What is faster - getting raw attribute value or use collection?

Question

I need to load one attribute value(weight) for many product IDs.

Say list of the product ids is in array $productIDs.

In general case, which approach is faster,

1. Loading raw attribute value for each product, without loading products:

   foreach ($productIds as $productId) {
       $weight[$productId] = Mage::getResourceModel('catalog/product')->getAttributeRawValue($productId, 'weight', $store_id );
   }
   

2. Using collection:

   $collection = Mage::getModel('catalog/product')->getCollection()->addAttributeToFilter('id', array('in' => $productIds));
   foreach ($collection as $product) {
       $weight[$product->getId()] = $product->getWeight();
   }
   

Answer 1

Well you can test is easily, create the following test.php file in your root folder:

<?php
//some settings
error_reporting(E_ALL | E_STRICT);
define('MAGENTO_ROOT', getcwd());
$mageFilename = MAGENTO_ROOT . '/app/Mage.php';
require_once $mageFilename;
Mage::setIsDeveloperMode(true);
ini_set('display_errors', 1);
umask(0);
//instantiate the app model
Mage::app();

$productIds = array();
$storeId = 1;
$weight1 = array();
$weight2 = array();

Varien_Profiler::enable();
Varien_Profiler::start("getattributerawvalue");
foreach ($productIds as $productId) {
    $weight1[] = Mage::getResourceModel('catalog/product')->getAttributeRawValue($productId, 'weight', $storeId);
}
Varien_Profiler::stop("getattributerawvalue");
Varien_Profiler::start("collection");
$collection = Mage::getModel('catalog/product')->getCollection()->addAttributeToFilter('id', array('in' => $productIds));
foreach ($collection as $product) {
    $weight2[$product->getId()] = $product->getWeight();
}
Varien_Profiler::stop("collection");

$timers = Varien_Profiler::getTimers();

#$out = '<div style="position:fixed;bottom:5px;right:5px;opacity:.1;background:white" onmouseover="this.style.opacity=1" onmouseout="this.style.opacity=.1">';
#$out = '<div style="opacity:.1" onmouseover="this.style.opacity=1" onmouseout="this.style.opacity=.1">';
$out = "<a href=\"javascript:void(0)\" onclick=\"$('profiler_section').style.display=$('profiler_section').style.display==''?'none':''\">[profiler]</a>";
$out .= '<div id="profiler_section" style="background:white; display:block">';
$out .= '<pre>Memory usage: real: '.memory_get_usage(true).', emalloc: '.memory_get_usage().'</pre>';
$out .= '<table border="1" cellspacing="0" cellpadding="2" style="width:auto">';
$out .= '<tr><th>Code Profiler</th><th>Time</th><th>Cnt</th><th>Emalloc</th><th>RealMem</th></tr>';
foreach ($timers as $name=>$timer) {
    $sum = Varien_Profiler::fetch($name,'sum');
    $count = Varien_Profiler::fetch($name,'count');
    $realmem = Varien_Profiler::fetch($name,'realmem');
    $emalloc = Varien_Profiler::fetch($name,'emalloc');
    if ($sum<.0010 && $count<10 && $emalloc<10000) {
        continue;
    }
    $out .= '<tr>'
        .'<td align="left">'.$name.'</td>'
        .'<td>'.number_format($sum,4).'</td>'
        .'<td align="right">'.$count.'</td>'
        .'<td align="right">'.number_format($emalloc).'</td>'
        .'<td align="right">'.number_format($realmem).'</td>'
        .'</tr>'
    ;
}
$out .= '</table>';
$out .= '<pre>';
$out .= print_r(Varien_Profiler::getSqlProfiler(Mage::getSingleton('core/resource')->getConnection('core_write')), 1);
$out .= '</pre>';
$out .= '</div>';
echo $out;

Replace the following two variables with the ones you want to use:

$productIds = array();
$storeId = 1;

Now access the file via a browser and see the results.

I tested on my side with 50 products:

• Using getAttributeRawValue : Time : 0.0939 seconds / Emalloc: 1,610,440
• Using collection : Time: 0.7858 seconds / Emalloc: 9,195,992

So the first method is almost 10 times faster and almost uses 10 times less memory.

Answer 2

Bit different numbers here. "xhprof'ed" with with ~700 products ....

$storeId = 1;
$productIds = Mage::getModel('catalog/category')->load(41)
    ->getProductCollection()
    ->getAllIds();

Xhprof_Profiler::start();
# ...
var_dump($weight);
Xhprof_Profiler::stop();

---

1) getAttributeRawValue()

$rm = Mage::getResourceModel('catalog/product');
foreach ($productIds as $productId) {
    $weight[$productId] = $rm->getAttributeRawValue($productId, 'weight', $storeId);
}

• Total Incl. Wall Time (microsec): 5,572,944 microsecs
• Total Incl. CPU (microsecs): 5,425,659 microsecs
• Total Incl. MemUse (bytes): 1,302,512 bytes
• Total Incl. PeakMemUse (bytes): 1,272,360 bytes
• Number of Function Calls: 254,235

2) Mage::getModel('catalog/product')->getCollection()

$collection = Mage::getModel('catalog/product')->getCollection()
    ->addAttributeToSelect('weight')
    ->addAttributeToFilter('entity_id', $productIds);

foreach ($collection as $product) {
    $weight[$product->getId()] = $product->getWeight();
}

• Total Incl. Wall Time (microsec): 2,588,250 microsecs
• Total Incl. CPU (microsecs): 2,549,021 microsecs
• Total Incl. MemUse (bytes): 3,717,824 bytes
• Total Incl. PeakMemUse (bytes): 3,834,616 bytes
• Number of Function Calls: 124,602

3) Mage::getResourceModel('catalog/product_collection')

$collection = Mage::getResourceModel('catalog/product_collection')
    ->addAttributeToSelect('weight')
    ->addAttributeToFilter('entity_id', $productIds);

foreach ($collection as $product) {
    $weight[$product->getId()] = $product->getWeight();
}

• Total Incl. Wall Time (microsec): 2,603,804 microsecs
• Total Incl. CPU (microsecs): 2,584,472 microsecs
• Total Incl. MemUse (bytes): 3,706,600 bytes
• Total Incl. PeakMemUse (bytes): 3,821,336 bytes
• Number of Function Calls: 124,554

5) using addIdFilter() ^*

$collection = Mage::getResourceModel('catalog/product_collection')
    ->addAttributeToSelect('weight')
    ->addIdFilter($productIds);

foreach ($collection as $product) {
    $weight[$product->getId()] = $product->getWeight();
}

• Total Incl. Wall Time (microsec): 2,439,878 microsecs
• Total Incl. CPU (microsecs): 2,419,062 microsecs
• Total Incl. MemUse (bytes): 3,689,440 bytes
• Total Incl. PeakMemUse (bytes): 3,700,200 bytes
• Number of Function Calls: 118,409

6) using getColumnValues() ^*

$collection = Mage::getResourceModel('catalog/product_collection')
    ->addAttributeToSelect('weight')
    ->addIdFilter($productIds);

$weight = array_combine(
    $collection->getColumnValues('entity_id'),
    $collection->getColumnValues('weight')
);

• Total Incl. Wall Time (microsec): 2,140,222 microsecs
• Total Incl. CPU (microsecs): 2,119,635 microsecs
• Total Incl. MemUse (bytes): 3,180,560 bytes
• Total Incl. PeakMemUse (bytes): 3,260,200 bytes
• Number of Function Calls: 104,722

---

^{* nearly same result for 'entity_id', array('in' => $productIds), but it's less to type and better readable}

Answer 3

In my opinion Mage::getResourceModel('catalog/product')->getAttributeRawValue($productId, 'weight', $store_id );

Is faster because here you go to that product and fetch a single attribute value only.

---

In other case you load collection(though its one product) and again in another call you get the value and also here you load all unnecessary data.