Tuesday, July 14, 2015

CPU Power Sagnation

A while a go (2008) I did a study on what drove me to upgrade my computer (the large green circles with date and MIPS labels). Typically I found this was around 5 to 10 times the speed of my previous computer was the tipping point to buy a new computer.  This has typically meant a three year life cycle for my computers.

I determined this from a simplistic analysis of the Drystone (MPS) benchmarks over the history of my computer life (roughly since 1982). Interestingly enough there seems to be a power function which has a reasonable fit over time for both the Drystone results and processor speed MHz.


My reason for revisiting this study is lately I have been underwhelmed by the latest mobile CPU's coming out from Intel and feel they have been stagnating or in some cases going backwards in performance. Interestingly enough I came across a similar study that concluded a 52% yearly increased per year prior to 2004 and 21% post 2004.

Even though my data is obviously much sketchier I think you can see a similar change in gradient pre and post 2004.  An interesting point is that we both developed similar graphs independently of each other pretty much interrogating the same questions about rate of change of CPU speed.


When you put these rates in terms of my tendency to update every 3 years it becomes quite dramatic. As you can see why I am questioning my normal 3 year update may now be a 5 year update.


1 Year Performance Improvement 3 Year Performance Improvement 5 Year Performance Improvement
Pre2004 1.52 3.44   7.85
Post2004 1.21 1.77    2.59

Interestingly enough my intuition that there was something going on with CPU speeds was enough to perk my interest sufficiently to investigate my computer upgrade history was only a few years after the 2004 tipping point observed in the study above.

I guess I am not the only one with shrinking  consumer demand and pushing product cycles.  Call me cynical ... is Intel's delay due to production or to extend current product life cycle.

Saturday, July 11, 2015

Beyond Maturity

Since the development of the Capability Maturity Model Integration (CMMI) by the Carnegie Mellon University Software Institute (SEI) and the US Air Force, every man and their dog have come up with their own maturity index's that are essentially the same as or close to the CMMI.

My premise is that there must be something beyond a simple assessment of a management process and I think that is the actualisation of that management process.  The idea is to not measure the theoretical effectiveness the organisation's management process but to measure how productive and effective is the output of the management process. ie the value of rapid actualisation.

 I think this draws some broad similarities to the ISO System Engineering, Design for Six Sigma, AGILE, CMMI and my own rapid value actualisation method.

ISO 15288 Six Sigma AGILE Capability Maturity Model Assessment Rapid Value
Actualisation
Stakeholder Requirements Definition Process
(Clause 6.4.1)
Requirements Analysis Process (Clause 6.4.2)
Define Discovery Initial Benefits Identification/
Earned Value Framework
Architectural Design Process (Clause 6.4.3) Measure Design Managed Model Based Design
Implementation Process (Clause 6.4.4)
Integration Process
(Clause 6.4.5)
Verification Process (Clause 6.4.6)
Analyse Develop Process Defined Rapid Prototyping/
Initial Benefits/EVM
Transition Process
(Clause 6.4.7)
Validation Process
(Clause 6.4.8)
Design Integrate & Test Quantitatively Managed Change Management
Operation Process
(Clause 6.4.9)
Maintenance Process (Clause 6.4.10)
Disposal Process
(Clause 6.4.11)
Verify Release Optimised Application Implementation/
Benefits Realisation

I would have liked to have worked Prince2 into the above table but its more a management framework to be overlaid onto a product development cycle. This article has the best view I have seen of the seven Prince2 Management processes.



Although this article does have a good view on overlaying Prince2 onto a project/product development process.