Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
Anyone with a Radeon VII should really be using The Pro Enterprise Software drivers. Im on 19Q4 version and it plays all the games without any problems or "thermal runaway"
The strange thing about Navi drivers is that not everyone has issues.
Anecdotal luck isn't proof of your assertion .Well, than the answer is not enough airflow in the case. Had all these cards with no issues
That doesn’t matter unless the PSU has multiple rails for the PCI-E connectors. Those cables can more than handle the wattage of two 8 pins going through them.
Last I checked... ATX spec still says 150 watts per rail. Granted the spec is old... and I have no doubt most PSUs these days go above.
Still 5700 xt is a 220 watt card... I would not use one 8 bin cable with a splinter.
Perhaps most modern power supplies will be fine... but if your running something a couple years old. Or your just barely meeting the 750 watt min most MFGs state as a requirement I would for sure use 2 rails unless I had zero choice.
I hope AMD figures it out. Once you lose a customer for something like this and the other side works... you’ll have a really really hard time winning them back. The average joe.
I didn’t have any blackscreens on my Radeon VII but that’s Vega not Navi...
I personally buy single rail PSUs. The idea of multiple rails feeding one device never felt right to me. I also believe it caused all sorts of issues with people using 2080tis and multi rail, corsair PSUs.
I agree about the splitter... it should be fine but it’d bug me. One well made connector in reality can handle 400W-ish. If you have a loose connection though... the risks aren’t worth it.
my RX580 works fine, but I am running the linux drivers
Except AMD acknowleged there’s issues...Yes, my anecdotal luck is more credible simply because I own these cards vs you repeating shit you read
Yep. I skipped that one for GTX 680s. Cheaper and a little faster.I remember way back when I had AMD's last fastest single GPU (7970's) and all the massive driver issues getting Crossfire and Eyefinity working. The amount of hours I spent was insane... AMD has always had a terrible GPU driver team.
Your PSU knowledge is a decade out of date, at minimum. Also, those recommendations are massively overblown. My computer doesn’t even consume 750 watts and it’s a hell of a lot more power hungry than what most people are using.Last I checked... ATX spec still says 150 watts per rail. Granted the spec is old... and I have no doubt most PSUs these days go above.
Still 5700 xt is a 220 watt card... I would not use one 8 bin cable with a splinter.
Perhaps most modern power supplies will be fine... but if your running something a couple years old. Or your just barely meeting the 750 watt min most MFGs state as a requirement I would for sure use 2 rails unless I had zero choice.
Anecdotal evidence is worthless. AMD acknowledged that there are problems. Stop acting like a fanboy, riding a company’s dick does no one any good, especially the customer. Being “loyal” or a fanboy and blatantly ignoring obvious problems only serves to give companies a free pass to screw people over.Yes, my anecdotal luck is more credible simply because I own these cards vs you repeating shit you read
Not to mention your long sad history of bashing AMD around this forum...
Something I've noticed is that AMD started to heavily monitor GPU temps since the introduction of Vega. For example the Vega 56/64 have HBM and Hotspot temperature sensors which show drastically different values from the core temp. I know this carried onto RX 5700's which now the Hotspot is called junction temp, and it can get stupidly hot. My Vega56 before a thermal paste swap would have a hotspot temp of 104C running Furmark. RX 5700's can achieve 110C in junction temp, which AMD considers within spec. There is no way that's normal to have a temperature of over 100C. To give you an idea the R9 Fury cards didn't measure HBM or any sort of hotspot or junction temperature.That is why I tend to think it might be more about cooling then real software issues at least for a good chunk of the reported issues. No doubt some are real software issues with specific games... every GPU vendor runs into that.
AMD went with a blower on their first cards cause they where clearly worried about heat issues.... I have a feeling most of the "software" issues with black screens, or crashes are related to heat. I know the 5700 I have if I where to unplug my 3 case exhaust fans... I have no doubt the inside of my case would turn into a nice warm fall day in hell. After playing Odyssey for a few hours... my top exhaust feels like a space heater. (and I know its cool to shit on AMDs software right now... but their game tailored fan curves work great. Nice option when only a couple games really push the card... nice to ramp them down for games where I have the frame rate locked anyway instead of driving 300 FPS for nothing)
Your PSU knowledge is a decade out of date, at minimum. Also, those recommendations are massively overblown. My computer doesn’t even consume 750 watts and it’s a hell of a lot more power hungry than what most people are using.
Anecdotal evidence is worthless. AMD acknowledged that there are problems. Stop acting like a fanboy, riding a company’s dick does no one any good, especially the customer. Being “loyal” or a fanboy and blatantly ignoring obvious problems only serves to give companies a free pass to screw people over.
This, if you are having this much trouble writing stable drivers for your HW it is an indiciation of the quirkiness of the HW.Is the hardware actually good or really just broken that they can't figure it out with just software?
Something I've noticed is that AMD started to heavily monitor GPU temps since the introduction of Vega. For example the Vega 56/64 have HBM and Hotspot temperature sensors which show drastically different values from the core temp. I know this carried onto RX 5700's which now the Hotspot is called junction temp, and it can get stupidly hot. My Vega56 before a thermal paste swap would have a hotspot temp of 104C running Furmark. RX 5700's can achieve 110C in junction temp, which AMD considers within spec. There is no way that's normal to have a temperature of over 100C. To give you an idea the R9 Fury cards didn't measure HBM or any sort of hotspot or junction temperature.
So temperature could be a serious problem, unless the black screens happen when you run a web browser as well, which was my problem. For a while I was certain this was an issue with FireFox and used Chrome for over a month without any issues, but the problem came back eventually and nearly everyday the driver would crash and I'd get a black screen. I'm not sure how hot Navi cards get when using a web browser but for me my Vega56 doesn't go above 33C when using Firefox. BTW the hotspot temp on my Vega56 hits 76C when playing a game, which is fanatically better than the 104C it got before. Thermal Grizzly Kryonaunt is really good thermal paste, which is what I'd recommend Navi users to try if they do reach insane junction temps. I don't think it'll solve the issue but I wouldn't want to use a card that reaches 110C.
Haha, by this logic you ride Nvidia's dick. I use both companies , so fanboy or loyal costumer is not my label
The spec hasn't changed.... perhaps most MFGs surpass the standard. But I wouldn't count on it... every major PSU MFG I'm aware of recommend GPUs with 2 power connectors, get their own plug. Sure it may work not doing that but... why the hell would you do that, if you buy a 200+ watt GPU get a proper PSU and plug it in as recommended.
MSI 8-pin x 1+ 6-pin x 1 750w recommended.
Asus 2 x 8-pin 600w recommended
Power Colour Two 8-pin PCI Express Power connectors 700w recommended.
Gigabyte 8 Pin*1, 6Pin*1 600w recommended
I agree 750w probably isn't required to run a 5700 xt... I know from AMDs driver that my card pulls around 220 watts... and I'm sure even counting for a bit of inefficacy 350 or so watts for everything else is fine. Having said that... I have no doubt there are people trying to run these things on 450 watt power supplies. Cause AMD has hit a sweet spot on price where a lot of people that are probably lower end GPU buyers might splurge a bit to get a 5700 or a XT. Especially when all the review sites come out and say its a killer value. Knowing its going to pull 250 watts at least on most peoples bronze and less then rated PSUs... I would say 700w is really the min I would consider for a 5700 class navi.
As for being out of date.... not really. The spec hasn't been updated since 2013... the last revision was 7 years ago. I really don't think many MFGs are going much further above and beyond the current spec... and the voluntary industry stuff like the bronze/gold/platinum efficiency badges don't exactly mean all that much.
https://web.archive.org/web/2017071...ocuments/guides/power-supply-design-guide.pdf
Is the hardware actually good or really just broken that they can't figure it out with just software?
I'm still using Radeon drivers from September 2018 on one system because Devinci Resolve doesn't recognize my card for OpenCL processing with anything newer.
well going on their BIOS fiasco with the 5600, their cards are just as back as the drivers.
The badges mean the PSU's meet efficiency standards, nothing more and nothing less.
Most PSUs these days put all of their PCI-E on a single rail, a lot only have a single rail period. It's been that way for a long time.
None of your links tell people to use two separate cables. I have NEVER seen a video card that tells users to use two separate cables.
The cards are fine. The BIOS fiasco was a huge mess, but that's really only a launch/launch window concern. It's not going to be a long term issue. Plus, that has nothing to do with hardware quality.
4. If you have dual monitor setup in extended configuration - the issues occur so often (almost every day with adrenalin 2020 drivers) especially with Freesync 2 and HDR enabled.
Last I checked... ATX spec still says 150 watts per rail. Granted the spec is old... and I have no doubt most PSUs these days go above.
Still 5700 xt is a 220 watt card... I would not use one 8 bin cable with a splinter.
Perhaps most modern power supplies will be fine... but if your running something a couple years old. Or your just barely meeting the 750 watt min most MFGs state as a requirement I would for sure use 2 rails unless I had zero choice.
I'm still using Radeon drivers from September 2018 on one system because Devinci Resolve doesn't recognize my card for OpenCL processing with anything newer.