o*********r
发帖数: 203 | 1 According to the IEEE Standard for Floating-Point Arithmetic (IEEE 754),
which is used by most of modern CPUs, .45 can't be exactly represented by:
floating_num = a0x(1/2)^(1) + a1x(1/2)^(2)+ ...
a0,a1,... = 0, or 1
So, .45 can't be stored exactly in binary format. Some numbers can be stored
exactly in binary:
0.5 = (1)x(1/2)
0.25 = (1)x(1/4)
0.125 = (1)x(1/8)
0.375 = (1)x(1/4) + (1)x(1/8) |
|
s*i
发帖数: 5025 | 2 var cluster = require('cluster'),
os = require('os');
if (cluster.isMaster) {
let files = ['1', '2', '3', '4', '5', '6', '7'];
let trySpawnNewWorker = function () {
if (files.length > 0) {
let params = {};
params.file = files.shift();
cluster.fork(params); //params becomes part of env
}
}
for (let i = 0; i < os.cpus().length; i++) {
trySpawnNewWorker();
}
// if one dies, spawn another to continue processing f... 阅读全帖 |
|
k****i
发帖数: 101 | 3 // tested, should work :)
(main = () => {
var cluster = require('cluster'),
fs = require('fs')
if (cluster.isMaster) {
(check = () => {
var mand = 'Usage: node thisScript inputDir outputDir ',
opt = '[optional: numberOfWorkerProcesses; default:
numberOfCpuCores]'
if (process.argv.length < 4) {
console.log(mand + opt)
process.exit(1)
}
})();
(schedule = () => {
var ind = process.argv[2],
files = fs... 阅读全帖 |
|
y**b
发帖数: 10166 | 4 perf不错，好像不能测试mpi程序。请看这个例子：
perf stat -p 48382 sleep 10
Performance counter stats for process id '48382':
4821.604141 task-clock (msec) # 0.963 CPUs utilized
[100.00%]
1,218 context-switches # 0.253 K/sec
[100.00%]
0 cpu-migrations # 0.000 K/sec
[100.00%]
0 page-faults # 0.000 K/sec
17,312,623,873 cycles ... 阅读全帖 |
|
m*****n
发帖数: 3575 | 5 https://www.quora.com/What-hardware-does-AlphaGo-run-on-Is-it-customized-
hardware-for-best-performance
AlphaGo uses a distributed system of 1920 CPUs and 280 GPUs. AlphaGo uses
Nvidia GPUs for the system. This is comparable to the TU Dresden's Taurus
supercomputer that uses 256 GPUs. In their research paper on Nature, they
talk about how AlphaGo can be run on a single computer, but would have only
a 23% win rate against the distributed system with multiple cores and GPUs.
So, you can view the s... 阅读全帖 |
|
E*********g
发帖数: 185 | 6 码农终于开发出了上吊的绳子
https://arxiv.org/abs/1709.05703
AI Programmer: Autonomously Creating Software Programs Using Genetic
Algorithms
Kory Becker, Justin Gottschlich
(Submitted on 17 Sep 2017)
In this paper, we present the first-of-its-kind machine learning (ML) system
, called AI Programmer, that can automatically generate full software
programs requiring only minimal human guidance. At its core, AI Programmer
uses genetic algorithms (GA) coupled with a tightly constrained programming
language that mi... 阅读全帖 |
|
Q**g
发帖数: 183 | 7 如果密码短可以一试。超过6位就放弃吧，呵呵。
Advanced RAR Password Recovery说：
Please take in mind, however, that RAR encryption is VERY strong, and even on
very fast CPUs it is possible to test not more than 2000 passwords per second.
So brute-force attack is effective for short passwords only (up to 5 or 6
characters), and if the password is longer and dictionary attack didn't
help -- you're out of luck. |
|
r***u
发帖数: 56 | 8 Is the cpu by amd? I feel that chrome has some locking issues with amd cpus. |
|
c*****t
发帖数: 1879 | 9 Do you actually believe any browser is 42x faster than the other?
The number itself is a strong indication of serious testing issue.
AND Apple is known to tweak performance numbers (like N years ago
apple cited a test to claim PowerPC was faster than Intel CPU,
but actually it was because the test specifically tweaked the library
used for PowerPC to remove lots of checks that were present in the
test for intel CPUs).
The article didn't say anything specific, although they did mention
that they u |
|
t******q
发帖数: 117 | 10 Reprogram your C code to multi thread,
and your unix should be aware of the multi CPU.
then, the os will auto assigned your code runing
on different CPUs.
Or read some parrallel processing material.
This is that area as I think. |
|
i*****d
发帖数: 5 | 11 how? I mean if some functions of C language can control
using multi-CPUs?
Thanks! |
|
p*u
发帖数: 2454 | 12 I am trying to find out how many CPUs are on one machine.
Does anybody know which command is to do this? Thank you for your help. |
|
h***o
发帖数: 539 | 13 the workstation has two cpus installed
I want the code to utilize both of them, instead of
only one by default. |
|
c******n
发帖数: 4965 | 14 【 以下文字转载自 Linux 讨论区 】
【 原文由 creation 所发表 】
is there some function
to call to return the # of CPUs on a machine?
I'm talking about a basic function that's
usable on all UNIX flavors,
thanks a lot
br /> |
|
T***B
发帖数: 137 | 15 I need to sort some very large files (30G for each) and want to estimate the
size of the temperary file created during the sorting. Is 3 or 4 times the
sorted file size large enough for the temperary file?
Since I want to know I could run how many sortings simultaneously if CPUs are
available. Thanks a lot. |
|
T***B
发帖数: 137 | 16 【 以下文字转载自 Linux 讨论区,原文如下 】
发信人: TFIIB (小明), 信区: Linux
标 题: logrotate
发信站: The unknown SPACE (Thu Jan 2 16:10:47 2003) WWW-POST
A process named logrotate created by the root has been running on my computer
for several days. It consumes ~90% of a CPU time ( 2 CPUs in total in this
computer ). I asked the system administrator to kill it. But it appeared again
soon.
I am using Red Hat 7.3.
Can anyone help? Thanks a lot. |
|
i******l
发帖数: 2 | 17 dependson which os u use, under solaris, use
uname -X to check # of CPUs and
/usr/sbin/prtconf | grep Mem to check memory |
|
T********r
发帖数: 6210 | 18 You know, for 24x7x365 up-time case, you should try IBM S390/AS400.
SUN may not be the same level as these systems. I forget the exact
number of the portion of OS/390 code that is dedicated to handle
fault. It is very impressive.
Also, I've heard that IBM's latest G6 (or G7?) CPU for S390 has two
CPUs to execute the same instructions simultaneously - in another
word, if one CPU happens to fail, another one takes over of it
immediately without any interruption. Any SUN's machine would 'waste'
it' |
|
i**o
发帖数: 13 | 19 想知道cpu的信息，在网上找到这个命令
/usr/platform/`uname -i`/sbin/prtdiag
得到的结果如下。那么是不是说cpu只有248mhz?
另外到底有几个cpu? 6?
谢谢。
System Configuration: Sun Microsystems sun4u 4-slot Sun Enterprise 3000
System clock frequency: 82 MHz
Memory size: 1280Mb
========================= CPUs =========================
Run Ecache CPU CPU
Brd CPU Module MHz MB Impl. Mask |
|
h**o
发帖数: 548 | 20 如果用: prstat -a >>prstat.log就能存入;
如果用: prstat -a |grep abc>>prstat.log就不能存入,是怎么回事那?(假设process
abc
存在). 应该怎样才能把结果存入prstat.log?
又, 如何显示process abc 所用的cpu那? 这台机器是multiple cpus, process abc 是
single thread 程序, 即只用一个cpu.我想知道它到底耗多少cpu, 但是用prstat 等总
是显示在
multiple cpu 中耗了多少. 请问哪个命令知道占所用cpu的多少? |
|
a**********0
发帖数: 422 | 21 cygwin在实际开发的过程中是做什么的呢？
经过编译的UNIX代码想在windows machine上运行就需要cygwin吗 据说没有编译过的用
cygwin也没有用
Cygwin is free software that provides a Unix-like environment and software
tool set to users of any modern version of MS-Windows for x86 CPUs 这是我看
的网上的定义 但是它做什么用呢？ |
|
a**********0
发帖数: 422 | 22 cygwin在实际开发的过程中是做什么的呢？
经过编译的UNIX代码想在windows machine上运行就需要cygwin吗 据说没有编译过的用
cygwin也没有用
Cygwin is free software that provides a Unix-like environment and software
tool set to users of any modern version of MS-Windows for x86 CPUs 这是我看
的网上的定义 但是它做什么用呢？ |
|
c***h
发帖数: 51 | 23 It depends which version of win2k you use. I believe win2k pro supports
2 CPU at most. |
|
B*******e
发帖数: 3882 | 24 control panel/system/hardware, "Device Manager", "Computer" 下面显示是什么？
双cpu 的机子显示的应该是 ACPI Multiprocessor PC.
Windows Task Manager, View/cpu history, 是“One Graph, All cpus" 还是”One g
raph per cpu" ?
performance |
|
s****0
发帖数: 89 | 25 I once tried to play a .3gp file with WMP 11. Just after that, My windows
media player 11 on Vista becomes mute when playing mp3 files. It seems the
player is playing the mp3 file, but no sound comes out. If you check the
task manager, you will see a process named Media Foundation Protected
Pipeline EXE occupy 50% of my Core 2 Duo CPUs. And if I use other player (
for example, foobar, winamp, or iTune) to play the same mp3 file, it works
without any problem.
System restore has been tried, unfort |
|
x******m
发帖数: 736 | 26 if you have lots of samples for whole genome seq, the computation is very
intensive. i remember some NG paper taking about this. Even for 100 cpu (
newest one), it still take about three days to finish alignment for about 50
genomes. But how many labs have the resources to run on clusters with 100
cpus? not so many! |
|
x******m
发帖数: 736 | 27 don't get it. 算三天跟算100天差别不大??? your boss can wait 3 days, not 100
days. it's based on 100 cpus. do you lab has 100 cpu cluster? |
|
h***o
发帖数: 539 | 28 I've turned the public workstations in our department as
a cluster...total 28 nodes, CPUs vary from 133MHz(super old)
to 450MHz...hoho
When my school's cluster is crowded..I use this one..
it works well....hiahia |
|
g****n
发帖数: 13 | 29 Hi
I am new to openMP. now I have some question about it.
I wrote a very simple program in C++.
#include
#include
main ()
{
int nthreads, tid;
int i;
omp_set_num_threads(2);
printf("Number of CPUS:%d\n",omp_get_num_procs());
/* Fork a team of threads giving them their own copies of variables */
#pragma omp parallel private(tid)
{
tid = omp_get_thread_num();
if(tid==0)
{
printf("tid=%d thread = %d\n", 0,tid);
printf("there are %d threads\n",omp_get_num_threads |
|
h***o
发帖数: 539 | 30 the workstation has two cpus installed
I want the code to utilize both of them instead
only one by default. |
|
t****n
发帖数: 39 | 31 Actually code optimization is machine dependent. However, for most popular
CPUs, the code written in a cache friendly manner could increase performance
significantly. For example, if you want calculate the following array
addition, C = A + B, and by switching off the compiler optimization,
code written as
do j = 1,jm
do i = 1,im
C(i,j) = A(i,j) + B(i,j)
end do
end do
would be way faster than
do i = 1,im
do j = 1,jm
C(i,j) = A(i,j) + B(i,j)
end do
end do
There are a lot of other t |
|
S***y
发帖数: 186 | 32
my
~~~~~~~~~
where? I don't think Athlon can beat P4C in large linear algebraic problems.
Actually, I never see a single CPU could beat P4 in such problems.
Of course, I'm referring the latest CPUs.
Alpha, Itanium, Power4, SGI, SUN ....
Maybe, Opetron can, but never have a chance to give it a try. :-)
no good things ba... for high-performance computing. |
|
f****r
发帖数: 27 | 33 Intel fortran compiler works great on Intel CPUs, but the performance
on AMDs are really bad(based on my experience). My solver compiled with ifc
on Athlon MP 2800+ has similar speed as on PIV 1.7G. But after I switched
to NAG F95, it doubled the speed. LOL |
|
Ag
发帖数: 481 | 34 even use double precision, you still can't get accurate result
for each step. As a result, gradually, the error will be bigger
and bigger. Different CPUs with various hardwares handle data in
different ways, you get different results, which is quiet normal. |
|
z***y
发帖数: 13818 | 35 But it's 8 Ultra Sparc II 375Mhz CPUs. I worry about that I may set up
something wrong. The CPU usage is always 60-85%. not 100%, not constant.
Maybe there's somethign wrong. |
|
s*****l
发帖数: 167 | 36 来自主题: Computation版 - 买机器建议 Can you have multiple CPUs? |
|
c*******o
发帖数: 1722 | 37 cannot get it run on multi CPUs using MPI?
想学习学习. |
|
D******n
发帖数: 2836 | 38 um...it is very cool when you just need to transfer small files...and it als
o depends how many cpus are on the pools. |
|
c****w
发帖数: 565 | 39 you're right, that's common phenomenon when using iterators under different
CPUs. However, my question is not platform or cpu-dependent. I just tested
under another machine and from intel fortran to compaq. The results are the
same, 52 for i, 20 for the number of function iterations.
Is there any chance to check IMSL's library? i would like to see the source
code to find out the reason. |
|
c*******h
发帖数: 1096 | 40 if you have a single core cpu, there is no way to do real parallelism.
(well, you can do some tricks on gpu and achieve somewhat parallelism,
but i don't think you can use matlab on gpus.)
parallel computing is for a multi-core cpu, or a cluster of cpus on either
one or many machines.
sing |
|
M*****t
发帖数: 120 | 41 Does it mean 60 CPUs are needed? |
|
f*****x
发帖数: 10 | 42 The job manager of matlab distributed computing toolbox queues the tasks and
submits them to the workers. The more CPUs in the cluster, the earlier the
job finishes.
FYI, matlab DCT will cost you (or your boss) a fortune. You'll probably want
to seek alternatives, for example R and Rmpi, if you prefer script
languages. |
|
c*******h
发帖数: 1096 | 43 man, some plugins do have been developed by third parties, but they are
either limited to a small range of operations (e.g., blas, fft), or do
not support double precision computations (though some pseudo-double-
precision
efforts have been made).
and when we are talking about multi-core, we mean cpus, not an (implicitly
multi-core) gpu. A gpu effectively has thousands of cores. You know the
difference between a multi-core cpu and a gpu. |
|
b****g
发帖数: 570 | 44 【 以下文字转载自 Faculty 讨论区 】
发信人: bluegg (Aggie), 信区: Faculty
标 题: 请推荐high performance computer
发信站: BBS 未名空间站 (Wed Jun 18 11:21:30 2014, 美东)
我的budget是主机不超过$7000，希望有尽量多的cpus, cores.
主要应用月CFD (Fluent)计算。
谢谢。 |
|
b******b
发帖数: 21 | 45 An ASIC flow is basically a set of EDA tools (softwares). Given a right ASIC
flow I can design a CPU in six months. But no one will buy the CPU because
it's going to 10x more expensive and 8x slower then a pentium 4. You won't
gain a lot of experience by using those tools. BTW, it took 1000+ engineers 3
years to design pentium 4.
之大者) 的大作中提到: 】
CPUs,
300M
前两天给中科院的一个同学打电话，发现北京中科院的几个所都在做芯片，cpu，dsp处理 |
|
L****a
发帖数: 572 | 46 see Amdahl's law :
http://en.wikipedia.org/wiki/Amdahl%27s_law
it's a law saying that when you stack up CPUs together, you can
only get sub-linear speed-up of the real computation power.
This is because of a few of bottlenecks, e.g. communication bottlenecks,
resource access bottlenecks ... |
|
f********e
发帖数: 48 | 47 本人MSEE with about ten years industry experience.
一个月前接受了A公司的offer，再过2个月去上班。 今天有来个B公司onsite面试邀请，真不知该不该去。
Company A: Analog and Pwr IC company, I will design DC DC converter, Hot swap, Pwr management related standard product.
Company B: Analog and Pwr IC company, I will design DC DC converter for
desktop/server. The group is focused on computing. The product are mainly DC DC converter for Intel and AMD CPUs in desktop. The design center has probably 30 people with about 10 designers. This desig |
|
g****t
发帖数: 31659 | 48 一般来说，例如你的工资是x,假如你自信A给你的平台，能让你每年给公司赚到5*x的利
润。那你的工作也会很稳定。B规模大，但如果闲人太多，也不一定稳定。
请，真不知该不该去。
swap, Pwr management related standard product. I will report to design
center director. I will be the 2nd employee of this design center.(First
one is the director).
DC DC converter for Intel and AMD CPUs in desktop. The design center has
probably 30 people with about 10 designers. This design center has a long
history but desktop biz is s
the industry. Company A's compensation is decent base, 15% bonus a |
|
g******z
发帖数: 495 | 49 ten years experience...
我要有你的1/5，工作也会好找得多...
请，真不知该不该去。
swap, Pwr management related standard product. I will report to design
center director. I will be the 2nd employee of this design center.(First
one is the director).
DC DC converter for Intel and AMD CPUs in desktop. The design center has
probably 30 people with about 10 designers. This design center has a long
history but desktop biz is s
the industry. Company A's compensation is decent base, 15% bonus and 12k
stock option. It will b |
|
c*******r
发帖数: 275 | 50 你都工作10年了，就别在这里问了。这里大部分都是小孩子，给的意见都想当然。
btw：暴露信息太多也不好，这圈子很小。
请，真不知该不该去。
swap, Pwr management related standard product. I will report to design
center director. I will be the 2nd employee of this design center.(First
one is the director).
DC DC converter for Intel and AMD CPUs in desktop. The design center has
probably 30 people with about 10 designers. This design center has a long
history but desktop biz is s
the industry. Company A's compensation is decent base, 15% bonus and 12k
stock opti |
|
