No argument. I could be “wronger” than a 13 year old male prostitute discovered in a hotel room with your preacher. Why I think it is threading. Setting to 20 fixes it. FR is legacy code. FR used to be magnificently fast. HT had lots of OS issues. Older compilers have no idea of the multicore nature of todays CPUs even run time ones. The repeatability at a point of time of the failure and that it is not consistent over time. Have seen it over and over again. IMHO it is a problem with FR running on a multicore system. Best fix IMHO is run FR in a vm emulating a single core system and let the OS and VM handle the switching.
Have seen it over and over again. IMHO it is a problem with FR running on a multicore system. Best fix IMHO isIt is a threading problem. Multicore CPU’s don’t play well with legacy code. I know. I code. Have John hunt around in BIOS and turn of hyperthreading and anything to do with multi threads. Add an API call to bind any single request to a single core.5 posted on 7/14/2017, 4:37:10 PM by LesbianThespianGymnasticMidget
run FR in a vm emulating a single core system and let the OS and VM handle the switching.24 posted on 7/14/2017, 6:03:49 PM by LesbianThespianGymnasticMidget
I claim no expertise. The discussion puts me in mind of when my boss’s boss’s boss commented to me about a meeting in which I had argued a technical point about aircraft vibration. He said he wasn’t qualified to understand what I was saying, but with his legal background (!) he had asked every question he could think of every way he could, and he could tell that at least my answers were consistent.But the solution LTGM described in post 24 sounds like:
- it could be a solution, and
- implementation probably would not be nearly as hard as a lot of other possibilities, including post 5.