Questions for smart guys

You guys seem smart. Here's some questions:

0. Suppose you had an application that ran on a dual pentium server.
Then you migrated it to a quad Xeon server. You notice that the
application exhibits no increase in speed for a length batch process.

What do you conclude?


1. This application uses java. On the quad xeon system, for a field in
a database that is type float, for a certain percentage of imported
records shows an odd point displacement -- so 275.5 in a csv file
becomes 0.002755 in the database. But only for 1% of the records. On
the dual pentium system, there is no such problem.

What could be the reason?

John Bailo wrote:


That JBailo is too stupid to really move the app to a different server. It
still runs on the old one


JBailo is an incompetent Java programmer
--
This problem was sponsored by Microsoft

John Bailo wrote:


the app only has at most two threads running... and so cannot take advantage
of the extra CPUs...

--
COMPUTER POWER TO THE PEOPLE! DOWN WITH CYBERCRUD!

John Bailo wrote:


There are a number of conditions:

(1) Is the application threaded? If not, you will not see much performance
difference unless it has multiple threads or multiple processes which can
each be executed on one of the processors.

(2) How well does the JVM support threading? If you have a crappy JVM, java
threads won't actually scale very well.

(3) Do you use a lot of shared resources? You might be in a "false
parallelism" mode where, on a practical level, it looks threaded, but in
fact, the threads are serialized through a shared resource like a single
database connection or file.

John Bailo wrote:
That your a bad programmer, any good programmer wouldn't need to ask for
help on a simple problem. Yes, I do know the answer to the problem,
but if you know Java, you would know the answer also. It might not be
easy to implement the fix, but it could be done.
I have done Java programming in the past if you really need or want help
their are much better places to find help, then a Linux Newsgroup.

Most like a logical bug you have in the code. It shouldn't be hard to
make sure the data is imported correctly but you skills are so poor you
wouldn't be able to fix the code.

On Fri, 30 Jan 2004 23:05:11 +0000, Quantum Leaper wrote:


How come the quiet ones like mlw tell me a very intelligent and probably
accurate answer and the boastful ones such as /Quantum Leaper/ parade
around as if there is some big mystery.



My question was posted to a java newsgroup as well.

I am wondering if the answer has to do with the Windos threading model (
did I mention it's a windos application ? ) and that perhaps the 2.6
kernel threading would make a difference.

John Bailo wrote:

It not a big mystery, just common sense, which you lack.... I have had
problems like yours in the past, and I have never needed outside help.

Which doesn't exist on my server or easynews, why not try comp.lang.java.*
newsgroups, or some of the java message boards around the web.


I have no idea what OS your running the Java program on, so I couldn't
answer that but why not try Sun's JVM with Windows, I would say they know
how to program a JVM. If it Java just run in on a Linux machine its not
hard...

Did you MS Java or real Java? If it real Java it will run on Linux just
fine, since your running it in a VM.

Quantum Leaper wrote:


LOL...the WHOLE ****ING POINT OF A NEWSGROUP IS TO DISCUSS THESE
ISSUES!!!! If you don't want to help others with your knowledge why
the **** are you here??? Jesus Christ Almight !!!


Wait. You're saying that I shouldn't post to a particular java
newsgroup because your newsserver doesn't have it and therefore it means
my post wasn't posted to a java group?

WTF are you ??????



its jre on windows 2k.


real java 1.3


--
*W '04* <:> Open Source <:> Open Competition

John Bailo wrote:
Very true, you missing the point, you should ask for questions in the
CORRECT Newsgroup, asking about Java problem in a Linux newsgroup is NOT
the correct newsgroup.


No, but if you want help, you might try a popular newsgroup. Easynews has
just about EVERY Newsgroup available, and it didn't have your Java
newsgroup, which means in NOT widely distributed or local to your server.
I guess you really didn't want help, most don't ask question in an Advocacy
or physics group unless they want to make trouble.


Human, unlike you, who is a Troll....
It always helps to post enough information, that way you get a quick
answer, which you didn't really do.



Good, atleast you got one thing correct....

IBM has ascertained that it can achieve 210 GIGAHERTZ absolute MAX.*
1000 Gigahertz seems just beyond their grasp.*
*
Meanwhile, our prototype NAST-CPU runs at a solid 2,000,000 Gigahertz.
and uses a 10,000,000 Gigahertz NAST-Oscillator as a clock, implying we
could achieve 4,000,000 Gigahertz in a single NAST-CPU if we tried (and
we aren't even trying to optimize or implement multiple execution units
or anything).* It implements the following functions:
*
(a) add* (varying word length up to 256 bytes per number, varying
formats: real, integer, double or scientific)
(s) subtract (same as add but subtract)
(m) multiply (ibid)
(d) divide (ibid)
(c) string compare
(p) copy (mem to mem), or move, or combine under mask
(u) various substring and unstring functions
(v) convert (num to string, and back)
(i) if
(t) then
(e) else
(w) do while
(h) switch statement
(r) start program
(z) end program
(b) begin block
(y) end block
(o) output to device
(n) input from device
(x) control device or obtain status from device
(b) branch conditional or unconditional
(l) call subroutine
(f) return from subroutine
(g) call protected subroutine
(ls) load stack frame
(ps) pop stack frame
(gs) get stack frame address
(is) increment from stack frame address
(ds) decrement stack frame address
(cs) copy stack frame
(rs) replace stack frame
(fs) reference stack frame value
(ts) test stack frame value
(bool) perform boolean function
(rr) read results registers
(sr) set results registers
(br) branch on results registers
(am) allocate memory block
(dm) deallocate memory block
(cm) consolidate memory block
(*k) various mask functions
(err) various error and error response state functions
(rst) reset and init
and a few others for esoteric machine control and setting up masks and
responding to error states and so on.


As the device uses other NAST's as memory, it has no need of CPU
Registers or Cache, it just processes everything memory to memory and
device to CPU and back.*

Very simple, elegant*design with*addresses as wide as we need.* I
believe we can get several thousand such CPUs going on a single large
100 terabyte memory plane of NAST memory, all running simultaneously, if
we try.* Which multiplies the 2Million gigahertz throughput by a
hundred orders of magnitude.* Much much faster than the Human Brain,
btw.*

In fact, one might be able to build one so bit and*fast, it could
outthink all of humanity at once.* In a box the size of a Toaster
Oven.
*
Since JP Morgan/Chase/Rockefeller/GE (IBM) doesn't own or control it,
they will neither use NAST nor allow me the fiscal resources to develop
it.* If they can't have it, no one will, with an exception to be known
soon.

߃--¹¹