Advanced Computing in the Age of AI | Friday, March 29, 2024

HPC Budgets: Where is the Money Going? 

Perhaps not surprisingly hardware still accounts for the lion’s share (48%) of HPC budgets according to Intersect360 Research’s just released 2015 report – HPC Budget Allocation Map: HPC Budget Distribution. Server spending has been the top item the last five years, and edged up about two percent over 2014. Storage was second, again a repeat.

More interesting is that while spending generally was up among site survey respondents cloud computing spending for HPC was not. Cloud computing spend has averaged about 3% over the last five surveys with minor variations in any given year rather than a trend upwards or downwards.

“We do continue to have a few respondents report that they will be increasing their use of cloud computing (without indication of private versus public cloud or which type of cloud services) in the next two years,” according to the report, “However, to see a significant increase in the overall spending for cloud computing, the usage would have to be more widespread, and we have not seen that over the last five years.

“Only about a quarter of the respondents report spending any money on cloud computing. Of these, roughly half reported spending predominantly on public cloud offerings. The share of respondents who are predominantly using public cloud has remained about the same over the last several surveys. Therefore, we continue to believe that most of the spending increase will be directed towards private cloud rather than public cloud for their HPC compute and storage needs.”

Intersect 360 noted that 18 percent of the survey respondents were from the commercial sector.

The five cloud computing spending categories measured were defined as follows:

  • Raw cycles: Purchase of computing cycles for which the organization supplies the applications software.
  • Cloud storage: Purchase of data storage space located at a cloud provider’s facility.
  • Software as a service/applications support: Purchase of computing cycles and ISV software
  • within the cloud.
  • Environment hosting: Use of a service to support virtually all computational tasks, with servers, storage, and software all being maintained by a third party.
  • Other utility/outsourcing costs: Anything not covered above.

I360 Cloud Budgets 2015 PM

Staffing is the second-largest overall expense at HPC sites, accounting for 20% of the budget. System management and operations is the primary component of this category. Software is the third-largest overall expense at HPC sites at 12% of the budget, with software tools as the major budget item. The top four HPC categories (server, storage, systems management and operations personnel, and networks) account for half (51%) of the total HPC budgets in 2015. Servers, storage, and networks account for 64% of all HPC spending on products and services (excludes facilities and staffing).

“In this year’s and last year’s Budget Map study, we were provided a glimpse into the additional cost associated with using accelerators. A few respondents listed their expenditures for these special processing units in the other hardware category. For these respondents, the additional spending on accelerators ranged from 5% to as high as 78% of the entire hardware spend. This implies users are purchasing accelerators to gain additional performance rather than to replace CPUs. If this holds true, the use of accelerators, FPGAs, and coprocessors will shift an increasing share of the budget towards server spend,” according to the report.

I360 HPC Budget 2015Intersect360 Research’s 2015 survey received 175 responses from all geographies, sectors, and budget levels. North America accounts for 48% of the respondents’ site locations in 2015. Thirty-three countries are represented in 2015, and 45 countries in all years. Academic sites represent 69% of the 2015 respondents. Please note that the number of respondents answering each question varies.

For the full report, visit Intersect360: http://www.intersect360.com/industry/reports.php?id=37

EnterpriseAI