Need help for benchmark (office purpose)

[Kardekton]

Hi community!

As a part of a thesis I have to conduct a comparison between 2 different groups of computers. The computers are focussed to be used in an office environment and what I wanted to do is to compare them based on this.

I've been reading that programs such as SysMark 2014 SE or PCMark 8 are good alternatives (even if there is debate about it), as they both have an option which simulates the load. I was thinking in using PC Mark 8 (as I have the license) in a conventional mode (without activating OpenCL) so I can have the marks and then validate the results creating some scenarios with different Office/Adobe programs. Battery tets will be performed as well.

My questions are:

1. Would you reccomend to use any complementary tool/step?

2. Do you think it is acceptable to assume that the use of OpenCL for office software would have little impact on the results? (So the conventional mode is acceptabe)

3. Are there more aspects that I should look at?

Thank you in advance!

 

[TheELF]

1. realbench it does tests with freeware, like image editing/manipulation with gimp video transcoding with x.264 and opencl too, and a multitasking test where they do it all together.

2.depends on the office...browsing,watching videos,image manipulation,rendering 3d or video and even database workloads can be accelerated,especially battery drain can be very different if you do things with the iGPU.

3.You should get into more detail of what the assignment is.

 

[Kardekton]

First of all, thank you very much for your quick and efficient reply. I've been looking some alternatives and I didn't find realbench then (and it really seems to be a very good complement).

Regarding the 3rd point, the main purpose of the study is to see if a group of "modified" computers -computers remanufactured from older computers- can perform as good as others, new computers in the market with close specs. The target of these computers are schools, MBE and corporates, so for example, putting special care in the GPU performance seems to be less important, due to the target.

After having read a bit about how benchmarks work (as I'm not an expert on it, as you can see), I believe that the best way of conducting this is to use application-based benchmarks, rather than comparing individually components. So, it would be more about comparing these benchmarks and getting the different measurements (i.e. spreadsheets, media, browsing...) in order to be able to compare them, so it can be concluded in which scenarios these computers will perform better or worse.

From the point of view of someone more experienced in this area, do you believe it may be a good approach for the objectives I'm trying to reach?

Once again, thank you for your help and time

 

[deustroop]

Do you need to test ? Pretty sure that the point of the testing may be determined a priori on the basis of technology: A spec is a spec and best spec wins, modified system or new. Performance is based on science and then technology,yes ?, i.e., the specs.If someone is looking to sell old computers as new, just rely on the specs.

 

[Kardekton]

I believe I do. I mean, this computers are meant to be "fit for purpose" which means that they're offered to fulfill a certain objective. You may have better -let's say- CPU and GPU which at it maximum usage would show great differences (i.e. games) but for some other functions the difference may be less significant (i.e. the running time for being able to do a certain workload in MS Office software). I will, for sure state the differences in specs, but I think that it may be not enough to show what I'm trying to.