The RX Vega 64 Owners Thread

LtMatt · 8 Sep 2017 at 12:38

I now have three RX Vega 64's and my third 64 clocks just as well as the one leading the Vega pack on the scoreboards. It reaches exactly the same clocks, but appears to run a dash cooler and only a 1c difference in CoreVsHBM temps from what i have seen.

kundi · 8 Sep 2017 at 12:44

AMDMatt said:
I now have three RX Vega 64's and my third 64 clocks just as well as the one leading the Vega pack on the scoreboards. It reaches exactly the same clocks, but appears to run a dash cooler and only a 1c difference in CoreVsHBM temps from what i have seen.

Well lucky you? None of us have been able to hit your clocks on our LC 64s and many of us have had to return our cards.

LtMatt · 8 Sep 2017 at 12:46

kundi said:
Well lucky you? None of us have been able to hit your clocks on our LC 64s and many of us have had to return our cards.

Perhaps yours weren't faulty after all? Saw your earlier post.

kundi said:
Yeah. My AIO 64 also stopped crashing on the new driver as well. Unfortunately, I already have an Air 64 on the way to replace it.

pandem0nium · 8 Sep 2017 at 12:52

pig666eon said:
yesterday was the first time i had time to sit down for the day and put in some hours gaming, what i did notice was that regardless of what setting on auto fan i put in the fan wouldnt ramp up past 1000rpm resulting in the pc to shut down after a few hours of battlefield, it happened 3 times each time i tweaked the settings before i had enough and set it to max speed where the pc didnt shut down after that. i tried all of the settings also inc power save and the bios switch, the card was roasting to the touch and the 290x i had previous in the system it never shut down like that

ive just installed the new driver so im going to give that a whirl and see if that has solved the issue but i was just wondering if anyone else was having this issue? the temp sensor is defo working and its just weird that its not ramping the fans at higher temps

I had this problem recently with an older card, a 7870. Cards fans wouldn't ramp up on auto until it overheated and shut down. Had to create a custom fan profile in afterburner.

schneeb · 8 Sep 2017 at 12:54

ubersonic said:
Not trying to be rude but, you realise you're asking for a feature to be added so that clean installs can be set not to install cleanly right?

Well its not part of the driver is it? :rolleyes:

JediFragger · 8 Sep 2017 at 12:59

AMDMatt said:
Perhaps yours weren't faulty after all? Saw your earlier post.

But that could also be turned around to say that AMD should sort out their obvious compatibility problems?

Probably a bit of both tbf.

mattyc123 · 8 Sep 2017 at 13:09

I think my card is one of the worse ones for overclock, I get 1687 at 1.165mv and +50 power limit on the core and 1075 on the memory if I go to 1692 it crashes and if I go to 1.16 it crashes too. Max temp is 43C (EK block)
I might go up on voltage but I don't want to push it too hard and am quite happy with the performance. I am looking to get a good balance between power and performance for when I get a second card at some point.

alec · 8 Sep 2017 at 13:12

ubersonic said:
To extrapolate on the above, here's the TDP limits for the different BIOS/Wattman configs

And no other differences between them? Clocks, voltages all same?

Biffa · 8 Sep 2017 at 13:19

ubersonic said:
Obvious question, but have you tried taking the fan off the radiator and putting it on the other side (so it's pulling air through the radiator instead of pushing air through it) and mounting the fan to the case instead of the rad?

I unscrewed the fan, but I'd have to cut all the cable ties and even then I'm not sure if it would stretch, mind you this was around midnight last night, so I'll have another look today. Cards reaching 70degC at the moment with the cooler in the base of the case. :eek:

it is being worked pretty hard however

LoadsaMoney · 8 Sep 2017 at 13:22

Biffa said:
Well apart from the fact that I bought a new case with a 120mm rear fan placement just for the card and the radiator doesn't fit the space. Card is in and working, albeit with the rad on the floor of the case blowing out the lower vent.

Ordered a new hdmi cable or display link as once the drivers installed I get a blank screen. Can remote in to the machine and everything is working fine so left it folding for 24 hours as a stress test

Nice one

Biffa · 8 Sep 2017 at 13:24

Its a C0

Oh and I forgot to say, the box it came in has to be the be the biggest box for a GPU I've ever seen.

kundi · 8 Sep 2017 at 13:37

AMDMatt said:
Perhaps yours weren't faulty after all? Saw your earlier post.

The first one definitely was. It wouldn't run anything but balanced or less on any driver without crashing and even then it would still crash sometimes.

The 2nd one seems like it could've been an driver issue but I haven't had enough time with the new driver to know for sure. That said, I'm not alone with this issue and AMD could've issued some sort of statement or note with the driver saying there's an issue with the driver and some of the LC 64 cards. Instead, customers like me have been beating their heads against the wall trying to figure out what's going on with our expensive GPUs. I don't want to return cards, I cannot express what a pain in the butt it is to do it, especially since I bought mine in a bundle as it was pretty much the only way to purchase one in the US. I'm what one might call an AMD fanboy, I regularly defend them on forums/reddit and I'm also a shareholder with a significant amount of stock. From a consumer standpoint though, the LC 64 issues are unacceptable, especially without comment from AMD.

heartburnron · 8 Sep 2017 at 13:45

AMDMatt said:
I now have three RX Vega 64's and my third 64 clocks just as well as the one leading the Vega pack on the scoreboards. It reaches exactly the same clocks, but appears to run a dash cooler and only a 1c difference in CoreVsHBM temps from what i have seen.

Jeeze the luck the gods you have Matt. The new drivers allow me to bench at 1777 with the voltage at 1230 but the highest I've seen it actually boost to is 1730, the highest on the previous drivers was 1724. I can lower the voltages to get a more solid boost clock at around 1700 but the performance for this compared to 1777 and extra voltage seems to be on a par...

LtMatt · 8 Sep 2017 at 13:46

heartburnron said:
Jeeze the luck the gods you have Matt. The new drivers allow me to bench at 1777 with the voltage at 1230 but the highest I've seen it actually boost to is 1730, the highest on the previous drivers was 1724. I can lower the voltages to get a more solid boost clock at around 1700 but the performance for this compared to 1777 and extra voltage seems to be on a par...

That's pretty good good to be fair.

chaosophy · 8 Sep 2017 at 14:48

shankly1985 said:
Thanks a lot for doing this. Seems your Memory is locked max. What resolution you running? and what is your GPUs max memory speed?

I running 1440p - Memory is at 945

Res was 1440p, Mem at 1040 iirc. FH3 was maxed out settings x4AA, unlocked framerate.

Was late when I did it so might re-run again if I get time, also forgot how nice the game looks at times and so want to try it out at 4K on the TV at some point.

pig666eon · 8 Sep 2017 at 14:53

so with the new driver the fan is now doing what it should be but here is a new thing now that the memory is just tanking randomly while playing games, its not thermal throttling as you can see peak temp was 76. there is no tweaking applied and its set to balanced, games are unplayable fps just tanks from 200 to 1 for a good 20 seconds every so often

TonyTurbo78 · 8 Sep 2017 at 15:28

I'm debating whether or not to bang an LC bios onto mine. The card seems like it will lap it up. But then there's the other side that says no need, 1700/1100 undervolted and 39c max temp is more than good enough.

Tinek · 8 Sep 2017 at 16:17

kundi said:
The first one definitely was. It wouldn't run anything but balanced or less on any driver without crashing and even then it would still crash sometimes

That's what mine was doing with standard settings. Balanced, crash. Power saver, less so but at times crashed too. Turbo - instant crashes.

Undervolted it, decreased clocks a tad and all was good, with really nice memory oc and undervolt. So I was happy. New driver and the same card works with 0 crashes in any of the power modes! I was sure it's faulty but it would seem drivers were at fault all along... I want to be upset with AMD but instead I'm just happy I don't have to RMA it and it works really well now.

Too bad so many people sent them with RMA when in fact they might have been ok, just drivers weren't... Waste of time for these people :/

Stamper_ · 8 Sep 2017 at 16:35

TonyTurbo78 said:
I'm debating whether or not to bang an LC bios onto mine. The card seems like it will lap it up. But then there's the other side that says no need, 1700/1100 undervolted and 39c max temp is more than good enough.

Do not think it will be worth it. The voltage to the HBM2 is the same. The only "upside" is the higher power limit and as you undervolted yours, I do not think you are currently running into any power limits? The registry hack with the power tables seems like another option to increase the power limit if you really want to. Source the almightly buildzoid: https://www.youtube.com/watch?v=SlpkmPEYTGE

Steampunk · 8 Sep 2017 at 16:35

Tinek said:
That's what mine was doing with standard settings. Balanced, crash. Power saver, less so but at times crashed too. Turbo - instant crashes.

Undervolted it, decreased clocks a tad and all was good, with really nice memory oc and undervolt. So I was happy. New driver and the same card works with 0 crashes in any of the power modes! I was sure it's faulty but it would seem drivers were at fault all along... I want to be upset with AMD but instead I'm just happy I don't have to RMA it and it works really well now. Too bad so many people sent them with RMA when in fact they might have been ok, just drivers weren't... Waste of time for these people :/

Well that's AMD's fault. We even had AMDMatt here saying that if it crashes at normal clocks with +50 percent power, it's faulty and should be RMA'd. Just goes to show how the drivers and hardware are intertwined and if one doesn't work, then the other is useless too.