AMD Vega Hardware Reviews

[Malo]

GN tested the Samsung Freesync monitor from the Vega rebates which was touted as having flickering issues (even in their own documentation). Looks like the monitor is basically unusable on Polaris in freesync mode but is fine on Vega 56. Strangely there were some issues on Vega FE as well, though not nearly as bad as Polaris, with Steve at GN stating that FE uses a Polaris-branch driver, not the same branch as Vega RX. No idea if issues with Polaris on this monitor (and others being reported) are fixable.

 

[Grall]

Why doesn't the driver framedouble automatically when there is the risk of flicker?

Ideally, monitors should have a built-in framebuffer DRAM, but since that would cost a penny they'd cut that, and instead sell a bad product which can't even function as advertised... Typical. Sometimes capitalism really sucks.

 

[Malo]

Why doesn't the driver framedouble automatically when there is the risk of flicker?

Ideally, monitors should have a built-in framebuffer DRAM, but since that would cost a penny they'd cut that, and instead sell a bad product which can't even function as advertised... Typical. Sometimes capitalism really sucks.

You're pretty much describing g-sync I believe? The extra hardware and the QA is why we have g-sync tax. Without that, results may vary (a lot). At least it's a lower cost alternative providing a lower barrier of entry to VRR gaming. The pros and cons of capitalism as you mention

 

[Grall]

You're pretty much describing g-sync I believe?

Well, at least in early G-sync monitors there was a FPGA module and some other stuff on there, not just a DRAM hooked up to the onboard panel driver IC. Plus NV being greedy dicks as usual, THAT'S where the tax comes from mostly; proprietary NV vendor lock-in fee.

Also, didn't I read something about (at least some) G-sync monitors also flickering at low framerates...? *shrug* Maybe it was early units and that's been fixed now.

 

[Kaotik]

GN tested the Samsung Freesync monitor from the Vega rebates which was touted as having flickering issues (even in their own documentation). Looks like the monitor is basically unusable on Polaris in freesync mode but is fine on Vega 56. Strangely there were some issues on Vega FE as well, though not nearly as bad as Polaris, with Steve at GN stating that FE uses a Polaris-branch driver, not the same branch as Vega RX. No idea if issues with Polaris on this monitor (and others being reported) are fixable.

Wasn't the whole flickering issue just drivers recognizing the display timings wrong, which could be fixed by custom timings no matter which gen card you have? (and that fix will more likely find it's way to drivers too, maybe it already has in Vega-branch)

 

[Malo]

Wasn't the whole flickering issue just drivers recognizing the display timings wrong, which could be fixed by custom timings no matter which gen card you have? (and that fix will more likely find it's way to drivers too, maybe it already has in Vega-branch)

Not sure. I do remember seeing that but I don't remember it being tested and substantiated. GN tested with the earliest and latest drivers for Polaris so if it's a driver update then AMD really should have fixed it there too.

Anyway I'm just reporting findings.

 

[Anarchist4000]

Why doesn't the driver framedouble automatically when there is the risk of flicker?

Depends on the issue, could be a simple QoS issue with Infinity. May also be a scheduling issue will chill-like functionality or a new way of outputting the frame. Freesync2 for example where the shader array is used for tonemapping, color correction, etc and the timing is off.

 

[CarstenS]

Wasn't the whole flickering issue just drivers recognizing the display timings wrong, which could be fixed by custom timings no matter which gen card you have? (and that fix will more likely find it's way to drivers too, maybe it already has in Vega-branch)

It would have been nice, if at least for Vega cards (in order to keep QA costs lower), they would have included this fix for the bundled and discounted display in the release driver. Yeah, I get that people with other displays would not like it so much, but at least it would have made the Pack more attractive.

 

[Gubbi]

Wasn't the whole flickering issue just drivers recognizing the display timings wrong, which could be fixed by custom timings no matter which gen card you have? (and that fix will more likely find it's way to drivers too, maybe it already has in Vega-branch)

It's probably something like that. When I have my desktop at 144Hz, my monitor flickers, when its 143.86Hz it doesn't.

Cheers

 

[Deleted member 13524]

It would have been nice, if at least for Vega cards (in order to keep QA costs lower), they would have included this fix for the bundled and discounted display in the release driver.

Isn't this exactly what gamers nexus found out? That the current RX Vega driver fixes the flicker?
It was just a matter of detecting the wrong panel timings in that specific model.

 

[Deleted member 13524]

Here it is, fixed in the latest driver:

https://support.amd.com/en-us/kb-ar...mson-ReLive-Edition-17.8.1-Release-Notes.aspx

• FreeSync brightness or flickering issues have been resolved on a small amount of Samsung FreeSync enabled displays that may have been experiencing issues.

 

[Malo]

lol nice timing

 

[Grall]

Here it is, fixed in the latest driver:

I wonder if the latest driver also fixes the driver crashes (and even bluescreens) I've had with my R9 390X when mixing 3D rendering and video playback. Happens way too much on web pages that constantly re-load their ad banners as I'm watching an inline streaming video, and one ad randomly using WebGL acceleration, or perhaps simply too much hardware accelerated 2D compositing.

Suddenly there's corruption flashing across the screen briefly in a thin band (or entire browser window fills with junk), sometimes the PC freezes, and then it either resets video driver or there's a full-blown bluescreen.

Blah... Updating video drivers is such a drag!

 

[Deleted member 13524]

I wonder if the latest driver also fixes the driver crashes (and even bluescreens) I've had with my R9 390X when mixing 3D rendering and video playback. Happens way too much on web pages that constantly re-load their ad banners as I'm watching an inline streaming video, and one ad randomly using WebGL acceleration, or perhaps simply too much hardware accelerated 2D compositing.

Suddenly there's corruption flashing across the screen briefly in a thin band (or entire browser window fills with junk), sometimes the PC freezes, and then it either resets video driver or there's a full-blown bluescreen.

Blah... Updating video drivers is such a drag!

Have you seen other people in the internet with the same problem? I've used a 290X for over two years, part of it in crossfire, and I don't remember ever crossing with that problem..

 

[Grall]

Nop, can't say I have, but it didn't start until I installed driver 17.7.2 only a short while ago. Before then I was on some older driver from this past spring, it worked fine. Of course, I had to go and delete that one - not that I enjoy rolling back to old software... It feels like defeat.

(Also, I wonder how many people are still using R9 290/390 series cards? They weren't all that common even at their peak.)

 

[eastmen]

I wonder if the latest driver also fixes the driver crashes (and even bluescreens) I've had with my R9 390X when mixing 3D rendering and video playback. Happens way too much on web pages that constantly re-load their ad banners as I'm watching an inline streaming video, and one ad randomly using WebGL acceleration, or perhaps simply too much hardware accelerated 2D compositing.

Suddenly there's corruption flashing across the screen briefly in a thin band (or entire browser window fills with junk), sometimes the PC freezes, and then it either resets video driver or there's a full-blown bluescreen.

Blah... Updating video drivers is such a drag!

I used to have my videos play black or garbled and sometimes they would crash browsers. It turned out the card was gonig bad

 

[Deleted member 13524]

Yup. exactly what I was going to say. Sounds like you might have a dying card on your hands @Grall ...

 

[lanek]

Yup. exactly what I was going to say. Sounds like you might have a dying card on your hands @Grall ...

Can allready try to disable hardware acceleration on browser, and uncheck the video acceleration too... it was even worst with flashplayer and Firefox not so long ago.

 

[seahawk]

With some of the latest RX Vega under volting tests, the card looks even stranger.

RX56 performs equal to RX64 within 2-3% but uses 25% less power...

 

[Entropy]

With some of the latest RX Vega under volting tests, the card looks even stranger.

RX56 performs equal to RX64 within 2-3% but uses 25% less power...

I believe you are referring to a test where they are undervolting the V56 and V64 by different amounts.
Hardwareluxx

Am I correct?