Linux Drivers Suggest An AMD RX Vega Dual-GPU, Liquid-Cooled Solution

Some of the more recent Linux patches have already given us quite a lot of information on AMD’s upcoming Vega graphics cards. The May 10th Linux patch has added two more device IDs to the Vega family of products: 0x6864 and 0x6868. So things brings the Vega device ID’s to 9, still less than the 12 for Polaris.

radeon vega

First let’s talk about the liquid cooling. These two lines of code suggest it:

– table->Tliquid1Limit = cpu_to_le16(tdp_table->usTemperatureLimitLiquid1);
– table->Tliquid2Limit = cpu_to_le16(tdp_table->usTemperatureLimitLiquid2);

These seems to be like some sort of temperature thresholds. Now this is where it gets interesting. The The “Tliquid1Limit” and “Tliquid2Limit” could mean the same temperature threshold for two different GPUs, or two different thresholds for a single GPU. So when the GPU reaches a certain temperature it triggers a check of a pre-configured TDP table with values for cooling curves. Although we do see another line of code that makes the dual-GPU idea even more evident:

– table->FanGainPlx = hwmgr->thermal_controller. advanceFanControlParameters.usFanGainPlx;table->TplxLimit = cpu_to_le16(tdp_table->usTemperatureLimitPlx);

The “Plx” could mean PLX as in the ASIC bridge that is used to connect two different GPUs inside a single graphics card. Now these chips do heat up when in use so this piece of code could guarantee that fan speed increase solely based on the PLX chip’s temperature.

I guess time will only tell, but are you guys still interested in a dual-GPU solution?

About Author