avatar
有没有sun unix的高手呀?# Unix - 噫吁兮,危乎高哉
z*y
1
刚搞了一台SUN enterprise 3500, 8个CPU, 5G RAM.
我们打算用它来搞 CFD simulation, 主要是大量的
浮点计算。
结果速度令人失望,仅仅和AMD 3000+的台式机差不多。
是不是我们的设置有问题。高手能否指点以下?
多谢!!
avatar
w*n
2
Definitely a bad choice if your intention is for floating points
computing.
Sun unix(sparc cpu) is good as a robust (file/web) server but bad for computing

【在 z***y 的大作中提到】
: 刚搞了一台SUN enterprise 3500, 8个CPU, 5G RAM.
: 我们打算用它来搞 CFD simulation, 主要是大量的
: 浮点计算。
: 结果速度令人失望,仅仅和AMD 3000+的台式机差不多。
: 是不是我们的设置有问题。高手能否指点以下?
: 多谢!!

avatar
b*e
3
what do you mean 差不多? did you run 8 simulations at the same time
for both the Sun and the AMD, and got the same results?

【在 z***y 的大作中提到】
: 刚搞了一台SUN enterprise 3500, 8个CPU, 5G RAM.
: 我们打算用它来搞 CFD simulation, 主要是大量的
: 浮点计算。
: 结果速度令人失望,仅仅和AMD 3000+的台式机差不多。
: 是不是我们的设置有问题。高手能否指点以下?
: 多谢!!

avatar
z*y
4
不是的,我们高的是pararell运算,就是一个simulation同时
在8个CPU上跑,和同样的这个运算在一个AMD 3000+上跑差不多。

【在 b*****e 的大作中提到】
: what do you mean 差不多? did you run 8 simulations at the same time
: for both the Sun and the AMD, and got the same results?

avatar
j*e
5
sounds interesting.
even only for memory, it should be faster.
there must be something wrong in your setup bah.

【在 z***y 的大作中提到】
: 刚搞了一台SUN enterprise 3500, 8个CPU, 5G RAM.
: 我们打算用它来搞 CFD simulation, 主要是大量的
: 浮点计算。
: 结果速度令人失望,仅仅和AMD 3000+的台式机差不多。
: 是不是我们的设置有问题。高手能否指点以下?
: 多谢!!

avatar
w*n
6
Guess this is an 8x400MHz model, meaning each cpu runs at 800Mhz. Is your AMD
3000 64-bit?

【在 z***y 的大作中提到】
: 刚搞了一台SUN enterprise 3500, 8个CPU, 5G RAM.
: 我们打算用它来搞 CFD simulation, 主要是大量的
: 浮点计算。
: 结果速度令人失望,仅仅和AMD 3000+的台式机差不多。
: 是不是我们的设置有问题。高手能否指点以下?
: 多谢!!

avatar
w*n
7
On the other hand, you should compare the performance of parallel version vs.
serial version, just make sure your parallelization actually works.
And try use sun's workshop compiler with optimization option.

【在 z***y 的大作中提到】
: 不是的,我们高的是pararell运算,就是一个simulation同时
: 在8个CPU上跑,和同样的这个运算在一个AMD 3000+上跑差不多。

avatar
z*y
8
Is Sun so weak in computing? I mean, this machine is an enterprise level
server, has 8*375MhzUltraSparcII CPUs, 5G RAM, even as weak as my AMD 2800+
notebook computation power?
I use commercial CFD code which has MPI function. It's designed to run
under UNIX environment, multiple CPUs. I believe it has been optimized for
parallel.
Is there anybody knows much about SUN MPI processing? That's why I post here

【在 j******e 的大作中提到】
: sounds interesting.
: even only for memory, it should be faster.
: there must be something wrong in your setup bah.

avatar
T*r
9
enterprise computing does not equal to scientific computing

【在 z***y 的大作中提到】
: Is Sun so weak in computing? I mean, this machine is an enterprise level
: server, has 8*375MhzUltraSparcII CPUs, 5G RAM, even as weak as my AMD 2800+
: notebook computation power?
: I use commercial CFD code which has MPI function. It's designed to run
: under UNIX environment, multiple CPUs. I believe it has been optimized for
: parallel.
: Is there anybody knows much about SUN MPI processing? That's why I post here

avatar
m*m
10
sun 3500 是几百年前的产品了, 早过时了. 而且ultra sparc 的 int, float
benchmarck 从来都是倒数第一, 第二... 换 alpha server 吧...

【在 z***y 的大作中提到】
: Is Sun so weak in computing? I mean, this machine is an enterprise level
: server, has 8*375MhzUltraSparcII CPUs, 5G RAM, even as weak as my AMD 2800+
: notebook computation power?
: I use commercial CFD code which has MPI function. It's designed to run
: under UNIX environment, multiple CPUs. I believe it has been optimized for
: parallel.
: Is there anybody knows much about SUN MPI processing? That's why I post here

avatar
n*t
11
actually, sun is a crap in High performance Scientific computing.

【在 z***y 的大作中提到】
: Is Sun so weak in computing? I mean, this machine is an enterprise level
: server, has 8*375MhzUltraSparcII CPUs, 5G RAM, even as weak as my AMD 2800+
: notebook computation power?
: I use commercial CFD code which has MPI function. It's designed to run
: under UNIX environment, multiple CPUs. I believe it has been optimized for
: parallel.
: Is there anybody knows much about SUN MPI processing? That's why I post here

相关阅读
logo
联系我们隐私协议©2024 redian.news
Redian新闻
Redian.news刊载任何文章,不代表同意其说法或描述,仅为提供更多信息,也不构成任何建议。文章信息的合法性及真实性由其作者负责,与Redian.news及其运营公司无关。欢迎投稿,如发现稿件侵权,或作者不愿在本网发表文章,请版权拥有者通知本网处理。