How to modify __m128i data?

Question

0.00/5 (No votes)

See more:

Hi all.

I have a question on SSE operations.

I have 'm128i_i8 of __m128i' data that have as following.

C++

__m128i Mydata;
Mydata.m128i_i8[0] : 90
Mydata.m128i_i8[1] : 0
Mydata.m128i_i8[2] : 1
Mydata.m128i_i8[3] : 0
Mydata.m128i_i8[4] : 114
Mydata.m128i_i8[5] : 1
Mydata.m128i_i8[6] : 34
Mydata.m128i_i8[7] : 1
Mydata.m128i_i8[8] : 85
Mydata.m128i_i8[9] : 2
Mydata.m128i_i8[10] : 68
Mydata.m128i_i8[11] : 2
Mydata.m128i_i8[12] : 103
Mydata.m128i_i8[13] : 2
Mydata.m128i_i8[14] : 119
Mydata.m128i_i8[15] : 1

※ Question 1.
Can I convert data following as ?

C++

__m128i result;
result.m128i_i8[0] : 90
result.m128i_i8[1] : 1
result.m128i_i8[2] : 114
result.m128i_i8[3] : 34
result.m128i_i8[4] : 85
result.m128i_i8[5] : 68
result.m128i_i8[6] : 103
result.m128i_i8[7] : 119
result.m128i_i8[8] : 0
result.m128i_i8[9] : 0
result.m128i_i8[10] : 0
result.m128i_i8[11] : 0
result.m128i_i8[12] : 0
result.m128i_i8[13] : 0
result.m128i_i8[14] : 0
result.m128i_i8[15] : 0

※ Question 2.
If I have as following data.

C++

Mydata.m128i_i16[0] : 90
Mydata.m128i_i16[1] : 1
Mydata.m128i_i16[2] : 114
Mydata.m128i_i16[3] : 222
Mydata.m128i_i16[4] : 341
Mydata.m128i_i16[5] : 444
Mydata.m128i_i16[6] : 359
Mydata.m128i_i16[7] : 119

Can I convert as following?
(Threshold is 255)

C++

__m128i result;
result.m128i_i8[0] : 90
result.m128i_i8[1] : 1
result.m128i_i8[2] : 114
result.m128i_i8[3] : 222
result.m128i_i8[4] : 255
result.m128i_i8[5] : 255
result.m128i_i8[6] : 255
result.m128i_i8[7] : 119
result.m128i_i8[8] : 0
result.m128i_i8[9] : 0
result.m128i_i8[10] : 0
result.m128i_i8[11] : 0
result.m128i_i8[12] : 0
result.m128i_i8[13] : 0
result.m128i_i8[14] : 0
result.m128i_i8[15] : 0

My code:

C++

__m128i result = _mm_setzero_si128();
result = _mm_packs_epi16(Mydata,_mm_setzero_si128());

Please advise me. Thank you. :)

What I have tried:

I tried to use '_mm_packs_epi16' operation but result is strange.

Posted 10-Apr-16 21:42pm

kgg124

Updated 11-Apr-16 13:58pm

v6

Add a Solution

Comments

nv3 11-Apr-16 6:47am

You have to show your code for someone being able to help you. Yes, _mm_packs_epi16 does pack your 16-bit integers into 8-bit integers, but with saturation. That means that for example a value of 346 (represented as bytes 90 1) is converted to 127, which is the highest signed 8-bit integer. Is that what you mean by "result is strange"?

kgg124 11-Apr-16 7:57am

Thank you for your reply.
I have same result. If value is larger than 127, it converted to 127.

1 solution

Add a Solution

Add your solution here

Treat my content as plain text, not as HTML

Preview 0

…

Existing Members

Sign in to your account

...or Join us

Download, Vote, Comment, Publish.

Your Email
Password
Forgot your password?

Your Email
This email is in use. Do you need your password?
Optional Password

I have read and agree to the Terms of Service and Privacy Policy
Please subscribe me to the CodeProject newsletters

When answering a question please:

Read the question carefully.
Understand that English isn't everyone's first language so be lenient of bad spelling and grammar.
If a question is poorly phrased then either ask for clarification, ignore it, or edit the question and fix the problem. Insults are not welcome.
Don't tell someone to read the manual. Chances are they have and don't get it. Provide an answer or move on to the next question.

Let's work to help developers, not make them feel stupid.

This content, along with any associated source code and files, is licensed under The Code Project Open License (CPOL)

nv3 · Answer 1 · 2016-04-11T06:35:00

To my knowledge there is no SSE instruction that does the packing without saturation. But I am admittedly no expert at SSE2.

See here for the SSE2 reference on the Microsoft website:

Integer Intrinsics Using Streaming SIMD Extensions 2[^]

Of course you can use conventional instruction to accomplish what you want. Just treat the memory representation of your 128-bit item as byte array and do the packing in a little loop. That's just two lines of code and if you don't do that extensively it will be not that much slower than an SSE2 instruction.