.

But what are the chances that either of the people that play it have a Vega 64?

gupsterg · 1 Feb 2018 at 23:22

Wanted to wrap up why MemtestCL maybe erroring for some members.

If you use MemtestCL from say Majorgeeks, as you are searching for windows exe it is old and errors like this.

If you get the one from github, all is fine, like this.

On github, on right is button "Clone or download", click and select "Download ZIP", once you unzip download, the exe you want is in memtestCL-master\binaries.

LtMatt · 2 Feb 2018 at 08:01

GoogalyMoogaly said:
But what are the chances that either of the people that play it have a Vega 64?

Unlikely, but an FYI. I had to sit and play it, well watch it, for 30 minutes. It's a hard life and a weird game.

gupsterg said:
Wanted to wrap up why MemtestCL maybe erroring for some members.

If you use MemtestCL from say Majorgeeks, as you are searching for windows exe it is old and errors like this.

If you get the one from github, all is fine, like this.

On github, on right is button "Clone or download", click and select "Download ZIP", once you unzip download, the exe you want is in memtestCL-master\binaries.

I did use the one from Majorgeeks so that must be it, good find.

gupsterg · 2 Feb 2018 at 08:54

AMDMatt said:
I did use the one from Majorgeeks so that must be it, good find.

NP

, how you finding v18.2.1 driver? went on that last night, seems good so far, still VRM info access is borked

.

iihavetoes · 2 Feb 2018 at 19:41

AMDMatt said:
I did use the one from Majorgeeks so that must be it, good find.

I was also using the Majorgeeks version. And at this point the card is in Sapphire's hands for RMA. Fine by me, I still think there was something wrong with it.

nashathedog · 4 Feb 2018 at 21:43

iihavetoes said:
I was also using the Majorgeeks version. And at this point the card is in Sapphire's hands for RMA. Fine by me, I still think there was something wrong with it.

Let us here know the response you get from Sapphire when they get back to you please.

lordhawkwind · 4 Feb 2018 at 22:21

Help I've got Nvidiots trollling me on here from Rage3D!!!

Nathan S · 4 Feb 2018 at 23:45

lordhawkwind said:
Help I've got Nvidiots trollling me on here from Rage3D!!!

I've seen this. Sadly, over at R3D they are stumped as to what he is talking about.* Edit- We were all stumped! Even me! lol.

Even I am stumped.. He called me an "Nvidiot" for wanting to save money by buying the Vega Liquid card last year before the Crypto miners got them all soaked up and running it on my FX-8320 till the 12nm stuff came out for TR and DDR4 prices inch their way down a bit. So can you all see where I am stumped as to how I am an "Nvidiot"?

LtMatt · 5 Feb 2018 at 15:36

Trender said:
Sadly I can't it as I only have this one my pc and my friends pc are just old asf, it would be a disgrace if it were faulty because its the "Limited Edition" I bought with love, btw if you wonder yes im the same guy I posted in the amd community forum).
Also I dont know if its coincidence, but definitely when im with relive and/or amd osd, both or only 1, cod ww2 just crash faster and im telling you thats tested because with both it crashes under 5-7 mins several times I tried it and withouth it takes more time to crash (also all of this is with hbcc)

Trender said:
Hey man I got some news, not bad not good either.
I did gaming session of CoD ww2 while recording and look what it didn't crashes, but with HBM memory locked at 800 mhz , any hintsigh?

I managed to reproduce this issue on my system, good catch so I've raised a ticket with driver engineering.

I see it slightly different from how you reported it. In short, HBCC segment + ReLive + COD WW2 causes a application crash. If you disable one of either HBCC, or ReLive, the game works without crashing. Strange bug.

gupsterg · 6 Feb 2018 at 10:38

@AMDMatt

Sweet

.

After what had gone on with differing MemtestCL I felt it was necessary to do more testing

. I wanted to know if the Github version was capable of catching HBM errors. I knew from past testing that HBM 1125MHz would show an issue on my card. It may not be apparent straight away, but later I may see issues.

So I run MemtestCL 2x 3.5GB twice, no issues. I then did 3DM FS combined test looped, then 3DM FS GT1 looped, total length ~45min, no issues. I played SWBF ~1hr, again no issues. Rerun MemtestCL 2x 3.5GB 1x error, instance 1, iteration 31, Memtest86 Modulo-20: 1 errors (4625 ms). ZIP of testing.

TrixX · 9 Feb 2018 at 16:43

@AMDMatt Just wondering if AMD are looking to fix support for Stellaris. Currently Paradox are still recommending the 17.9.3 drivers which are now very very outdated. Game loads up to a point but never makes it into the game actual just main menu and freeze during game load when using DirectX. With OpenGL it works fine for around 1-2hrs before randomly crashing to desktop. Running 18.2.1 WHQL drivers. Apparently Vega has never played well with Stellaris and neither AMD nor Paradox have managed a fix. Paradox are saying it's a driver issue. Just wanted AMD's take on it

LtMatt · 9 Feb 2018 at 17:08

TrixX said:
@AMDMatt Just wondering if AMD are looking to fix support for Stellaris. Currently Paradox are still recommending the 17.9.3 drivers which are now very very outdated. Game loads up to a point but never makes it into the game actual just main menu and freeze during game load when using DirectX. With OpenGL it works fine for around 1-2hrs before randomly crashing to desktop. Running 18.2.1 WHQL drivers. Apparently Vega has never played well with Stellaris and neither AMD nor Paradox have managed a fix. Paradox are saying it's a driver issue. Just wanted AMD's take on it

I'm not aware of anything, but that doesn't mean nothing is happening.

The Stellaris twitter account tweeted that they were fixing bugs relating to Vega and provided a workaround, that was the last i heard about it.
https://community.amd.com/message/2835008

If they need support from us they should work with our developer relations team directly, not sure who their contact is.

pete910 · 9 Feb 2018 at 17:19

TrixX said:
@AMDMatt Just wondering if AMD are looking to fix support for Stellaris. Currently Paradox are still recommending the 17.9.3 drivers which are now very very outdated. Game loads up to a point but never makes it into the game actual just main menu and freeze during game load when using DirectX. With OpenGL it works fine for around 1-2hrs before randomly crashing to desktop. Running 18.2.1 WHQL drivers. Apparently Vega has never played well with Stellaris and neither AMD nor Paradox have managed a fix. Paradox are saying it's a driver issue. Just wanted AMD's take on it

Use oss driver and linux

TrixX · 9 Feb 2018 at 17:35

AMDMatt said:
I'm not aware of anything, but that doesn't mean nothing is happening.

The Stellaris twitter account tweeted that they were fixing bugs relating to Vega and provided a workaround, that was the last i heard about it.
https://community.amd.com/message/2835008

If they need support from us they should work with our developer relations team directly, not sure who their contact is.

Thanks for getting back to me, the comment from the support guy on the forums would highlight a lack of communication or a breakdown of such. I'll grab the quote:

"AndrewT (Forum Super Mod and also Helpdesk respondent):

Your problem is the AMD Vega card driver, not the game. The AMD Vega drivers have pretty much never worked for this game.

But I have one report that their 17.9.3 version works, so I'd suggest you try that..."

This is on the Paradox forums and indicates they aren't actively attempting a fix or they are not communicating well with AMD. As a Vega64 user I don't really feel I should have to downgrade to the 17.9.3 drivers for the possibility of a working game in DirectX or even worse have to just switch out for my R9 290 instead for that game! I'm truly not sure what's gone on but it doesn't seem like much since that tweet

(here's the link to the thread on the forum:
https://forum.paradoxplaza.com/forum/index.php?threads/game-wont-play.1067691/)

TrixX · 12 Feb 2018 at 07:24

@AMDMatt Latest reply from our good friends at Paradox:

FrostyYETI (Paradox Interactive)

Feb 11, 21:57 CET

Hello Phillip, This is a known issue for a good bit of games with AMD Drivers. We have no control over it.

FrostyYETI
Paradox Interactive

Why do they have no control over it? To me that's just handwavium to get rid of a problem they don't want to address. Any info from the AMD side of things?