lets continue where we left off
我们接着上次的继续。
>>109421894
嵌入式系统。
>>109471957
there are multiple contenders that do basically the same thing
有好几个竞争者都在做差不多的事。
for MPU systems I would like ThreadX to be the standard
对于MPU系统,我希望ThreadX能成为标准。
ThreadX is nice, clean, good docs, certification, open source, fast
ThreadX很好,干净、文档齐全、有认证、开源、速度快。
better than FreeRTOS (and its derivative, SafeRTOS)
比FreeRTOS(以及它的衍生品SafeRTOS)强。
u-VelOSity is... a bit obtuse. weird optimizations and non-optimizations made. but it is military grade and Green Hills also has a qualified compiler, toolchain, debug tools, good support and so on. strong contender
u-VelOSity嘛……有点难搞。做了一些奇怪的优化和非优化。但它是军工级,而且Green Hills还有合格的编译器、工具链、调试工具、好的技术支持等等。硬核竞争者。
theres also PikeOS that seems nice but very... german. I liked it though.
还有个PikeOS,看起来不错,但非常……德式。不过我挺喜欢它的。
i don't see that C# can be a good option for most systems. not sure how you would link code, symbols, libraries to a specific memory region for spatial separation on an MPU
我不觉得C#对大多数系统是个好选择。不确定你怎么把代码、符号、库链接到特定内存区域来做MPU上的空间隔离。
C is the golden standard, C++ works great also but slightly more inconvenient in some cases
C是黄金标准,C++也很棒,但有些场景稍微麻烦点。
for MMU systems there are, to my knowledge, fewer options.. PikeOS, Integrity, QNX
对于MMU系统,据我所知选项更少……PikeOS、Integrity、QNX。
Zephyr is a bit of a loser here due to its monolithic kernel
Zephyr在这块有点拉胯,因为它是单内核架构。
its not necessarily a disadvantage, but he industry is fearmongering to disregard anything bigger than a microkernel due to safety concerns
这不一定是个劣势,但行业里在渲染恐慌情绪,出于安全考量,凡是比微内核大的东西都被否定。
i know Zephyr also goes onto MPU systems, but its massive and much slower that the options listed so I dont see the use case
我知道Zephyr也能跑在MPU系统上,但它太臃肿,比上面列的那些慢多了,所以我看不出它有啥用武之地。
regardless, there will not be a standard. they all work pretty much the same, and you should use what you are comfortable with
不管怎样,不会有统一标准的。它们工作原理基本一样,你该用你觉得用得顺手的。
if you are a company you will probably buy a package, QNX+SafeRTOS, uVelOSity+Integrity, PikeOS, or some other
如果你是公司,你大概会买整套方案,QNX+SafeRTOS、uVelOSity+Integrity、PikeOS,或者其他什么。
>>109476088
How do you guys come up with project ideas? Not asking for any cookie-cutter tutorials or specific ideas, but just need a thought process to expand my portfolio to include something where I actually design something as a solution a problem or optimization that I identified.
你们都是怎么想到项目点子的?不是要那种模板化教程或者具体想法,就是想要个思路,好让我的作品集里能加点真正是为解决问题或优化而设计出来的东西。
I have access to a NXP MCU board, an Artix-7 FPGA, and a HackRF. I'm mostly interested in applying DSP concepts if that matters.
我手头有块NXP MCU板子、一块Artix-7 FPGA,还有个HackRF。如果有关的话,我主要对应用DSP概念感兴趣。
>>109476088
why would you want to get into embedded systems?
你为啥想搞嵌入式系统?
you know the pay is dogshit right?
你知道这行工资低得离谱吧?
>>109476409
think of something that would be nice to have, a silly problem you could make a bit more convenient, or just replicate something you think is cool. I made a smart powerstrip for my server rack because i thought it would be cool to know how much each one was consuming and to be able to remotely shut a server off while also protecting them from surges
想个“有了会方便”的东西,一个能稍微改善的小破问题,或者干脆复刻个你觉得很酷的东西。我给我的服务器机柜做了个智能排插,因为我觉得能知道每个设备耗多少电、还能远程关机顺便防浪涌,挺带劲的。
>>109477341
it's very fun imo, specially if you also like electronics. if not, try osdev
我觉得挺好玩的,尤其你也喜欢电子的话。不喜欢的话,试试搞OS开发。
>>109477341
>muh pay
>钱钱钱
spiritual jeet. white men do things out of curiosity and interest
精神修炼罢了。白人做事是出于好奇心和兴趣。
>>109476100
IMEO I remain a bit sceptical that defense contractors are buying the proprietary RTOS's because they are so determinisitic/fast/efficient.
依我看,我还是有点怀疑防务承包商买专有RTOS不是因为它们有多确定/快/高效。
IME working with customers in defense running VxWorks, they were really buying it because that is what the previous code base ran on. That code base got certified/tested/passed regulatory hurdles, and they are not spending the money to port it to Linux (even though they keep getting screwed in the long run by the RTOS vendor).
我跟做防务的客户打交道,他们跑VxWorks,买它纯粹是因为之前的代码库就跑在这上面。那套代码过了认证、测试、监管关卡,他们不会花钱移植到Linux(哪怕长期被RTOS厂商坑得死死的)。
Back in the 80's/90's when these project started RAM was limited, swap space was painfully slow and unpredictable with timing, so an RTOS made sense. Today, those limitations are pretty much gone, and if your really need determinism/fast response there is RT linux. If that is not fast enough/deterministic enough you need to use an FPGA.
这些项目起步于80/90年代,那时候RAM有限,交换空间慢得要命且时序不可预测,用RTOS合情合理。现在这些限制基本没了,真要确定性和快速响应有RT Linux。那还不够快/够确定的话,就该上FPGA了。
That's why I see the open source options taking over industry slowly. They may not be better in performance/features/support then the proprietary paid RTOS, but they are free, and IME the support from the paid vendors is pretty crappy/non-existent anyways.
所以我觉得开源方案会慢慢霸占这个行业。它们可能在性能/功能/支持上比不上收费的专有RTOS,但免费啊,而且以我的经验,付费厂商的支持本来也烂得跟没有似的。
I mostly see freeRTOS and zepher staying the dominant free options out there. FreeRTOS is the lightest and simplest thing you can practically want from an RTOS, and zepher is the more feature rich one for those sick of reinventing the wheel when they need some basic OS/system features not packaged into freeRTOS.
我看FreeRTOS和Zephyr会是免费阵营的两大主力。FreeRTOS是最轻最简、实际需求里最贴合的RTOS,Zephyr则是功能更全,适合那些需要FreeRTOS没打包进来的基础OS/系统功能、厌倦重复造轮子的人。
>>109476409
If you're interested in DSP and have an FPGA maybe a simple synth? You'd get then into some digital filter design.
你对DSP感兴趣又有一块FPGA,要不要搞个简单的合成器?这样就能上手数字滤波器设计了。
How do I get into FPGA's without breaking the bank? A simple dev kit would be nice
怎么才能不花大钱入坑FPGA?有个简单开发板就好。
>>109478134
That powerstrip idea sounds pretty awesome. Thanks anon for the advice, I should just be more creative.
那个排插点子听起来真牛。谢了匿名老哥的建议,我就是该更有创意点。
>>109478612
I'll give that a try and have some fun with it, though I do plan into getting into something in RF.
我会试试好好玩一把,不过我也确实打算往RF方向搞点东西。
>>109480189
How much is too much for you? You can always simulate your RTL designs for free through software, though you may enjoy that a lot less. The cheapest FPGA dev board out there is 4 dollars, but most beginners start with something that is around 100 dollars like the DE-10 lite or Arty S7.
对你来说多少钱算多?你总是可以通过软件免费仿真你的RTL设计,虽然你可能没那么享受这个过程。市面上最便宜的FPGA开发板只要4美元,但大多数初学者会从100美元左右的板子开始,比如DE-10 lite或Arty S7。
You can probably find what you need through here: https://boards.fpgadeveloper.com/find?sort=price
>>109476088
>osdev
Crowded by faggots btw
顺便说一句,这里挤满了傻逼
(如果你觉得这篇文章有启发,可以点击这里付费)