Terrible experience with XFX RX 580

[fullinfusion]

Last month I ordered a brand new XFX Radeon RX 580 GTS XXX Edition 8GB (RX-580P8DFD6) from Amazon during Black Friday to replace aging GTX 970.

I've been using the card for a week and my experience with it been terrible. Unbelieve amount of issues.

* Poor rendering performance of 640x480 (BIOS, POST, DOS, Linux).
* The card is buzzing when the computer is POWERED OFF if plugged to a monitor via DP (outlet is grounded and everything is plugged to a UPS).
* Black screen on Windows' startup when using dual monitors via DVI and HDMI.
* AMD WattMan cannot overclock the memory (Hynix) more than 100 MHz (hard lockup). MSI Afterburner works fine.
* AMD Overlay and ReLive sometimes crash games.
* PC automatically powers on after a shutdown (WTF).
* Random "CPU overclocking failed" errors at POST.
* Fans are loud and sometimes run at 100 percent during boot.
* Artifacts appear in low-perf 3D (PUBG in-game inventory screen) on stock clocks.

Today I plugged the GTX 970 back, everything works perfectly again.

I already requested a replacement from Amazon... hope its just a defective card and not a flawed design. @XFXSupport

You know what buddie?

Without me going through the entire thread I'll just say.... Just RMA the card! I hax XFX a while back and yes it worked but it was slow.
In the end I got a fast ass card,
RMA and be done with it

 

[Aquinus]

You know what buddie?

Without me going through the entire thread I'll just say.... Just RMA the card! I hax XFX a while back and yes it worked but it was slow.
In the end I got a fast ass card,
RMA and be done with it

You must have missed the part where the OP already tried that. This is card #2.

 

[Ferrum Master]

Voltage reflection appears on all ports. DP makes it worse.

I've had maybe 30 VGAs in my life. No card reflected voltage like this.

2 case fans keep spinning when the system is powered down and PSU AC switch is off.

There is no such thing as voltage reflection here lol. You disable current source. It logically leaks according to Kirchoff law everywhere in any voltage rail through grounding termination resistors.

XFX did nothing wrong. In engineering there are often cases like that. The thing needed is overenginering to foresee bad engineering in stupid end devices not following specs. That may be the case, but you won't prove it to XFX and all your efforts are futile, they did nothing wrong. Also the thing should be more popular. Like any user should have this issue, with the same PCB design. I tend to believe there is something else more.

HDMI and DVI also has powered +5V PIN and feeds back.

The older devices you mentioned didn't even implement such features, like powering up dongles and converters, capture cards etc, so it is quite normal you endure it only now. The monitors also didn't have powered MHL type HDMI's.

Find cables without power pins. Also look, double check your grounding scheme at home.

 

[fullinfusion]

You must have missed the part where the OP already tried that. This is card #2.

Did you read where I said without going through this entire thread?

 

[Regeneration]

You must have missed the part where the OP already tried that. This is card #2.

This is a replacement from Amazon. Not RMA.

I'd definitely open a dialog with XFX.

This is exactly what I'm planning to do.

XFX did nothing wrong. In engineering there are often cases like that. The thing needed is overenginering to foresee bad engineering in stupid end devices not following specs. That may be the case, but you won't prove it to XFX and all your efforts are futile, they did nothing wrong. Also the thing should be more popular. Like any user should have this issue, with the same PCB design. I tend to believe there is something else more.

GTX 970 works perfectly fine. I have high-end UPS and PSU.

There is no reason for the card to feed DP/DVI/HDMI voltage to the +12v rail. The lack of isolation can fry the whole system.

There are 15 percent negative 1-star reviews on Amazon for this specific item.

After about 2 months of having this card, it had fried itself. I was playing fallout when my computer crashed and upon trying to start it up, my USB keyboard and mouse would get power for their LED's but the computer would not start, no fans would spin. A few days later I tried it again to which sparks shot out of my video card. It would have likely fried my whole computer if I hadn't unplugged my power strip from the wall. I later tried another video card that a friend gave me to which I downloaded the drivers for and it worked pretty much fine. Very disappointed in this product as it nearly ruined my entire computer.

The card fried itself within an hour of being in my case. Very disappointed. To be fair, Amazon ships the product in retail packaging. There is very little protection within the box itself and I imagine it had a bumpy ride during the shipping process.

The card made a few sparking noises and my machine crashed. The computer would not boot until I removed the card and replaced it with my old one. There was a noticeably burn smell in my office and the card had a very strong smell when it was removed and packaged for a return.

Thankfully Amazon quickly accepted the return and it will be going back tomorrow.

 

[TheoneandonlyMrK]

This is a replacement from Amazon. Not RMA.



This is exactly what I'm planning to do.



GTX 970 works perfectly fine. I have high-end UPS and PSU.

There is no reason for the card to feed DP/DVI/HDMI voltage to the +12v rail. The lack of isolation can fry the whole system.

There are 15 percent negative 1-star reviews on Amazon for this specific item.

You say you tried different cables , how different , howd ya know it is different, older spec cables made cheap can have all pins connected.

A different cable is not automatically better, get a decent late spec display port cable and hdmi , their Are subtle differences.

 

[Ferrum Master]

There is no reason for the card to feed DP/DVI/HDMI voltage to the +12v rail. The lack of isolation can fry the whole system.

Draw a scheme. With 5 current sources. +3.3V,+5V, +12V and the monitor(3.3v or 5V depending or DP or HDMI) , they share common 0 rail. With resistors, some other active elements pulled to ground in each branch. Cut off all sources except the monitor(the thing you do when shutdown). What kind of potential will be across those resistors, internal diodes? You did it at school, didn't you? Here is no different. Things will bleed through and you can elevate the voltage in those rails while being turned off. Some fans can start to work from 3-4V already...

Well I very very simplified it, but it is not what you think it is.

The 970 may have older specification, or the power pins are left disconnected at all as seldom who uses those adapters and etc things. It could be that they are using an additional LDO for those voltages. But I must check the schematics then for both.

The only other cuckoo scenario are seen when there is no ground. And two devices generate and induct a loop. Thus causing anomalies like that. No matter how good your PSU or UPS is. The ground will pollute Y cap leaked current summing up from each SMPS based device at the same branch. And causes things spin for sure.

Maybe try the duct tape method. Isolate the Pin20 for the DP cable. Just for science.

 

[Regeneration]

OK, I solved the DP issue temporary.

Now here's another problem:

DVI plugged to ASUS VG248QE 1920x1080@144hz (defined as primary)
HDMI plugged to LG 4K TV 1920x1080@60hz

Black screen on Windows' startup. OS unresponsive.

Possible to reconfigure in VGA mode. Looks like a driver issue.

 

[EarthDog]

Does DVI (any DVI) doesn't support 1080p 144Hz....? I thought it was 1080p 120Hz max?

I don't think its a driver issue if the above is true.

 

[Regeneration]

Dual-link DVI works with 144hz, and the configuration works fine, but restart results in a black screen on both monitors.

 

[EarthDog]

nd the configuration works fine, but restart results in a black screen on both monitors.

So, to be clear... it works, but a restart (soft boot?) results black screens?

 

[Regeneration]

So, to be clear... it works, but a restart (soft boot?) results black screens?

Yes.

 

[EarthDog]

Ok, that wasn't mentioned initially.

 

[Ferrum Master]

You did boot with only your primary screen on? From cold boot?

Then in windows pugged the TV? Then reboot?

Or the problem is, that the 4K LG does work in FHD?

 

[Regeneration]

Ok, I found a way to avoid the black screen of death.

Apparently it happens if the TV is set to full RGB in AMD Settings.

Hardware bugs, software bugs. Hell of a purchase.

 

[EarthDog]

So... not drivers?

 

[Regeneration]

It's a driver bug. The LG TV has no problem displaying full RGB. It works with PS4, GTX 970, and even RX 580, unless I reboot.

I forced it to YCbCr 4:4:4 in the AMD control panel. No black screen on Windows' startup anymore.

 

[Ferrum Master]

Do you have any HDCP settings? In the TV?

 

[Regeneration]

Do you have any HDCP settings? In the TV?

Nope. HDCP was reported working on Nvidia.

Anyway, I moved to another issue: GPU being overvolted by default on stock clocks.

 

[Ferrum Master]

Nope. HDCP was reported working on Nvidia.

Anyway, I moved to another issue: GPU being overvolted by default on stock clocks.

Ain't it by design?

 

[EarthDog]

Nope. HDCP was reported working on Nvidia.

Anyway, I moved to another issue: GPU being overvolted by default on stock clocks.

LOL, jesus man...

How do you know it is being over volted?

 

[Regeneration]

GPU-Z, MSI Afterburner, voltage and temperature readings.

 

[EarthDog]

Those display voltage, yes.

But what is its stock voltage that its going over? All cards are not the same in their voltage curves.

 

[Ferrum Master]

Aint it because of the dual display mode?

 

[Regeneration]

Aint it because of the dual display mode?

Multi-monitor has nothing to do with GPU core voltage.

For some reason, the vGPU jumps between 1.17v to 1.23v under load for stock clocks. WattMan reports that stock voltage should be 1.15v.