CONNECT SPEED - Users are STOP using CE

Hello Ian,

We do know that one source of the slowness is caused by the model picker from the Open page on the backstage.  We are currently investigating and working on a resolution to this for the MicroStation CONNECT Edition 10 timeframe.  We are still investigating other sources of slowness that have been mentioned on Bentley Communities as well and trying to track those down in order for us to resolve them as well in future updates.

Thanks,

Dan

Hi Dan,

I am feeling more and more disappointed by Bentley's reactions to users complains. In the past I was (or I think I was ;-) an advocate why to migrate to CONNECT Edition and because of MicroStation complexity, I accepted a first year after a release can be a bit painfull until code is cleaned and optimized.

But looking back how MicroStation has been developed, it seems to me like huge development failure. The problems with speed (open file, references, ribbon interactivity...) have been reported from very beginning, so to discuss specific scenarios and tickets after two years from release does not look credibly and trustworthy.

As developer I know after any code is implemented and working, it's not finished yet, because it can be (and often should be) refactored and optimized towards to low memory consumption and high performance. This is often tough and time consuming task requiring deep knowledge how used langauge (C++, C# etc.) works inside. But it makes from average product the top one. It looks like Bentley resigned to invest to any optimization and "it works" code is acceptable (which was not true in the past).

Ribbon (and the whole GUI) is good example: Response time is terrible (why to experience a delay with every tab or workflow change?) and it's not caused (as often people thougt) it's implemented in WPF and C#, because C# can be as fast as C++, but because not optimized code.  Funny / crazy thing is I remember Keith Bentley information long time ago that based on internal testing, a migration from 32bit to 64bit code brings about 10% performance increase (don't remember exact number, but it was not small). But the reality is competely opposite.

Resources are always limited, so when there is no effort in optimization and performance, I would expect focus on new features. But the list what has been added is not any "wow" and ideas and examples what is great in another solutions, described by many users, are ignored. So MicroStation is not anymore Bentley core product and it will be changed step by step to average "mostly works" product like so many others on market?

With regards,

  Jan

Jan Šlegr said:

I remember Keith Bentley information long time ago that based on internal testing, a migration from 32bit to 64bit code brings about 10% performance increase

If comparing like-for-like code (i.e. apples vs. apples), I'm sure that's true.  Unfortunately, Bentley Systems did not develop a 64-bit version of MicroStation V8i.  Instead, they developed a new product that uses many new technologies.  The new product has so many bells & whistles that it's slower than its predecessor.  We're looking at apples vs. oranges.

Compare that approach with, say, Intel's tick-tock strategy.  Intel's processors depend on two primary technologies: architecture and silicon. A tick-tock cycle lasts about two years.  The 'tick' refers to improved architecture.  The 'tock' refers to improved silicon.  They don't do both simultaneously.  They introduce each separately, about a year apart.  That gives Intel time to refine each change before proceeding to the next.  Has that strategy been successful for Intel and its customers?

Drawing an analogy with MicroStation, a 'tick' would have been a 64-bit port of V8i.  A 'tock' would have been the addition of bells & whistles to 64-bit V8i to develop CONNECT.  Unfortunately that didn't happen.

Jan Šlegr said:

I remember Keith Bentley information long time ago that based on internal testing, a migration from 32bit to 64bit code brings about 10% performance increase

If comparing like-for-like code (i.e. apples vs. apples), I'm sure that's true.  Unfortunately, Bentley Systems did not develop a 64-bit version of MicroStation V8i.  Instead, they developed a new product that uses many new technologies.  The new product has so many bells & whistles that it's slower than its predecessor.  We're looking at apples vs. oranges.

Compare that approach with, say, Intel's tick-tock strategy.  Intel's processors depend on two primary technologies: architecture and silicon. A tick-tock cycle lasts about two years.  The 'tick' refers to improved architecture.  The 'tock' refers to improved silicon.  They don't do both simultaneously.  They introduce each separately, about a year apart.  That gives Intel time to refine each change before proceeding to the next.  Has that strategy been successful for Intel and its customers?

Drawing an analogy with MicroStation, a 'tick' would have been a 64-bit port of V8i.  A 'tock' would have been the addition of bells & whistles to 64-bit V8i to develop CONNECT.  Unfortunately that didn't happen.