l*******o
发帖数: 250 | 1 not every time, but happened several times at least. Only got issues with
this site.
C:\dev\vert.x>gvim bin\vertx.bat
C:\dev\vert.x>vertx run Server.java
"C:\Java\jdk1.7.0_45/bin/java.exe" -Djava.util.logging.config.file="c:\
dev\vert.x\bin\..\conf\logging.properties" -Dvertx.home="c:\dev\
vert.x\bin\.." -Dvertx.clusterManagerFactory="org.vertx.java.spi.cluster.
impl.hazelcast.HazelcastClusterManagerFactory" -classpath ";c:\dev\
vert.x\bin\..\conf;c:\dev\vert.x\bin\..\lib\*" org.vertx.java.... 阅读全帖 |
|
L***s
发帖数: 1148 | 2
I used to compile and install ATLAS,
and I was aware of two kinds of optimization:
1. Matrix/vector multiplication is done at the block-level,
as opposed to element-wise. So it has to be knowledgeable
on your memory hierarchy, e.g., speed and sizes of your
register, L1/L2/L3 caches, main memory, etc.
ATLAS picks the best preset parameters accordingly.
2. ATLAS performs local search on the best preset parameters,
conducts actual performance profiling, and picks the best set
of empirical paramet... 阅读全帖 |
|
a*********a
发帖数: 3656 | 3 also some times const vars don't have to be allocated at all. they can be
placed inline by the compiler. the compiler can even do some computation at
compile time.
static const int a = 5;
static const int b = 2;
...
int c = a+b;
I suspect the compiler can just init c to 7, no run time operations. if this
is the only place a and b are used, no address would be allocated for a and
b on stack at all. |
|
v******y
发帖数: 84 | 4 这是我的理解，不对的请指出，谢谢
The usage of rvalue reference &&
Define a class
class MyClass{
public:
MyClass(){/*require for const MyClass cMyClass init. Without this default
constructor, const MyClass will have to be declared and defined as such:
const MyClass cMyClass=MyClass(). MyClass() here will call synthesized
constructor automatically generated by compiler. This synthesized
constructor can not directly initiate const. So following is illegal without
default constructor: const MyClass cMyClass;*/}
};... 阅读全帖 |
|
w***g
发帖数: 5958 | 5 java按官方说法是先compile成字节码，然后用VM执行。应该算是compile，跟解释还是
有本质区别的。不排除compiler做的很烂跑得比解释执行还慢的可能。
Java的字节码原来是VM解释执行的，嫌慢。JIT是指just in time把字节码翻译成机器
码后再执行。但是字节码设计得实在很烂，跟主流CPU架构很不一样，导致翻译过程比
较复杂，所以才有compilation一说。 |
|
F****n
发帖数: 3271 | 6 cache access 和 memory barrier 从reorder 优化的角度是一个道理
为什么recorder优化可行？是因为compiler发现你程序某些部分并不互相关联
所以把顺序换换也没关系。但在并行计算的情况下compiler很难通过语义分析（对C/C+
+来说）确定关联性，所以需要额外的memory barrier，它的一个作用就是把一定的顺
序强加在使用不同cache但访问同样内存地址区的并行进程。
为什么决大多数情况下用volatile退出线程不对？ 简单的说，因为有数据一致性
要求时volatile不工作，没有数据一致性要求时，你根本不需要volatile。大多数情况
下你总不可能什么都不做就直接quit吧？总要对一些数据进行操作和判断然后决定是否
quit。举个不太恰当的例子：
Connection conn; // assumed to be a db connection;
bool volatile quit;
...
while (!quit) {
conn.doSomething(); // error if conn is closed
... 阅读全帖 |
|
|
z*******w
发帖数: 79 | 8 /*
* !!!! Private do not distribute !!!!
*
* <1080r.c> socks5 remote exploit / linux x86
*
* Usage:
* $ ./1080r [offset]
*
* Vulnerables:
* socks5-v1.0r10 (compiled on a turbolinux 4.0.5) => 0
* socks5-v1.0r9 (compiled on a turbolinux 4.0.5) => 0
* socks5-v1.0r8 (compiled on a turbolinux 4.0.5) => 0
* socks5-v1.0r10 (compiled on a redhat 6.0) => 400
* socks5-s5watch-1.0r9-2 (redhat-contrib) => no?
* socks5-0.17-1 (redhat 4.2) => no
* socks5-1.0r10-5 (redhat-contrib) => no??
* so |
|
c*******o
发帖数: 1722 | 9 dunno if you guys use vim latex-suite or not.
if no, you should check it out. it is so neat.
in the visual mode, select the part you want to
compile and check for changes, do \ll (\ll is the
latex-suite command for compile, of cuase you can
configure what \ll really does). then only the
visually selected part will be compiled and you can
view it too by \lv (again, latex suite command).
it will automatically choose to use your original
preambles for the compile.
after you exit vim, latex-suite wi |
|
c*****t
发帖数: 1879 | 10 #pragma are compiler directives. They are compiler specific. So,
you need to know the orginal compiler and the current compiler
and then read their manuals for more information. |
|
D**e
发帖数: 10169 | 11 gcc是个C compiler, g++是C++ compiler.
它们偏偏又都是GCC的一部分. 鬼知道GNU
怎么把名字搞得这么乱.
GCC这么大写, 是GNU COMPILER COLLECTION
的缩写. 小写, gcc, 呢就是C compiler.
TMD, 这帮人真TMD. |
|
p****o
发帖数: 220 | 12 Bloodshed Dev-C++
Bloodshed Dev-C++ is a full-featured Integrated Development Environment (IDE)
for the C/C++ programming language. It uses Mingw port of GCC (GNU Compiler
Collection) as it's compiler. Dev-C++ can also be used in combination with
Cygwin or any other GCC based compiler.
Features are :
- Support GCC-based compilers
- Integrated debugging (using GDB)
- Project Manager
- Customizable syntax highlighting editor
- Class Browser
- Code Completion
- Function listing
- Profiling support |
|
t**********g
发帖数: 85 | 13 My experience is, even if you compile and run it using different compilers and
in different UNIX systems, you could get very different results after some
iterations. It's hard to say which one is better, because compilers might has
different explain of your source code. So, I think the most reliable results
come from the compiler and system in which you write your code, because,
generally, you will test it step by step to see if it creat reasonble results.
BTW, my experience based on FORTRAN, C+ |
|
r****y
发帖数: 1437 | 14 For single precision, the machine precision is about 1e-7. Your result
is correct within machine precision.
For double precision, if you explicitly use 2.0D0, 1.0D-4. You will
get higher precision than single precision. If just 2.0, then your compiler
might convert it to 2.0000011111.
Different compilers can be a problem as well. Recently I have a code
working perfectly on SGI, Sun UNIX, Linux gnu compiler, but Linux PGI compiler
will give me a wrong answer. |
|
O******e
发帖数: 734 | 15 If you really need to handle general signals using native Fortran,
you will need a compiler that supports the PXFSIG* subroutines
defined in the POSIX Fortran binding standard IEEE 1003.9-1992.
It might be easier to use mixed Fortran/C++ compilation to do
what you want, if your compiler supports mixed language calling.
If you only want to deal with numerical exceptions, check whether
your compiler supports the Fortran 2003 ieee_exceptions intrinsic
module. |
|
h***z
发帖数: 233 | 16 A few things to try:
1. Compiler options: Turn on compiler optimization as well as any SIMD
2. Libraries: Link against pre-compiled libraries tuned for your computer
architecture and tune parameters of any libraries that you compile yourself.
3. Your code: Keep in mind the cache size of your CPU and write your code to
avoid cache thrashing. |
|
b******e
发帖数: 34 | 17 C is a derived class from B, which of the following is true:
B* C:: fn(C* c)
{
if (c !=0)
{
c->clear();
return c;
}
else{
return NULL;
}
}
1. This will compile and run successfully
2. This will compile, but may cause an addressing exception at run time.
3. This will not compile.
4. More info is needed to determine whether this will compile and run. |
|
y***s
发帖数: 23 | 18 I happened to test the 1st program a few days ago.
1. First compile the functions needed in the main program:
g++ -c function.cpp (-c: just compile to generate binary file. No linking
here)
2. compile and link the main program
g++ main.cpp function. o
This is the idea(compiling and linking). |
|
z****g
发帖数: 1978 | 19 depends on the memory. The default compilation is 30M or 40M. You can
compile the source code yourself and set a specific MACRO for the memory
size. Current R source code is stable for 32 bit compiler, and the memory
upper limit is the same as the standard 32 bit application, which is 2G.
64bit compilation is not recommended as there may be unknown type bug.
Also, there is no sparse matrix implementation for R. For further
optimization, you can try to build the ATLAS DLL on your machine to optim... 阅读全帖 |
|
G*****7
发帖数: 1759 | 20
legacy code. old timers like bloomberg, ansys and nastran (siemens plm) have
gazillion tons of legacy code. rewriting them in c++ is simply financially
in-feasible. that's why nvidia and pgi invested in the new openacc standard
to enable compilation from tagged legacy fortran code into gpu-parallel
binaries.
open source tools come at the price of business-hostile, sadistic licensing
terms.
not 转向. ms still invests in its cl(vc) compiler for c/cpp. they have
produced F# as well, which doesnt nec... 阅读全帖 |
|
t******g
发帖数: 372 | 21 于server local与否无关，或者至少给些error log 别人也好知道你尝试了啥，
S/W Requirements:
-----------------
MySQL client library ("libmysqlclient.so") and header files
(see www.mysql.com). For Linux/intel you may use the binary rpms
client+devel+shared, and make sure you also have the header files,
they are *NOT* included in the following
MySQL-client-4.1.7-1.i386.rpm
MySQL-shared-4.1.7-1.i386.rpm
but they are included in the latest 4.1.7 rpm versions.
The library libz. This is a stan... 阅读全帖 |
|
z*******n
发帖数: 1034 | 22 http://www.infoq.com/research/mobile-apps-stack
by Ralph Winzinger on Mar 18, 2014
Native toolkits
Apple Xcode (iOS): Apples full featured development environment lets you
create mobile apps as well as desktop applications. Xcode comes with
integrated support for Git source-code repositories, graphical editor to
build user interfaces, instrumentation and debugging tooling and integrated
documentation. You need Mac OS X to run XCode and it is free of charge.
IntelliJ AppCode (iOS): Jetbrains, cre... 阅读全帖 |
|
z*******n
发帖数: 1034 | 23 https://stackoverflow.com/questions/24101718/swift-performance-sorting-
arrays#comment37183103_24102237
tl;dr Swift without aggressive compiler optimizations at this stage is very
slow; with them it is very fast. Keep it in mind.
Here is an in-place quicksort in Swift:
func quicksort_swift(inout a:CInt[], start:Int, end:Int) {
if (end - start < 2){
return
}
var p = a[start + (end - start)/2]
var l = start
var r = end - 1
while (l <= r){
if (a[l] < p){
... 阅读全帖 |
|
m***f
发帖数: 732 | 24 【 以下文字转载自 Outdoors 讨论区 】
发信人: motif (魔地佛), 信区: Outdoors
标 题: 写了个 SAC的抓deal code
发信站: BBS 未名空间站 (Wed Nov 3 17:01:39 2010, 美东)
抓到你想要的deal自动给你发email或短消息，有兴趣的下去改了用。直接拷贝下面这
段code修改后存为sac.py,然后运行python sac.py即可。在python 2.7下测试通过。
#!/usr/bin/python
import re
import urllib
import smtplib
import time
from email.mime.text import MIMEText
target=[['trekking', 'poles'],['Napali', 'Backpack'],['north', 'face'],['
oakley','battalion']] #The items you want to monitor, in every set of [] AND
logical will be us... 阅读全帖 |
|
f******d
发帖数: 2394 | 25 公开点名中俄，为示公平，搭上小兄弟沙特。在中国股市开始下跌前，中国政府被人民
币国际化和亚投行的“光辉”绑架不能主动让人民币崩盘前，是最后的逃离机会。
Once Over $12 Trillion, the World’s Currency Reserves Are Now Shrinking
Bloomberg Business
The decade-long surge in foreign-currency reserves held by the world’s
central banks is coming to an end.
Global reserves declined to $11.6 trillion in March from a record $12.03
trillion in August 2014, halting a five-fold increase that began in 2004,
according to data compiled by Bloomberg. While the drop may be overstated
becaus... 阅读全帖 |
|
n******g
发帖数: 17225 | 26 北京大学周二(5月5日)上午为一座以德国哲学家卡尔·马克思的名字命名的教学楼
举行奠基仪式，以庆祝其诞辰197周年。
北京大学马克思主义学院的副院长孙代尧称，这将是世界上首座以马克思命名的大
楼，位于大学南门附近，将在明年年底前完成。
英文原文：
Peking University lays Marx center foundation
Peking University held a foundation laying ceremony Tuesday morning for
an academic building named after German philosopher Karl Marx, in
celebration of the 197th anniversary of his birth。
Sun Daiyao, the associate dean of Peking University's School of Marxism,
said this will be the world's first building named after Marx, news ... 阅读全帖 |
|
x*****z
发帖数: 787 | 27 北京大学周二(5月5日)上午为一座以德国哲学家卡尔·马克思的名字命名的教学楼举行
奠基仪式，以庆祝其诞辰197周年。
北京大学马克思主义学院的副院长孙代尧称，这将是世界上首座以马克思命名的大
楼，位于大学南门附近，将在明年年底前完成。
英文原文：
Peking University lays Marx center foundation
Peking University held a foundation laying ceremony Tuesday morning for
an academic building named after German philosopher Karl Marx, in
celebration of the 197th anniversary of his birth。
Sun Daiyao, the associate dean of Peking University's School of Marxism,
said this will be the world's first building named after Marx, news ... 阅读全帖 |
|
发帖数: 1 | 28 我学理科，又是做学术的，过去20年，所有版本的C++，我都用过。这是学术圈做理论
的科研工作者的标配。就连cross compiler，我都做过。
从新版的compiler是构造上越来越复杂，但功能越来越powerful，运算效果越来越强，
效率越来越高。
高性能计算如果再用以前的老版，会对新版的Ubuntu，新版MacOS和iOS的Xcode里的
compiler有严重的incompatible issue，根本没法execute了。
做研究的，这是基本功。 |
|
n********g
发帖数: 6504 | 29 俺们烂校CS本科也需要写File System和进程调度子系统。用C在Unix终端上写。用VI码
code。
俺们烂校CS也需要每人自己写Compiler。包括能编译算术表达式、函数那种。
相比之下，电脑图形课太简单了。都忘了怎么水过。
引文作者已经是总经理级别。印证了他的行业观察，做管理的都是代码写得不行的。留
下写代码的都是代码写得好的。
以下为引文：
我在布朗大学电脑科学系读博士修课的时候，程序作业的份量非常重。上作业系统(
OperatingSystem)课的时候，教授要我们每个人独立写一个包含File System与
Process System的迷你Unix作业系统；
上编译程序(Compiler)课的时候，教授要我们每个人独立写一个 Compiler，而且每一
个模组还必须用两个以上的方法写，然后互相比较；
而上 Andy Van Dam 教授的电脑图学，那简直就是人间炼狱。当时助教发问卷调查，发
现每位学生每周花在写那堂课的程序的平均时间超过四十个小时，学生们几乎要群起造
反。 |
|
发帖数: 1 | 30 你这里有几个问题。你说的 编译器 如果只是写front end 那个和写一个完整的编译器
差别大了去了。
第二个问题 知识的普及程度问题。现在的互联网上面 各种示例代码开源项目 文章介
绍 随处可见。三十年前 写一个迷你os 和 现在写一个迷你 os 的难度差别大了去了
[在 niuheliang (别问我是谁) 的大作中提到：]
:俺们烂校CS本科也需要写File System和进程调度子系统。用C在Unix终端上写。用VI
码code。
:俺们烂校CS也需要每人自己写Compiler。包括能编译算术表达式、函数那种。
:相比之下，电脑图形课太简单了。都忘了怎么水过。
:引文作者已经是总经理级别。印证了他的行业观察，做管理的都是代码写得不行的。
留下写代码的都是代码写得好的。
:以下为引文：
:我在布朗大学电脑科学系读博士修课的时候，程序作业的份量非常重。上作业系统(
:OperatingSystem)课的时候，教授要我们每个人独立写一个包含File System与
:Process System的迷你Unix作业系统；
:上编译程序(Compiler)课的时候，教授要我们每个人独... 阅读全帖 |
|
d*********8
发帖数: 2192 | 31 The original and reference implementation Java compilers,
virtual machines, and class libraries were originally released by
Sun under proprietary licenses. As of May 2007, in compliance with the
specifications of the Java Community Process, Sun relicensed
most of its Java technologies under the GNU General Public License.
Others have also developed alternative implementations of these Sun
technologies, such as the GNU Compiler for Java (byt... 阅读全帖 |
|
m******r
发帖数: 485 | 32 程序语言其实也是讲语言的基本概念，跟compiler有交集，其实也没有compiler难，编
compiler你基本上定义了一个语言。 |
|
B*Q
发帖数: 25729 | 33 你也可以写个西语言的编译器
俺觉得比自定义的难
自定义的简化了不少
[在 matrixtr (密) 的大作中提到：]
:程序语言其实也是讲语言的基本概念，跟compiler有交集，其实也没有compiler难，
编compiler你基本上定义了一个语言。 |
|
l*****0
发帖数: 64 | 34 提供一个公司急招信息：招QA Engineer Full time. New grades 或 experienced 的
都可以。要求如下：这个要的急，站内联系。谢谢！大公司
希望是有相关QA经验的，或是熟悉Compiler的。真实信息。
The job details are showing below.
1. Software Quality Engineer (Based in Santa Clara, California)
Description
Debugs software products through the use of systematic tests to develop,
apply, and maintain quality standards for company products. Develops,
modifies, and executes software test plans. Analyzes and writes test
standards and procedures. Maintains documentation of te... 阅读全帖 |
|
c****p
发帖数: 32 | 35 嗯，但是也可以从另一个方面来说不需要recompile.
如果已经explicitly 有了其他ctor，加入新的没有影响。
这个基本上是要重新compile，一个例外是加入原来的class本来就只是method的封装比
如：
class Methods
{
public:
staic void test1();
}
这样加入new member fields也没什么关系（但是如果client code有类似sizeof(
Methods)的就不行）
这个一定要重新compile.其他的例子都可以找到反例（虽然有点急转弯的意思），但是
因为dtor是一定要被用到的，如果原本的dtor变成virtual，那么client端整个调用过
程就变了。原来是：
$Class_Dtor();
变成virtual后变成了
pClass->vtbl[index_of_dtor]();
所以client端必须重新compile.
反例是如果这个member function根本就没用到:D |
|
x****k
发帖数: 2932 | 36 第一题应该是 static const int SIZE = 100;（Effective C++, 3rd edition
, item 2）
具体结果depends compiler和编译选项。32 24是可能值之一。g++下是28 20，static
const int size不占sizeof的空间，按照4byte alignment。
可以用 -fdump-class-hierarchy dump出class的layout文件。
第二题在C++ compiler下应该是不确定。但我见过的编译器都是从右到左进行参数计算
，但g++会给warning “test.cpp:28: warning: operation on ‘i’ may be
undefined”，
test.cpp line 28 is like "fun(i++, i++)"
Effective C++, 3rd edition, item 17, P76
C++ compilers are granted considerable latitude in determining the ord |
|
l********n
发帖数: 54 | 37 父类的函数(pure virtual)抛出一个异常，子类继承。有如下代码
class Foo
{
public:
void virtual func() throw (int, float) = 0;
};
class Bar : public Foo
{
public:
void virtual func() throw(short); // line 1: compile error "
// looser throw specifier"
void virtual func() throw(); // line 2: can compile
void virtual func() throw(float, int); // line 3: can compile
void virtual func() throw(float |
|
z***e
发帖数: 5393 | 38 之前我发过个帖子讨论科班和非科班的问题，这就是个例子。
不要被什么vtbl之类的束缚了，想一下如果你写一个compiler会怎么处理。
compiler读到这个a.Fun1()，首先去看A是否有matching function signature（注意，
compiler不会在B的function table里面找，因为a是type A)，找到了，然后看用什么
参数，生成大概如下代码：
push 10
call A::Func1
这里default parameter只是帮助source code少打几个字而已。 |
|
F**********r
发帖数: 237 | 39 try it in gcc compiler, it doesn't compile - how will compiler deduce the
return type is ? |
|
g**u
发帖数: 583 | 40
In fact, it is not exactely true.
In C++(Visual studio.net 2008), the first assignment, the compilation error
will complain for the first case, from b* --> a object. If you change it to
a *A=new b(), then the function called will be the one in the base because
the function in the base is non-virtual by default.
The second assignment, the compiler will error out again: can not convert
from b*-->a*.
BUT in Java, it is different, all these two will pass the compile time
test, and the output depe... 阅读全帖 |
|
y******5
发帖数: 43 | 41 Sorry for confusion. I know your code is for java, but I am not familiar
with it. So, I automatically using C++. Of course in C++ we need to change
the code from a to a*. Because size of b is no less than a, if we assign
address of an object of b to a-type pointer, b's own fields will be hidden.
That is why I said it is cut shorter.
If we assign address of an object of a to b-type pointer, there will always
be an error in C++.
In fact, it is not exactely true.
In C++(Visual studio.net 2008), th... 阅读全帖 |
|
h**********d
发帖数: 4313 | 42 有一个叫 method-local inner class 的东西
the inner class object cannot use the local variables of the method the inner class is in.
class MyOuter2 {
private String x = "Outer2";
void doStuff() {
String z = "local variable";
class MyInner {
public void seeOuter() {
System.out.println("Outer x is " + x);
System.out.println("Local variable z is " + z); // Won't Compile!
} // close inner class method
} // close inner class definition
} // close outer class meth... 阅读全帖 |
|
a******u
发帖数: 239 | 43 I think my answers to Question 7 and 8 are correct, but they said they are
not correct, why?
I already tested the code by Visual C++ (c compiler).
Thank you very much.
Line in file Contents
30 int * someIDs, theFirst, *r;
110 someIDs =GetSomeIDs(); /* defined below */
111 theFirst = someIDs [0];
112 r= ReorderIDs(someIDs);
113-150 /* we want to use ‘theFirst’ and ‘r’ here*/
499 /*-------- GetSomeIDs-----*/
500 int * GetSomeIDs()
501 {
502 int ids[8];
503-5... 阅读全帖 |
|
b*f
发帖数: 3154 | 44 mutable is to tell the compiler to be careful when optimizing this code as
the value of this variable can be changed outside of the current context.
For example,
int x = 0;
while(x == 0)
{
...
}
compiler will optimize away the comparison because within the context there
is no other way can change the value of x. But if you put mutable, compiler
will not do such optimization. |
|
g****e
发帖数: 141 | 45 工作地点在northern Virginia 有意者请发resume至g****[email protected]
本地、外地均可申请。
本人n年上站一次，所以请勿发站内信。谢谢！
The successful individual will be an energetic self-starter, with a
strong
desire to learn new things, and delve into unusual problems. They will
be a
member of the Software Services Team, whose focus is on build system
management and functionality, rather than the typical break-fix, user
administration of a typical Linux environment, although we do a little
of
that as well.
The selected individual will also ... 阅读全帖 |
|
b*****e
发帖数: 22 | 46 Probably due to the compiler you used. Empty base class optimization and
memory alignment all depend on compilers. Try print sizeof of all three
classes. Your compiler probably gives 4 bytes for all. I used GCC and got 4,
4,8 and therefore segfault at setValue. |
|
j********x
发帖数: 2330 | 47 写了一个小时，没考虑特别的corner case，O(num of intervals of "except")，用了
上面提到的binary search：
#include
#include
#include
#include
#include
struct compareable_interval {
int start;
int end;
int real_start;
int real_end;
compareable_interval(int lhs, int rhs) : start(lhs), end(rhs), real_
start(lhs), real_end(rhs) {
}
compareable_interval(int l, int r, int rl, int rr) : start(l), end(r),
real_start(rl), real_end(rr) {
}
b... 阅读全帖 |
|
j********x
发帖数: 2330 | 48 写了一个小时，没考虑特别的corner case，O(num of intervals of "except")，用了
上面提到的binary search：
#include
#include
#include
#include
#include
struct compareable_interval {
int start;
int end;
int real_start;
int real_end;
compareable_interval(int lhs, int rhs) : start(lhs), end(rhs), real_
start(lhs), real_end(rhs) {
}
compareable_interval(int l, int r, int rl, int rr) : start(l), end(r),
real_start(rl), real_end(rr) {
}
b... 阅读全帖 |
|
f****l
发帖数: 8042 | 49 难道CMU的CS MSc是类似CIS的方向？？但是他家的可选课程看上去是CS的典型课程设置
啊。求解惑，不想申请CIS方向。
Qualifying courses outside of CSD:
02-712 Computational Methods for Biological Modeling and Simulation
05-813 Human Factors
05-891 Designing Human-Centered Software
10-601 Machine Learning
10-702 Statistical Machine Learning
10-705 Intermediate Statistics
10-708 Probabilistic Graphical Models
10-725 Optimization
11-741 Information Retrieval
11-772 Analysis of Social Media
16-642 Manipulation, Mobility and Control
16-720 Computer Visi... 阅读全帖 |
|
h**********l
发帖数: 410 | 50 cornell cs就是这么奇葩，确实底层的软件课也不底层，所以我去面OS的职位被鄙视了
。我们compiler用java写project(practicum), 然后把他们自己发明的一个叫cubex的
鬼语言compile成C。。。没有assembly的事。。。
不过符合我转纯软件的目标。我觉得吧，上课没法建立软件体系，毕竟课的project规
模还是小，工作了找个软件工作就好，软件确实比硬件学得快多了。。。我也是上过c
，c++， data structure， algorithm，OS，compiler，embedded system， 硬件的工
作其实也都是python/perl的scritping，我觉得软件自学很容易啊，我在c#完全不懂的
情况下还修改了一下我research用的破软件的source code改了个bug呢。。。理论不会
，找葫芦画瓢正常人谁不会。。。 |
|
