PDA

View Full Version : Corporate environment: minimum score



chrisipeters
09-20-2011, 03:16 AM
The company I support is investigating using Passmark as a tool to benchmark our environment. What I'm looking to do is:

1. determine a 'topline' score for our PCs. Each Hardware model supported, each running our standard image.

2. determine a 'bottomline' score which to preserve an acceptable user experience we must stay above.

What I'm hoping to be able to do is to determine (by installation and running tests) the performance cost for each application. This would allow us to determine whch machines must be upgraded before they can run appX + appY + appZ.

The topline is easy enough to determine, simply running the performance tests on a freshly built machine.

What I am struggling with is the botomline number. Does anyone here have experience with doing something similar? Has anyone determined such a 'botomline' number which to preserve an acceptable user experience the machine must stay above?

Thanks in advance for your replies and any assistance / information you all can provide.

David (PassMark)
09-20-2011, 04:35 AM
As what is acceptable is based on user expectations and your budget, it will be hard to define a number that will apply to any more than small user groups.

640KB of RAM and a 100Mhz CPU were very acceptable... some years ago...

chrisipeters
09-20-2011, 04:46 AM
Thanks for the quick reply! let me try and clarify:

I'm not looking for minimum hardware requirements, Im trying to build a performance costing model for applications.

regardless of hardware (and Im making these numbers up) if a machine comes back with a score of 1500 with full software load, is this an acceptable score for user experience.
If a machine has that score of 1500, should we say that it is maxed out, and it cannot handle any additional software load?

I want to get to the point where:
AppX will lower the score of a machine by 100
AppY will lower the score of a machine by 500
AppZ (think video editing / compiling) will lower the score of a machine by 1,000

new hardware comes out, gets passmark score of 4500. I install all 3 apps on it, effectively lowering its score by 1,600. It now has a score of 2,900, which is more than our bottomline 1500.

Old hardware has a passmark score of 2500 and it meets the hardware requirements of AppZ. I install AppX and appY leaving it with a score of 1,900. the user requests appZ, to which IT responds they need a new machine first because their current machine cannot handle this load.

Does this make sense?

David (PassMark)
09-20-2011, 07:38 AM
I think you are under the incorrect assumption that the installation of software will slow the hardware down. With a few exceptions, this is not the case.

The installation of a compiler, to use your example, will take up disk space and RAM (if loaded and idle), but won't impact the CPU speed or disk speed.

The only way your scheme would kind of work would be if the applications where active (e.g. in the middle of compiling code, or doing video encoding) when you did the test. Which would be very tricky to arrange, but still futile. It would be futile as you would then be measuring the load of the task, but not the duration of the task. So a task that uses 4 cores in a CPU might place a very heavy load on the system for a short period. But the same task on 1 core might run much longer at lighter load.