Last week, we took a look at the Asus ROG Astral GeForce RTX 5080 OC Edition. It has a powerful cooler, but by default, the fan control pushes speeds high, and card is anything but quiet. So, I also measured how the card’s behavior changes when lowering fan speeds, how different power limit settings affect performance, and what the OC scanner does to the clock speed/voltage curve and how that impacts performance.
Looking at the huge Asus ROG Astral GeForce RTX 5080 cooler, it’s clear that pushing the big fans over 1500 or even 22000rpm like fan management does won’t be necessary. Let’s see how lowering their speeds impacts temperatures and clock speeds. And then we’ll look at how performance is affected if you lower the power limit of the card.
Test build
I don’t test graphics cards openly on a desk like most reviewers, I use a classic gaming PC for measurements. The components are mounted in a Fractal Design Meshify 2 case. To give you an idea, for now I’ll borrow a photo of another build in the same case, which you may have come across in articles dedicated to system cooling. It also features a quartet of 140mm fans in all available positions. All system fans have their speeds reduced to about 780 rpm. The build is not completely silent, but it can cool even high performance cards with high power consumption.
Inside is the newer Aorus X870 Elite WiFi7 motherboard, which is equipped with an AMD Ryzen 9 9800X3D processor. The cooling of the processor is taken care of by the new Noctua NH-D15 G2 in the offset position. The processor is complemented by 64GB of DDR5-6000 memory in two modules from Kingston, and the system and data are stored on a 4TB SSD from the same manufacturer. The whole setup is powered by the top-of-the-line 1300W Be quiet! Dark Power Pro 13 PSU.
Below is a picture of the configuration and current system settings and the version of Windows used for testing. The memories are set to an EXPO profile with DDR5-6000 32-38-38-38-1,350V with XMP/EXPO High Bandwidth Support set to Enabled, Infinity Fabric Frequency and Dividers manually set to 2000 MHz and the UCLK DIV1 MODE divider set to UCLK=MEMCLK.
The SVM Enable setting is deactivated in the advanced processor settings for higher performance and the “Core Isolation” setting in Windows Security is also deactivated.
Furthermore, the Radeon graphics adapter integrated in the processor and the integrated SATA controller are manually deactivated.
Benchmark for tests
For the tests, I adopted the benchmark that I use to measure operating characteristics. It’s a built-in benchmark from F1 24. It was run at 3840×2160 resolution with settings at ultra, it’s the Singapore circuit with weather set to wet, 5 laps.
Its advantage is that it is a game and not a synthetic workload and there is no pause between each run to load the test, but it runs continuously for ten minutes.
⠀
