Vega 64 Performance Issues

The issue I'm having is the performance difference between AMD driver 18.12.1 and any version after it.


Shadow of the Tomb Raider: 18.12.1(69fps) 19.2.1(58fps)
Resident Evil 2 Remake: 18.12.1(85fps) 19.2.1(72fps)
Shadow of War: 18.12.1(72fps) 19.2.1(62fps)


The percent differentials are so close to each other that it doesn't seem to be just a coincidence.


I'm at a loss as to what to do to try and fix this issue.



A couple of places I've looked have shown their Vega 64 getting a performance boost from these drivers.


I've tried a clean driver install with the same result multiple times. I've also updated the BIOS on my motherboard just to rule that out.


The same overclocked settings are used in both drivers. I monitored them during the benchmarks to make sure the clocks matched up and they did.


Any ideas?

Have you checked the clock speeds?

Have you tried 19.1.2? There's a chance 19.2.1 just doesn't agree with your system. This does happen to people in the form of performance loss and instability.

If performance is OK then you know to just skip 19.2.1. If you still have issues, try DDU to remove old driver files. The new version is very good compared to back in the day! Failing this, if you use a Wattman profile, try resetting it and dial in your settings again, manually. Check clock speeds while your at it and ensure you've not got anything running in the background other than the OS, Steam and your game, if possible.

I just hope you disconnected from the internet when installing the GPU driver otherwise MS drivers could be causing an issue with the VGA's performance due to conflicting drivers ext. Just DL AMD's drivers, save to desktop, then disconnect from the web. You'll be safe then, to remove what you want and try whatever driver you like. Reconnect once the driver is installed.

When I meant clean install I did in fact use DDU and booted into safe mode without the network connectivity to install the new drivers.

I also did my overclock over again in WattMan since they did add new settings. I monitored clock speed while the benchmark was running and still had the same results with the proper core and memory clock speeds showing up.


I tried 19.1.2 and got the same results as 19.2.1.


Not sure if this will point to anything, but my basic PC specs are:


Ryzen 7 1700(4Ghz)
Radeon RX Vega 64(1600/1000 actual speeds)
16gb 3000Mhz dual channel memory

What are you using to monitor frame rates?

Check the clock speeds. I had the same issue with 19.2.1 & had to roll back as the cards kept down-clocking every couple seconds in all games no matter what.

Edit: Nvm, just saw you had checked the clocks.

I'll use MSI Afterburner or Wattman. When I ran the Tomb Raider and Shadow of War benchmarks I used the FPS average at the end for comparison.


If any of you guys have a Vega 64 with either of these two games could you tell me what you get for any average for either of these two at 1440p max settings running the benchmark?

Use afterburner to check. I have my money on it's down clocking to some extent in similar fashion as mine w/ 19.2.1. Also, make sure you're unplugging your Ethernet or disabling wifi during uninstall. I usually go disable internet>clean uninstall>run DDU for left overs> install new driver.

I have a similar system and also had a bad experience with the latest drivers. The problem I had was that the fans would not rev up causing the Vega 64 to run too hot and downclock.



The 18.12.1 drivers allow the fans to rev up and cool better, resulting in better clocks and frame rates.

I made sure to follow your instructions to a T and still get the same performance degradation. I also monitored clocks speeds while running the benchmarks and actually get around 1610/1000 instead of 1600/1000 which is slightly higher.




I have my Vega 64 water-cooled so the fan speed not spinning up and causing lower clock speeds shouldn't affect me.



I did do a Timespy benchmark comparison and with 18.12.1/19.2.1 drivers I got 7879/7148, respectively. Seems like I'm losing performance across the board.


I'm just at a lose at this point.