nomic-ai/kompute
1
0
Fork 0
mirror of https://github.com/nomic-ai/kompute.git synced 2026-05-11 00:49:58 +00:00
Code Issues Packages Projects Releases Wiki Activity

Updated documentation examples

Browse Source
This commit is contained in:
Alejandro Saucedo
2021-02-21 12:12:16 +00:00
parent 29c50e5728
commit f474d21088
3 changed files with 22 additions and 271 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
README.md
View File

@@ -78,7 +78,7 @@ int main() {
// 3. Run operation with string shader synchronously
mgr.evalOpDefault<kp::OpAlgoBase>(
{ tensorInA, tensorInB, tensorOut },
std::vector<uint32_t>(shaderString.begin(), shaderString.end()));
kp::Shader::compile_source(shaderString));
// 4. Map results back from GPU memory to print the results
mgr.evalOpDefault<kp::OpTensorSyncLocal>({ tensorInA, tensorInB, tensorOut });

279
docs/overview/advanced-examples.rst
View File

@@ -45,7 +45,7 @@ Pass compute shader data in glsl/hlsl text or compiled SPIR-V format (or as path
auto tensorB = std::make_shared<kp::Tensor>(kp::Tensor({ 0., 0., 0. }));
// Create tensors data explicitly in GPU with an operation
mgr.evalOpDefault<kp::OpTensorCreate>({ tensorA, tensorB });
mgr.rebuild({ tensorA, tensorB });
// Define your shader as a string (using string literals for simplicity)
// (You can also pass the raw compiled bytes, or even path to file)
@@ -67,7 +67,7 @@ Pass compute shader data in glsl/hlsl text or compiled SPIR-V format (or as path
// Run Kompute operation on the parameters provided with dispatch layout
mgr.evalOpDefault<kp::OpAlgoBase>(
{ tensorA, tensorB },
std::vector<char>(shader.begin(), shader.end()));
kp::Shader::compile_source(shader));
// Sync the GPU memory back to the local tensor
mgr.evalOpDefault<kp::OpTensorSyncLocal>({ tensorA, tensorB });
@@ -105,7 +105,7 @@ Record commands in a single submit by using a Sequence to send in batch to GPU.
sq->begin();
// Record batch commands to send to GPU
sq->record<kp::OpMult<>>({ tensorLHS, tensorRHS, tensorOutput });
sq->record<kp::OpMult>({ tensorLHS, tensorRHS, tensorOutput });
sq->record<kp::OpTensorCopy>({tensorOutput, tensorLHS, tensorRHS});
// Stop recording
@@ -146,7 +146,7 @@ You can submit operations asynchronously with the async/await commands in the kp
auto tensor = std::make_shared<kp::Tensor>(kp::Tensor(std::vector<float>(10, 0.0)));
// Create tensors data explicitly in GPU with an operation
mgr.evalOpAsyncDefault<kp::OpTensorCreate>({ tensor });
mgr.rebuild(tensor)
// Define your shader as a string (using string literals for simplicity)
// (You can also pass the raw compiled bytes, or even path to file)
@@ -174,6 +174,8 @@ You can submit operations asynchronously with the async/await commands in the kp
}
)");
std::vector<uint32_t> spirv = kp::Shader::compile_source(shader);
// We can now await for the previous submitted command
// The first parameter can be the amount of time to wait
// The time provided is in nanoseconds
@@ -182,7 +184,7 @@ You can submit operations asynchronously with the async/await commands in the kp
// Run Async Kompute operation on the parameters provided
mgr.evalOpAsyncDefault<kp::OpAlgoBase>(
{ tensor },
std::vector<char>(shader.begin(), shader.end()));
spirv);
// Here we can do other work
@@ -234,7 +236,7 @@ Back to `examples list <#simple-examples>`_.
auto tensorB = std::make_shared<kp::Tensor>(kp::Tensor(std::vector<float>(10, 0.0)));
// We run the first step synchronously on the default sequence
mgr.evalOpDefault<kp::OpTensorCreate>({ tensorA, tensorB });
mgr.rebuild({ tensorA, tensorB });
// Define your shader as a string (using string literals for simplicity)
// (You can also pass the raw compiled bytes, or even path to file)
@@ -262,17 +264,19 @@ Back to `examples list <#simple-examples>`_.
}
)");
std::vector<uint32_t> spirv = kp::Shader::compile_source(shader);
// Run the first parallel operation in the `queueOne` sequence
mgr.evalOpAsync<kp::OpAlgoBase>(
{ tensorA },
"queueOne",
std::vector<char>(shader.begin(), shader.end()));
spirv);
// Run the second parallel operation in the `queueTwo` sequence
mgr.evalOpAsync<kp::OpAlgoBase>(
{ tensorB },
"queueTwo",
std::vector<char>(shader.begin(), shader.end()));
spirv);
// Here we can do other work
@@ -308,7 +312,7 @@ We also provide tools that allow you to `convert shaders into C++ headers <https
: OpAlgoBase(physicalDevice, device, commandBuffer, tensors, "")
{
// Perform your custom steps such as reading from a shader file
this->mShaderFilePath = "shaders/glsl/opmult.comp";
this->mShaderFilePath = "shaders/glsl/opmult.comp.spv";
}
}
@@ -323,7 +327,7 @@ We also provide tools that allow you to `convert shaders into C++ headers <https
auto tensorOut = std::make_shared<kp::Tensor>(kp::Tensor({ 0., 0., 0. }));
// Create tensors data explicitly in GPU with an operation
mgr.evalOpDefault<kp::OpTensorCreate>({ tensorLhs, tensorRhs, tensorOut });
mgr.rebuild({ tensorLhs, tensorRhs, tensorOut });
// Run Kompute operation on the parameters provided with dispatch layout
mgr.evalOpDefault<kp::OpMyCustom<3, 1, 1>>(
@@ -334,258 +338,3 @@ We also provide tools that allow you to `convert shaders into C++ headers <https
}
Logistic Regression Example
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Logistic regression is oftens seen as the hello world in machine learning so we will be using it for our examples. Back to `examples list <#simple-examples>`_.
.. image:: ../images/logistic-regression.jpg
:width: 300px
In summary, we have:
* Vector ``X`` with input data (with a pair of inputs ``Xi`` and ``Xj``\ )
* Output ``Y`` with expected predictions
With this we will:
* Optimize the function simplified as ``Y = WX + b``
* We'll want our program to learn the parameters ``W`` and ``b``
We will have to convert this into Kompute terminology.
First specifically around the inputs, we will be using the following:
* Two vertors for the variable `X`, vector `Xi` and `Xj`
* One vector `Y` for the true predictions
* A vector `W` containing the two input weight values to use for inference
* A vector `B` containing a single input parameter for `b`
.. code-block:: cpp
:linenos:
std::vector<float> wInVec = { 0.001, 0.001 };
std::vector<float> bInVec = { 0 };
std::shared_ptr<kp::Tensor> xI{ new kp::Tensor({ 0, 1, 1, 1, 1 })};
std::shared_ptr<kp::Tensor> xJ{ new kp::Tensor({ 0, 0, 0, 1, 1 })};
std::shared_ptr<kp::Tensor> y{ new kp::Tensor({ 0, 0, 0, 1, 1 })};
std::shared_ptr<kp::Tensor> wIn{
new kp::Tensor(wInVec, kp::Tensor::TensorTypes::eStaging)};
std::shared_ptr<kp::Tensor> bIn{
new kp::Tensor(bInVec, kp::Tensor::TensorTypes::eStaging)};
We will have the following output vectors:
* Two output vectors `Wi` and `Wj` to store all the deltas to perform gradient descent on W
* One output vector `Bout` to store all the deltas to perform gradient descent on B
.. code-block:: cpp
:linenos:
std::shared_ptr<kp::Tensor> wOutI{ new kp::Tensor({ 0, 0, 0, 0, 0 })};
std::shared_ptr<kp::Tensor> wOutJ{ new kp::Tensor({ 0, 0, 0, 0, 0 })};
std::shared_ptr<kp::Tensor> bOut{ new kp::Tensor({ 0, 0, 0, 0, 0 })};
For simplicity we will store all the tensors inside a params variable:
.. code-block:: cpp
:linenos:
std::vector<std::shared_ptr<kp::Tensor>> params =
{xI, xJ, y, wIn, wOutI, wOutJ, bIn, bOut};
Now that we have the inputs and outputs we will be able to use them in the processing. The workflow we will be using is the following:
1. Create a Sequence to record and submit GPU commands
2. Submit OpCreateTensor to create all the tensors
3. Record the OpAlgo with the Logistic Regression shader
4. Loop across number of iterations:
4-a. Submit algo operation on LR shader
4-b. Re-calculate weights from loss
5. Print output weights and bias
1. Create a sequence to record and submit GPU commands
.. code-block:: cpp
:linenos:
kp::Manager mgr;
if (std::shared_ptr<kp::Sequence> sq =
mgr.sequence("createTensors").lock())
{
// ...
Submit OpCreateTensor to create all the tensors
.. code-block:: cpp
:linenos:
{
// ... continuing from codeblock above
sq->begin();
sq->record<kp::OpCreateTensor>(params);
sq->end();
sq->eval();
Record the OpAlgo with the Logistic Regression shader
Once we re-record, all the instructions that were recorded previously are cleared.
Because of this we can record now the new commands which will consist of the following:
.. code-block:: cpp
:linenos:
{
// ... continuing from codeblock above
sq->begin();
sq->record<kp::OpTensorSyncDevice>({wIn, bIn});
sq->record<kp::OpAlgoBase>(
params,
false, // Whether to copy output from device
"test/shaders/glsl/test_logistic_regression.comp");
sq->record<kp::OpTensorSyncLocal>({wOutI, wOutJ, bOut});
sq->end();
Loop across number of iterations + 4-a. Submit algo operation on LR shader
.. code-block:: cpp
:linenos:
{
// ... continuing from codeblock above
uint32_t ITERATIONS = 100;
for (size_t i = 0; i < ITERATIONS; i++)
{
// Run evaluation which passes data through shader once
sq->eval();
4-b. Re-calculate weights from loss
Once the shader code is executed, we are able to use the outputs from the shader calculation.
In this case we want to basically add all the calculated weights and bias from the back-prop step.
.. code-block:: cpp
:linenos:
{
// ...
for (size_t i = 0; i < ITERATIONS; i++)
{
// ... continuing from codeblock above
// Run evaluation which passes data through shader once
sq->eval();
// Subtract the resulting weights and biases
for(size_t j = 0; j < bOut->size(); j++) {
wInVec[0] -= wOutI->data()[j];
wInVec[1] -= wOutJ->data()[j];
bInVec[0] -= bOut->data()[j];
}
// Set the data for the GPU to use in the next iteration
wIn->mapDataIntoHostMemory();
bIn->mapDataIntoHostMemory();
}
5. Print output weights and bias
.. code-block:: cpp
:linenos:
std::cout << "Weight i: " << wIn->data()[0] << std::endl;
std::cout << "Weight j: " << wIn->data()[1] << std::endl;
std::cout << "Bias: " << bIn->data()[0] << std::endl;
Logistic Regression Compute Shader
----------------------------------
Finally you can see the shader used for the logistic regression usecase below:
.. code-block:: cpp
:linenos:
#version 450
layout (constant_id = 0) const uint M = 0;
layout (local_size_x = 1) in;
layout(set = 0, binding = 0) buffer bxi { float xi[]; };
layout(set = 0, binding = 1) buffer bxj { float xj[]; };
layout(set = 0, binding = 2) buffer by { float y[]; };
layout(set = 0, binding = 3) buffer bwin { float win[]; };
layout(set = 0, binding = 4) buffer bwouti { float wouti[]; };
layout(set = 0, binding = 5) buffer bwoutj { float woutj[]; };
layout(set = 0, binding = 6) buffer bbin { float bin[]; };
layout(set = 0, binding = 7) buffer bbout { float bout[]; };
float learningRate = 0.1;
float m = float(M);
float sigmoid(float z) {
return 1.0 / (1.0 + exp(-z));
}
float inference(vec2 x, vec2 w, float b) {
float z = dot(w, x) + b;
float yHat = sigmoid(z);
return yHat;
}
float calculateLoss(float yHat, float y) {
return -(y * log(yHat) + (1.0 - y) * log(1.0 - yHat));
}
void main() {
uint idx = gl_GlobalInvocationID.x;
vec2 wCurr = vec2(win[0], win[1]);
float bCurr = bin[0];
vec2 xCurr = vec2(xi[idx], xj[idx]);
float yCurr = y[idx];
float yHat = inference(xCurr, wCurr, bCurr);
float loss = calculateLoss(yHat, yCurr);
float dZ = yHat - yCurr;
vec2 dW = (1. / m) * xCurr * dZ;
float dB = (1. / m) * dZ;
wouti[idx] = learningRate * dW.x;
woutj[idx] = learningRate * dW.y;
bout[idx] = learningRate * dB;
}

12
docs/overview/async-parallel.rst
View File

@@ -64,7 +64,7 @@ Sequences can be executed in synchronously or asynchronously without having to c
:linenos:
// Create tensors data explicitly in GPU with an operation
mgr.evalOpAsyncDefault<kp::OpTensorCreate>({ tensor });
mgr.rebuild({ tensor });
While this is running we can actually do other things like in this case create the shader we'll be using.
@@ -125,7 +125,7 @@ Similar to above we can run other commands such as the `OpAlgoBase` asynchronous
// Run Async Kompute operation on the parameters provided
mgr.evalOpAsyncDefault<kp::OpAlgoBase<>>(
{ tensor },
std::vector<char>(shader.begin(), shader.end()));
kp::Shader::compile_source(shader));
// Here we can do other work
@@ -226,7 +226,7 @@ Similar to the asyncrhonous usecase above, we can still run synchronous commands
:linenos:
// We run the first step synchronously on the default sequence
mgr.evalOpDefault<kp::OpTensorCreate>({ tensorA, tensorB });
mgr.rebuild({ tensorA, tensorB });
// Define your shader as a string (using string literals for simplicity)
// (You can also pass the raw compiled bytes, or even path to file)
@@ -259,17 +259,19 @@ Now we can actually trigger the parallel processing, running two OpAlgoBase Oper
.. code-block:: cpp
:linenos:
std::vector<uint32_t> spirv = kp::Shader::compile_source(shader);
// Run the first parallel operation in the `queueOne` sequence
mgr.evalOpAsync<kp::OpAlgoBase<>>(
{ tensorA },
"queueOne",
std::vector<char>(shader.begin(), shader.end()));
spirv);
// Run the second parallel operation in the `queueTwo` sequence
mgr.evalOpAsync<kp::OpAlgoBase<>>(
{ tensorB },
"queueTwo",
std::vector<char>(shader.begin(), shader.end()));
spirv);
Similar to the asynchronous example above, we are able to do other work whilst the tasks are executing.