The Register recently published analysis from a recent reader survey on opinions and attitudes in HPC. El Reg’s readership is mostly (I would guess) from the traditional IT side of the house, so you HPC’ers might be interested in what the other 7/8s of the computer-related profession thinks of our community.
As a consequence the use of higher-performance computing platforms is becoming more important in ‘routine’ business operations. To determine whether the gap was closing between traditional HPC and more run-of-the-mill high-end computing, we ran a Reg survey and gathered information from 254 respondents, the majority IT professionals and systems architects from a mix of industries and company sizes.
The survey results aren’t earth shattering, and the sample isn’t large (254 respondents), but the graphs offer interesting, and sometimes arresting, information about the spread of opinions.
For example, Unix, Linux, and z/OS (uh, ok) all get high marks for “high-performance” requirements, but Windows is down in single digits of respondents who think it is useful for demanding requirements. Likewise, Unix- and Linux-based “clusters/grids” are used successfully by many of the respondents, but between 30 and 50% of respondents don’t use and wouldn’t consider Windows-based clusters or “specialist ‘super computers'” — those inside quotes around supercomputers are theirs, as is the (mis)spelling. If these results are representative, the Windows HPC team at least has a lot of outreach to do (note: Microsoft is a co-sponsor of the study).
More in the survey; I recommend a casual browse to let the graphs wash over you.
Note: As I was researching this I downloaded the full report, which includes the “Do you use any of the following platforms to deal with compute intensive workloads?” graph (Figure 10 in the full report) with different colors in the legend. That version of the graphic has the sense of the colors reversed, so that Windows-based clusters/grid are the most used segment of technology for compute intensive workloads. This is almost certainly an error, but at the least it is inconsistent with the graph shown in the excerpt article I linked to above.
UPDATE: Heard back from one of the authors, and the article is correct, not the full report. They are correcting the report.