Undervolting the AMD Radeon RX Vega: better performance at lower voltage

Higher clocking speeds at lower voltage - it works!

By


Performance and power consumption

To measure the effect of our tweaks, we have taken the top model of the air-cooled RX Vega and recorded results at several settings. We measured performance in both 3DMark and a few games while also recording the data for power consumption and how consistently the card can keep up the top clocking speed. The latter we look at more closely on the next page.

Without changing any setting, the score in 3DMark Time Spy is 7482 points while the voltage is at 1200 mV. The power consumption lies at 301 watts and the temperature reached 80’C. By raising the power limiter to 150%, the consumption becomes higher and the temperature rises but the performance does not improve any.

To bypass the temperature limiter, we raised the fan speeds to 100%, but even then there was no improvement. This is where we started to play around with the current. The 3DMark Time Spy runs stable at 1070 mV but the games do not run spotlessly yet. The stable game performance comes at 1090 mV, at the same point the 3DMark Time Spy score improves by 3%. The power consumption increases somewhat.

  RX Vega 64 @ stock RX Vega 64 @ +50% PWL RX Vega 64 @ 50% PWL + UV
3DMark Time Spy 7482 7480 7704
Hitman 4K ultra 61,4 fps 59,7 fps 63,6 fps
Rise of the Tomb Raider 4K ultra 42,8 fps 42,6 fps 43,6 fps
Spanning (P7) 1200 mV 1200 mV 1090 mV
Temperature 80 ºC 87 ºC 71 ºC (100% fan)
Power consumption 300,6 W 355,5 W 305,3 W

Compare

two products discussed in this review

  Product Lowest price

AMD Radeon RX Vega 56

Vega, 3584 cores, 1156 MHz, 8192 MB, 2048 bit, DirectX 12 fl 12_1, PCI-Express 3.0 x16

Specifications Test results Reviews

AMD Radeon RX Vega 64

Vega, 4096 cores, 1247 MHz, 8192 MB, 2048 bit, DirectX 12 fl 12_1, PCI-Express 3.0 x16

Specifications Test results Reviews

Also read these 3d chip articles on Hardware.Info

The Hardware.Info website uses cookies.
*