I find compressor diff funtion cost large time, but it seem not use.

Can you clarify? The compression is used here: <a href="https://

Can you clarify? The compression is used here: <a href=

<a target="_blank" rel="noopener noreferrer nofollow" href="https://user-images.github

Is compressor enable ? about imgui-ws HOT 5 CLOSED

ggerganov commented on July 20, 2024

Is compressor enable ?

from imgui-ws.

Comments (5)

ggerganov commented on July 20, 2024

Can you clarify?

The compression is used here:
https://github.com/ggerganov/imgui-ws/blob/master/src/imgui-ws.cpp#L343-L351

from imgui-ws.

Eragon-Brisingr commented on July 20, 2024

Can you clarify?

The compression is used here: https://github.com/ggerganov/imgui-ws/blob/master/src/imgui-ws.cpp#L343-L351

But dataRead.drawListsDiff can not find any read reference. And i cancel calculate diff imgui-ws seem render right.

imgui-ws/src/compressor-xor-rle-per-draw-list-with-vtx-offset.cpp

Lines 94 to 161 in adece8c

 // calculate diff 

 m_drawListsDiff.clear(); 

 m_drawListsDiff.resize(nCmdLists); 

 for (uint32_t iList = 0; iList < nCmdLists; ++iList) { 

 auto & bufferCur = m_drawListsCur[iList]; 

 auto & bufferDiff = m_drawListsDiff[iList]; 

 bufferDiff.clear(); 

 int32_t type = 1; // Run-Length Encoding 

 if (iList >= m_drawListsPrev.size() || m_drawListsPrev[iList].size() != bufferCur.size()) { 

 type = 0; // Full update 

 } 

 std::copy((char *)(&type), (char *)(&type) + sizeof(type), std::back_inserter(bufferDiff)); 

 if (type == 0) { 

 std::copy(bufferCur.begin(), bufferCur.end(), std::back_inserter(bufferDiff)); 

 } else if (type == 1) { 

 auto & bufferPrev = m_drawListsPrev[iList]; 

 uint32_t a = 0; 

 uint32_t b = 0; 

 uint32_t c = 0; 

 uint32_t n = 0; 

 for (int i = 0; i < (int) bufferCur.size(); i += 4) { 

 std::memcpy((char *)(&a), bufferPrev.data() + i, sizeof(uint32_t)); 

 std::memcpy((char *)(&b), bufferCur.data() + i, sizeof(uint32_t)); 

 a = a ^ b; 

 if (a == c) { 

 ++n; 

 } else { 

 if (n > 0) { 

 std::copy((char *)(&n), (char *)(&n) + sizeof(uint32_t), std::back_inserter(bufferDiff)); 

 std::copy((char *)(&c), (char *)(&c) + sizeof(uint32_t), std::back_inserter(bufferDiff)); 

 } 

 n = 1; 

 c = a; 

 } 

 } 

 if (bufferCur.size() % 4 != 0) { 

 a = 0; 

 b = 0; 

 uint32_t i = (bufferCur.size()/4)*4; 

 uint32_t k = bufferCur.size() - i; 

 std::memcpy((char *)(&a), bufferPrev.data() + i, k); 

 std::memcpy((char *)(&b), bufferCur.data() + i, k); 

 a = a ^ b; 

 if (a == c) { 

 ++n; 

 } else { 

 std::copy((char *)(&n), (char *)(&n) + sizeof(uint32_t), std::back_inserter(bufferDiff)); 

 std::copy((char *)(&c), (char *)(&c) + sizeof(uint32_t), std::back_inserter(bufferDiff)); 

 n = 1; 

 c = a; 

 } 

 } 

 std::copy((char *)(&n), (char *)(&n) + sizeof(uint32_t), std::back_inserter(bufferDiff)); 

 std::copy((char *)(&c), (char *)(&c) + sizeof(uint32_t), std::back_inserter(bufferDiff)); 

 } 

 } 

 return true;

from imgui-ws.

Eragon-Brisingr commented on July 20, 2024

when draw many shape, Incppect::update cost too much time.

from imgui-ws.

ggerganov commented on July 20, 2024

Probably, the compression is not efficient for your scene.

If it is possible, you can try to record your session and we can analyze the data to see what is the problem. To record a session, check the record-sdl2 tool. Basically, you need to grab the Session class from here and use it to save your frames to a file. The file will be probably big, but if you could send it to me somehow, I can try to improve the compression algorithm to perform better in your case. You can also run the compressor-benchmark tool to get some information about the compression rate and performance for you scene.

from imgui-ws.

Eragon-Brisingr commented on July 20, 2024

Maybe is stl debug too slow, I try unreal engine development settings, the performance is can accept.

from imgui-ws.

Is compressor enable ? about imgui-ws HOT 5 CLOSED

Comments (5)

Related Issues (17)

Recommend Projects

React

Vue.js

Typescript

TensorFlow

Django

Laravel

D3

Recommend Topics

javascript

web

server

Machine learning

Visualization

Game

Recommend Org

Facebook

Microsoft

Google

Alibaba

D3

Tencent

	// calculate diff

	m_drawListsDiff.clear();
	m_drawListsDiff.resize(nCmdLists);

	for (uint32_t iList = 0; iList < nCmdLists; ++iList) {
	auto & bufferCur = m_drawListsCur[iList];
	auto & bufferDiff = m_drawListsDiff[iList];

	bufferDiff.clear();

	int32_t type = 1; // Run-Length Encoding
	if (iList >= m_drawListsPrev.size() \|\| m_drawListsPrev[iList].size() != bufferCur.size()) {
	type = 0; // Full update
	}

	std::copy((char )(&type), (char )(&type) + sizeof(type), std::back_inserter(bufferDiff));

	if (type == 0) {
	std::copy(bufferCur.begin(), bufferCur.end(), std::back_inserter(bufferDiff));
	} else if (type == 1) {
	auto & bufferPrev = m_drawListsPrev[iList];

	uint32_t a = 0;
	uint32_t b = 0;
	uint32_t c = 0;
	uint32_t n = 0;

	for (int i = 0; i < (int) bufferCur.size(); i += 4) {
	std::memcpy((char *)(&a), bufferPrev.data() + i, sizeof(uint32_t));
	std::memcpy((char *)(&b), bufferCur.data() + i, sizeof(uint32_t));
	a = a ^ b;
	if (a == c) {
	++n;
	} else {
	if (n > 0) {
	std::copy((char )(&n), (char )(&n) + sizeof(uint32_t), std::back_inserter(bufferDiff));
	std::copy((char )(&c), (char )(&c) + sizeof(uint32_t), std::back_inserter(bufferDiff));
	}
	n = 1;
	c = a;
	}
	}

	if (bufferCur.size() % 4 != 0) {
	a = 0;
	b = 0;
	uint32_t i = (bufferCur.size()/4)*4;
	uint32_t k = bufferCur.size() - i;
	std::memcpy((char *)(&a), bufferPrev.data() + i, k);
	std::memcpy((char *)(&b), bufferCur.data() + i, k);
	a = a ^ b;
	if (a == c) {
	++n;
	} else {
	std::copy((char )(&n), (char )(&n) + sizeof(uint32_t), std::back_inserter(bufferDiff));
	std::copy((char )(&c), (char )(&c) + sizeof(uint32_t), std::back_inserter(bufferDiff));
	n = 1;
	c = a;
	}
	}

	std::copy((char )(&n), (char )(&n) + sizeof(uint32_t), std::back_inserter(bufferDiff));
	std::copy((char )(&c), (char )(&c) + sizeof(uint32_t), std::back_inserter(bufferDiff));
	}
	}

	return true;