Shareaza 2.5.2.0 advantage of SSE2

[trooty]

Hi, how exactly does Shareaza 2.5.2.0 take advantage of or benefit from usage of SSE2 which can be noticed during normal use?

[diztrancer]

No advantages, only disadvantage for some users

[ocexyz]

[trooty]

[diztrancer]

Shareaza don't have functions that can be optimized using SSE2. I doubt that even hashing routines can benefit from SSE2.

[trooty]

[trooty]

[cyko_01]

You can ignore reasonedmind* ... I mean diztrancer. He is just trolling, as usual.

SSE2 takes one repetitive operations and does them several times simultaneously. It is commonly used in software like photoshop and graphics software as well. In theory it should improve interface responsiveness as well as search result filtering, although we have no benchmarks to back this up.



Disclaimer: I am not a programmer, and these are only speculations based on my own research.
*Those who read torrentfreak will know who I am referring to.

[trooty]

[cyko_01]

[ocexyz]

SS2 and non SSE2 version are made only by different setting of compiler. This is also compiler related question if compiled code will be better or worse, as compiler is just a software and also is developed. SSE2 is very early (about 2000 year) implementation so should be in most of nowadays running machines, used are SSE7 and later, you can check details in public wikipedia. So SS2 is compatibility-friendly and most of users friendly I think. SSE idea is to do calculate the same operation on larger amount of data eg. diference like to calculate all pixells of screen with several instructions and calculate with one instruction whole group af data which makes together a screen, this allow also more effective hardware usage and benefits from it - this is not full and proper exactly explanation, shows only the general idea.

[trooty]

Thanks Cyko and Ocexyz.

[diztrancer]

Cyko: do you know assembler for IA32 ? did u write some projects directly in assmbler? did u write firmware ? did u write low-level windows software ? I do.

please, point me to that part of code that can directly benefit from SSE2 optimization

oce: secutiry rules code merely can be optimised, any second man vomits after looking at that part of code

[ocexyz]

[diztrancer]

[old_death]

[cyko_01]

[raspopov]

SSE1 and SSE2 widely used in Shareaza: all float point calculations, large memory blocks and strings operations. You can turn on assembler generation during project build and make sure yourself.

[diztrancer]

SSE2 code is not faster in most cases (that are applieble for shareaza). On many CPU's SSE2 can be even slower than MMX.

P.S. Anyone noticed performance gain in SSE2 version ?

[raspopov]

MMX is an obsolete technology not used together with SSE/SSE2. Read about float point intrinsic functions.