Airsys Proposes PCE Metric to Measure Compute Capacity Within Fixed Data Center Power Limits
Редакция ЦОД:Главное
Airsys Cooling Technologies has proposed power compute effectiveness (PCE), a metric described in an Open Compute Project publication to assess the share of a data center’s provisioned power that is structurally available to compute. The proposed measure is positioned as a complement to PUE, rather than a replacement.
- PCE is proposed as a complement to PUE, not a replacement.
- The metric addresses compute capacity available within a fixed provisioned-power limit.
- It may be most relevant to retrofits and expansions constrained by grid availability.
- The 100 MW, 70 MW and 0.70 PCE figures are hypothetical examples, not facility performance data.
- The available information does not show that PCE has been adopted as an industry standard or deployed at named sites.
Airsys Cooling Technologies has proposed power compute effectiveness (PCE) as a metric for evaluating how much of a data center’s provisioned power is structurally available to support compute workloads.
Described in an Open Compute Project publication, PCE is intended to distinguish the portion of a facility’s power allocation available to servers from the power consumed by cooling, electrical losses and other supporting infrastructure. It is not presented as a replacement for power usage effectiveness (PUE), but as a complementary measure addressing a different operating question: how much compute capacity can be supported within a fixed power limit.
That distinction is relevant for retrofit and expansion programs at sites where additional grid capacity cannot be obtained quickly and local generation is not an immediate option. In such cases, improvements to cooling or electrical infrastructure may free capacity for IT equipment without increasing the site’s provisioned power allocation. An AFCOM representative cited in the source said PCE could be useful in assessing upgrades and determining what share of provisioned capacity can support IT equipment.
The source illustrates the calculation with a hypothetical facility holding 100 MW of provisioned power, of which 70 MW is structurally available for compute. In that example, PCE is 0.70. These figures are illustrative only and do not describe an operating data center or a specific project.
PUE remains a widely used indicator of facility energy overhead. According to the source, Uptime Institute data show average PUE improving from roughly 2.5 to about 1.6 over approximately 20 years, while some facilities report PUE of 1.1 or lower. PCE would instead focus on the allocation of an already constrained power envelope between compute and non-compute loads.
If operators and designers recognize the metric, PCE could become a practical framework for ranking cooling and electrical upgrades by the IT capacity they can unlock. However, the available source does not establish PCE as an industry standard or confirm deployment at specific facilities.
For operators constrained by grid capacity and unable to secure additional supply or build local generation quickly, PCE could provide a separate measure of how much IT load can be added within an existing power envelope. If adopted, it could help prioritize cooling and electrical retrofits according to the IT capacity they release, rather than operational energy efficiency alone.
- — / 21 октября 2026 г. · АнглийскийПервоисточник →
Материал подготовлен редакцией ЦОД: Главное на основе указанных первоисточников.