* scattered in memory and thus don't interfere much with each
   * other. But Atomic objects residing in arrays will tend to be
   * placed adjacent to each other, and so will most often share
   * cache lines (with a huge negative performance impact) without
   * this precaution.
   *
   * In part because Cells are relatively large, we avoid creating
   * them until they are needed.  When there is no contention, all

}

// StringSym returns a symbol containing the string s.
// The symbol contains the string data, not a string header.
func StringSym(pos src.XPos, s string) (data *obj.LSym) {
	var symname string
	if len(s) > 100 {
		// Huge strings are hashed to avoid long names in object files.
		// Indulge in some paranoia by writing the length of s, too,
		// as protection against length extension attacks.
		// Same pattern is known to fileStringSym below.

		return out
	}

	return nil
}

// handleGateway resource adds a NetworkGateway for each combination of address and auto-passthrough listener
// discovering duplicates from the generated Service is not a huge concern as we de-duplicate in NetworkGateways
// which returns a set, although it's not totally efficient.
func (n *networkManager) handleGatewayResource(_ *v1beta1.Gateway, gw *v1beta1.Gateway, event model.Event) error {

			if cap, exists := allocatable[resource]; exists {
				container.Resources.Limits[resource] = cap.DeepCopy()
			}
		}
	}
}

// IsHugePageResourceName returns true if the resource name has the huge page
// resource prefix.
func IsHugePageResourceName(name v1.ResourceName) bool {
	return strings.HasPrefix(string(name), v1.ResourceHugePagesPrefix)

This is only an example to illustrate potential pitfalls.
In practice, larger libraries or frameworks can bring in a huge set of dependencies.
If those libraries fail to declare features separately and can only be consumed in a "all or nothing" fashion, excludes can be a valid method to reduce the library to the feature set actually required.

* **Machine Learning**: it normally requires lots of "matrix" and "vector" multiplications. Think of a huge spreadsheet with numbers and multiplying all of them together at the same time.
* **Deep Learning**: this is a sub-field of Machine Learning, so, the same applies. It's just that there is not a single spreadsheet of numbers to multiply, but a huge set of them, and in many cases, you use a special processor to build and / or use those models.

	// preprocessing step. The block based nature means that we can
	// preallocate fixed-size buffers and reuse them. However, the RLE
	// preprocessing would require allocating huge buffers to store the
	// maximum expansion. Thus we process blocks all at once, except for
	// the RLE which we decompress as required.
	n := 0
	for (bz2.repeats > 0 || bz2.preRLEUsed < len(bz2.preRLE)) && n < len(buf) {

				return <-con.errorChan
			}
		case <-con.stop:
			return nil
		default:
		}
		// If there wasn't already a request, poll for requests and pushes. Note: if we have a huge
		// amount of incoming requests, we may still send some pushes, as we do not `continue` above;
		// however, requests will be handled ~2x as much as pushes. This ensures a wave of requests

	//	case OpDiv8:
	//		if !isDivideByZero(argLt2.val.AuxInt8()) {
	//			res.val = newConst(argLt1.val.AuxInt8() / argLt2.val.AuxInt8())
	//		}
	//  ...
	// 	}
	//
	// However, this would create a huge switch for all opcodes that can be
	// evaluated during compile time. Moreover, some operations can be evaluated
	// only if its arguments satisfy additional conditions(e.g. divide by zero).

		name         string
		testKubelet  *TestKubelet
		initialNode  *v1.Node
		existingNode *v1.Node
		expectedNode *v1.Node
		needsUpdate  bool
	}{
		{
			name:        "no update needed when all huge page resources are similar",
			testKubelet: testKubelet,
			needsUpdate: false,
			initialNode: &v1.Node{
				Status: v1.NodeStatus{
					Capacity: v1.ResourceList{