Recently, Hardware Unboxed tested the effect of using resizable bar on their Z790 Intel test bench. They discovered that Gigabyte boards show substantial performance improvements in certain games over other brands, when paired with an RTX 4090. Primarily this was in Horizon Zero Dawn. Ultimately, they determined that the Gigabyte boards had resizable bar turned off whereas all other boards had it turned on.
I decided to do a quick test with my more modest setup to see if I could replicate this. I used Z690 boards paired with the Intel I9-12900K, to compare a RTX 3080 to Radeon 6900XT. This is an informal test, and I needed the RTX 3080 machine immediately operational. This meant I couldn’t put the RTX 3080 on the test bench, and it’s technically two different albeit similar machines. This means do not use this as a gold standard of benchmarks and treat is a just a fun exercise. Both had and Intel 12900K, MSI Unify Z690 motherboards, 32GB G.Skill DDR5-6400 CAS 32 RAM, Windows 11, and stock CPU settings.
First up was the Radeon 6900XT.
With the AMD 6900XT I saw an increase in performance in Horizon Zero Dawn enabling resizable bar. The average and 1% low frame times were 4.7% and 7.6% to the better with resizable bar active. There was a very large improvement to 0.1% low frame rates of 16%. That’s a difference you may be able to perceive.
So resizable bar works well for AMD GPUs on the Z690 platform Horizon Zero Dawn. Let’s look at the Nvidia RTX 3080.
Clearly this is the total opposite effect. Average frame rates and 1% lows both went down by 4.3%. The 0.1% lows took a far more worse hit at 9.2% loss in performance by turning resizable bar on.
That’s a pretty solid loss of performance by turning on resizable bar for the Nvidia GPU, and backs up some of Hardware Unboxed results.
Check out the Hardware Unboxed video below.
