FM3 Firmware Version 5.03 beta 6 (release candidate)

Status
Not open for further replies.
I'm respectfully curious to know what you disagree with.
You removed the context. :D
I was continuing a discussion with @hippietim about tight loops. Probably two different worlds. Mine in distributed systems, server side development (Locks, multithreading, synchronizations, volatiles), his probably embedded software (interrupts, semaphores, sleep)
 
Probably two different worlds. Mine in distributed systems, server side development (Locks, multithreading, synchronizations, volatiles), his probably embedded software (interrupts, semaphores, sleep)
Yes, very different worlds. Memory, storage and CPU are very constrained in the modelers, and we can’t simply throw virtual machines or VM servers at the problem.

It’s been years since I had to write in assembly language for tiny processors without expandable memory or storage, but we’d often be counting clock cycles and the sizes of the instructions to fit code into the available time and memory available. Fractal used to do that, but the new CPUs in the current hardware have better compilers so they can take advantage of their optimizations; the Wiki has quotes from Cliff about the differences in the generations of the hardware. It’s interesting reading.
 
You removed the context. :D
I was continuing a discussion with @hippietim about tight loops. Probably two different worlds. Mine in distributed systems, server side development (Locks, multithreading, synchronizations, volatiles), his probably embedded software (interrupts, semaphores, sleep)

If you need a loop, it should be tight. That doesn't mean poorly written or inescapable.

I just hit the 25 year milestone working at Microsoft. Before that I worked in the defense world - logistics, AEGIS radar, cruise missiles, etc. Most of my career has been writing C/C++ and C# code for PC/Windows platforms. But I've written code in more languages on more platforms than I feel like typing (or remembering).

In terms of server side development, I worked on Azure DevOps (and Team Foundation Server) for like 7 years and a very brief stint on Xbox Live. So I do get where you're coming from.

My point was that what you said were indicators of bad software design is not true in many scenarios. What you're striving for in your server development is completely inappropriate for a couple of the projects I've worked on the last few years. When the CPU isn't pegged, I try to find out how to use the spare cycles.
 
If you need a loop, it should be tight. That doesn't mean poorly written or inescapable.

I just hit the 25 year milestone working at Microsoft. Before that I worked in the defense world - logistics, AEGIS radar, cruise missiles, etc. Most of my career has been writing C/C++ and C# code for PC/Windows platforms. But I've written code in more languages on more platforms than I feel like typing (or remembering).

In terms of server side development, I worked on Azure DevOps (and Team Foundation Server) for like 7 years and a very brief stint on Xbox Live. So I do get where you're coming from.

My point was that what you said were indicators of bad software design is not true in many scenarios. What you're striving for in your server development is completely inappropriate for a couple of the projects I've worked on the last few years. When the CPU isn't pegged, I try to find out how to use the spare cycles.
Congrats on the 25th anniversary.
I started in '93 on os2/SunOS/freebsd (trading/realtime) and dabbed into windows for a year (a wasted one - '96).
 
My birthday present, a first generation FC-12, arrives today!
Not my birthday, but me too! I mean, come on - a close-out deal on a Mark 1 FC-12, for less than the price of a new FC-6!? Yes, please! I wasn't going to get an FC-12, but I couldn't pass that up.

Oh - and Happy Birthday! :)

EDIT: aaaand it's here! Woo hoo!! OMG15, baby!! Very thankful FAS put on this closeout deal!! :D
 
Last edited:
View attachment 102176

Maybe recalibration of the CPU meter is slowing them up? :p:tearsofjoy::p

EDIT: I may have posted this too soon. Just looked and there have been past FW Fridays at around lunchtime. We'll see.
Word on the street says they were going to do a release today. But then this endless discussion about how the CPU usage reads delayed it. It‘s now become a “major issue” that has diverted all available resources so they can have the fix out ASAP…
 
Not my birthday, but me too! I mean, come on - a close-out deal on a Mark 1 FC-12, for less than the price of a new FC-6!? Yes, please! I wasn't going to get an FC-12, but I couldn't pass that up.

Oh - and Happy Birthday! :)

EDIT: aaaand it's here! Woo hoo!! OMG15, baby!! Very thankful FAS put on this closeout deal!! :D
I wish I could benefit from this deal here in Europe 😔
 
On the technical side, I'll also just chip in that an embedded DSP system running some type of real-time OS is a world apart from a modern pre-emptively scheduled speculatively executed multicore "PC" type platform. Anyone know what the error bars on that CPU usage percentage? It's not an exact measurement, you know. Could be ±1%, or ±5%? We know there's background stuff going on in the processor (UI, USB, other?) - how tall and wide can the spikes be? In fact while we're on the subject, let's not forget that latency and throughput are separate and often competing quantities in tuning the performance of any given system design.

Myself, I'm coming from a professional game development background and while I understand many of the issues and discussion at a high level, I am definitively NOT a real time DSP programmer. From an engineering mindset I actually like the idea of setting the 100% mark at a nominal safe level and allowing the counter to exceed 100% into a "danger zone". This to me is a better expression not of the technical CPU cycle situation (which frankly is not very important to a user) but rather of the actual available processing power and risk level (which is extremely important). However I think you'd never really get away with that type of readout in a mass market device. It'd confuse more people than it would help.
 
I did provide references “word on the street”
There’s always a possibility that there will be a firmware release, so every day can be a “Software Friday”. That “word on the street” for any retooling of the CPU load reporting is unfounded without a reference, and without it can only be considered wishful thinking at best, or rumor mongering at worst.
 
There’s always a possibility that there will be a firmware release, so every day can be a “Software Friday”. That “word on the street” for any retooling of the CPU load reporting is unfounded without a reference, and without it can only be considered wishful thinking at best, or rumor mongering at worst.

Pretty sure he was speaking in jest, and never meant to be taken so "seriously" by the forum PoPo. :)
 
Pretty sure he was speaking in jest, and never meant to be taken so "seriously" by the forum PoPo. :)
I appreciate your words. Thanks.

Emoticons, and ensuring what we write matches the tone we want to convey, are important when operating behind the keyboard. I actually use Grammerly’s plug-in to double-check inside my desktop browser because it does a good job analyzing the tone of our speech to help catch unintended tone in what I write.
 
Status
Not open for further replies.
Back
Top Bottom