flat assembler
Message board for the users of flat assembler.

flat assembler > Main > What's the fastest?


What's the fastest/better?
Faster: floating point + FPU; Better: floating point + FPU
0%
 0%  [ 0 ]
Faster: floating point + FPU; Better: floating point + SSE4
0%
 0%  [ 0 ]
Faster: floating point + FPU; Better: fixed-point + GPR
0%
 0%  [ 0 ]
Faster: floating point + FPU; Better: fixed-point + SSE4
33%
 33%  [ 1 ]
Faster: floating point + SSE4; Better: floating point + FPU
0%
 0%  [ 0 ]
Faster: floating point + SSE4; Better: floating point + SSE4
0%
 0%  [ 0 ]
Faster: floating point + SSE4; Better: fixed-point + GPR
0%
 0%  [ 0 ]
Faster: floating point + SSE4; Better: fixed-point + SSE4
0%
 0%  [ 0 ]
Faster: fixed-point + GPR; Better: floating point + FPU
0%
 0%  [ 0 ]
Faster: fixed-point + GPR; Better: floating point + SSE4
0%
 0%  [ 0 ]
Faster: fixed-point + GPR; Better: fixed-point + GPR
0%
 0%  [ 0 ]
Faster: fixed-point + GPR; Better: fixed-point + SSE4
0%
 0%  [ 0 ]
Faster: fixed-point + SSE4; Better: floating point + FPU
33%
 33%  [ 1 ]
Faster: fixed-point + SSE4; Better: floating point + SSE4
33%
 33%  [ 1 ]
Faster: fixed-point + SSE4; Better: fixed-point + GPR
0%
 0%  [ 0 ]
Faster: fixed-point + SSE4; Better: fixed-point + SSE4
0%
 0%  [ 0 ]
Total Votes : 3

Author
Thread Post new topic Reply to topic
A$M



Joined: 29 Feb 2012
Posts: 94
Hello!
I have a little doubt.
What do you think faster:
- Use floating point numbers with FPU
- Use floating point numbers with SSE4
- Use fixed-point numbers with general registers
- Use fixed-point numbers with SSE4
(You don't need to be limited to these options)
Post 26 Sep 2013, 23:10
View user's profile Send private message Reply with quote
HaHaAnonymous



Joined: 02 Dec 2012
Posts: 1181
Location: Unknown
Stupid post removed.


Last edited by HaHaAnonymous on 28 Feb 2015, 19:49; edited 1 time in total
Post 27 Sep 2013, 00:52
View user's profile Send private message Reply with quote
A$M



Joined: 29 Feb 2012
Posts: 94
HaHaAnonymous wrote:
I doubt this poll will result in anything good. There are too many options.

Quote:
What do you think faster

It depends on your code. You can code pretty crap SSE4 "optimized" code and other people could write a code that is much more efficient than yours by using other means.

If you are going to include SSE4 code, please don't forget the processors that can't run it (e.g.: mine) and make it possible to run on these processors.

P.S. You are not obliged to do what I said.

Waiting for revolution or another expert... They might give you a decent answer.


Thanks for your answer. (my processor can run SSE4 Very Happy)
Post 27 Sep 2013, 01:39
View user's profile Send private message Reply with quote
revolution
When all else fails, read the source


Joined: 24 Aug 2004
Posts: 16782
Location: In your JS exploiting you and your system
The standard answer to "which is faster" is: we don't know, there are too many variables. When you look at only the instructions executing you get a very small view of what is happening in the system as a whole. Things like cache management might be more important, or not. Things like code layout and latency values might be more important, or not. Things like RAM timings and numbers of channels might be more impotant, or not. Things like which OS or GPU or even your cooling fan capability can affect the run speed of code, thses might be more important, or not. Like HaHaAnonymous mentions above, it depends upon your precise situation as to which code will be faster.
Post 27 Sep 2013, 04:36
View user's profile Send private message Visit poster's website Reply with quote
edfed



Joined: 20 Feb 2006
Posts: 4209
Location: 2018
the speed of your code is also the speed of the writing of the code. and as the writing speed is a very variable time, you can just try to determine the method where your writing is faster, and then, see what is the benefit to try anything else.

i think, when talking about Fixed and floating points, the fixed is very good and fast, but floating is also good and fast, it always depends on the dynamics of your signals.

if you should only cover a little count of digits, you should maybe use fixed point.
if you should manage values from nano to giga, maybe the floating point is better.

maybe none of them can suit your requirement, and then, you'll need a sort of bigDecimal type to define values with more than 19 decimal digits.
Post 27 Sep 2013, 20:45
View user's profile Send private message Visit poster's website Reply with quote
Display posts from previous:
Post new topic Reply to topic

Jump to:  


< Last Thread | Next Thread >
Forum Rules:
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum


Copyright © 1999-2019, Tomasz Grysztar.

Powered by rwasa.