Skylake tweaks improve discrete graphics performance

Skylake boxes

Tests on the first Intel processors running on its new Skylake architecture in early August gave some odd results regarding the performance of games using discrete graphics cards. In some benchmarks, both Haswell and Broadwell, the previous generation CPUs, performed marginally better than or tied with Skylake. When we wrote about Skylake, we mentioned that it was support for new memory and storage bandwidth, not pure CPU speed, that made it a valuable upgrade. Now we’ve got a bit more insight into why Skylake struggled to surpass its predecessors.

According to AnandTech, the FCLK (f-clock) wasn’t being set up properly during start-up. Asus helped explain the situation. The FCLK affects the transmission of data between CPU and GPU by controlling a ratio frequency setting which is tied to the BCLK (base frequency of the processor). It’s set at either 4x, 8x, or 10x for 400 MHz, 800 MHz, or 1000 MHz. Apparently the default setting of the FCLK is 800 MHz for Skylake processors; however, Intel’s recommended value for desktops was 1000 MHz, so the 10x ratio setting should be used. AnandTech understands that the 10x setting back in early August wasn’t functioning properly.

View more information:

Articles in this category: software

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button