z****e
发帖数: 54598 | 1 以后去写微波炉我都不奇怪
很正常嘛
不适合大项目的就适合搞一些小玩意 | k**********g
发帖数: 989 | 2
Any code needing numerical and signal processing would still need to write
in C or C++ or OpenCL via JNI.
For now, Java and Microsoft MSIL JIT still do not generate SIMD code
directly. Only Mono has SIMD support.
SIMD is significant for heavy numerical processing because it processes e.g.
4 - 16 values at once. GPU is even more powerful. These two techniques will
continue to scale up, in part closing the Moore's Law gap.
High-level libraries e.g. graphics, mathematical, financial, video
processing packages already wrap around these low-level details, so it is
possible that a Java or high-level programmer might not even know that a low
-level implementation is used indirectly.
However, there are always situations that need custom implementation that
isn't provided by existing libraries. Knowing how to get that done will open
you to a niche market that is quite stable. As long as you still know high-
level programming languages, and be comfortable doing the majority of work
there.
【在 z****e 的大作中提到】
: 以后去写微波炉我都不奇怪
: 很正常嘛
: 不适合大项目的就适合搞一些小玩意
| z****e
发帖数: 54598 | 3 graphics和math都快守不住了
unity做出了很好的表率
math的话，就看script engine了
你说的这些，聪明人其实看到了机会
聪明人会想办法去搞定这些东西
然后交给我们懒人去用
所以光喊说需要需要，是没有用的
现在需要，从长远看，只会越来越少
这个真的是顺之者昌，逆之者亡
聪明人就顺着推一把，然后给自己赚取足够的credits
比如renjin那些，这就是典型的math，统计就是典型的应用数学
他们已经在做了，你还等什么？
这不是我的领域，我更多的只是看个热闹，但是如果你从事的是这些领域
哼哼，那自求多福吧，实际上现在c++找工作跟十年前已经不是一回事了
low
open
high-
【在 k**********g 的大作中提到】
:
: Any code needing numerical and signal processing would still need to write
: in C or C++ or OpenCL via JNI.
: For now, Java and Microsoft MSIL JIT still do not generate SIMD code
: directly. Only Mono has SIMD support.
: SIMD is significant for heavy numerical processing because it processes e.g.
: 4 - 16 values at once. GPU is even more powerful. These two techniques will
: continue to scale up, in part closing the Moore's Law gap.
: High-level libraries e.g. graphics, mathematical, financial, video
: processing packages already wrap around these low-level details, so it is
| k**********g
发帖数: 989 | 4 No worries, I use C# too.
我会 MATLAB，不懂 R，这反而是件比较担心的事情。
The next wave of computing will be back to hard science | c*********e
发帖数: 16335 | 5 嵌入式的早就是c的市场了
【在 z****e 的大作中提到】
: 以后去写微波炉我都不奇怪
: 很正常嘛
: 不适合大项目的就适合搞一些小玩意
| z****e
发帖数: 54598 | 6 但是主要的已经不是实现过程了
而是理论了
你会r或者matlab没有用啊
最重要的是要会统计
写游戏倒是不需要啥特别高深的理论知识
next wave是啥都不要太care
说出来很多时候都不能保证的
跟算命差不多，最重要的是把握当下
现在什么能赚钱就赶紧
以后事真说不清楚
【在 k**********g 的大作中提到】
: No worries, I use C# too.
: 我会 MATLAB，不懂 R，这反而是件比较担心的事情。
: The next wave of computing will be back to hard science
| z****e
发帖数: 54598 | 7 对硬件不熟
但是很明显感觉到
c++的人说自己搞硬件
引起一堆人的排斥
感觉纯粹胡说八道
看起来c++属于典型的上不上，下不下
【在 c*********e 的大作中提到】
: 嵌入式的早就是c的市场了
|
|
