473,574 Members | 2,248 Online
Bytes | Software Development & Data Engineering Community
+ Post

Home Posts Topics Members FAQ

set, insert with hint

Some time ago I posted here about inserting into a set with a hint:

http://groups-beta.google.com/group/...8b8d0eadb38dbf

I quoted the SGI STL docs describing a.insert(p, t), where p is the hint
iterator and t is the inserted object:

"Insert with hint is logarithmic in general, but it is amortized
constant time if t is inserted immediately before p."

Several knowledgeable people asserted that this is contrary to the
standard which says that t should be inserted immediately after p.

I wrote a little program to test this out (see below). I then obtained
the following results, which suggest that the SGI docs are in fact
correct and that the insert should occur immediately before the hint.

Using gcc 3.2.2 with -O3:

start - no hint
finish: 14 seconds
start - insert after hint
finish: 14 seconds
start - insert before hint
finish: 3 seconds

Using Sun CC 7.1 ith -O5:

start - no hint
finish: 54 seconds
start - insert after hint
finish: 43 seconds
start - insert before hint
finish: 17 seconds

Was I just confused before and the standard does say that the hint
should point after the location of the insert? How else can these
results be explained?

Thanks,
Mark
TEST CODE BELOW:

#include <set>
#include <iostream>
#include <ctime>

int main() {

const int MaxVal = 30000000;
time_t startTime, endTime;
std::set<int> is;
std::set<int>:: iterator isIter;

// no hint

std::cout << "start - no hint" << std::endl;
time(&startTime );

for (int i = 0; i < MaxVal; ++i)
is.insert(i);

time(&endTime);
std::cout << "finish: " << difftime(endTim e,startTime) << " seconds"
<< std::endl;

// insert after hint

is.clear();
std::cout << "start - insert after hint" << std::endl;
time(&startTime );

isIter = (is.insert(0)). first;
for (int i = 1; i < MaxVal; ++i)
isIter = is.insert(isIte r,i);

time(&endTime);
std::cout << "finish: " << difftime(endTim e,startTime) << " seconds"
<< std::endl;

// insert before hint

is.clear();
std::cout << "start - insert before hint" << std::endl;
time(&startTime );

isIter = (is.insert(MaxV al)).first;
for (int i = MaxVal-1; i >= 0; --i)
isIter = is.insert(isIte r,i);

time(&endTime);
std::cout << "finish: " << difftime(endTim e,startTime) << " seconds"
<< std::endl;

}
Jul 23 '05 #1
6 6995
Mark P wrote:

I quoted the SGI STL docs describing a.insert(p, t), where p is the hint
iterator and t is the inserted object:

"Insert with hint is logarithmic in general, but it is amortized
constant time if t is inserted immediately before p."

Several knowledgeable people asserted that this is contrary to the
standard which says that t should be inserted immediately after p.

I wrote a little program to test this out (see below). I then obtained
the following results, which suggest that the SGI docs are in fact
correct and that the insert should occur immediately before the hint.

You would do better to consult the language definition. The 2003
standard gives the complexity of insert with hint as: "logarithmi c in
general, but amortized cosntant if t is inserted right after p."
How else can these results be explained?


Confusion.

--

Pete Becker
Dinkumware, Ltd. (http://www.dinkumware.com)
Jul 23 '05 #2
Pete Becker wrote:
Mark P wrote:

I quoted the SGI STL docs describing a.insert(p, t), where p is the
hint iterator and t is the inserted object:

"Insert with hint is logarithmic in general, but it is amortized
constant time if t is inserted immediately before p."

Several knowledgeable people asserted that this is contrary to the
standard which says that t should be inserted immediately after p.

I wrote a little program to test this out (see below). I then
obtained the following results, which suggest that the SGI docs are in
fact correct and that the insert should occur immediately before the
hint.


You would do better to consult the language definition. The 2003
standard gives the complexity of insert with hint as: "logarithmi c in
general, but amortized cosntant if t is inserted right after p."
How else can these results be explained?

Confusion.


I'll concede that I may be confused, but why does my test code run
fastest when t is inserted right before p?
Jul 23 '05 #3
Mark P wrote:
How else can these results be explained?


Confusion.


I'll concede that I may be confused, but why does my test code run
fastest when t is inserted right before p?


I meant confusion on the part of implementors. It could be that the
libraries you're using don't conform to the standard.

--

Pete Becker
Dinkumware, Ltd. (http://www.dinkumware.com)
Jul 23 '05 #4
Pete Becker wrote:
Mark P wrote:

How else can these results be explained?


Confusion.


I'll concede that I may be confused, but why does my test code run
fastest when t is inserted right before p?

I meant confusion on the part of implementors. It could be that the
libraries you're using don't conform to the standard.


Ah, I see. I find this surprising (which is not to say doubtful)
because my Sun compiler uses (I think) a RogueWave STL implementation
and my gcc uses an SGI STL implementation. Then again, they may derive
from common ancestry.

If anyone has a chance to try my test code on other compilers, I'd be
interested to see the results.
Jul 23 '05 #5
Mark P wrote:
Some time ago I posted here about inserting into a set with a hint:

http://groups-beta.google.com/group/...8b8d0eadb38dbf
I quoted the SGI STL docs describing a.insert(p, t), where p is the hint
iterator and t is the inserted object:

"Insert with hint is logarithmic in general, but it is amortized
constant time if t is inserted immediately before p."

Several knowledgeable people asserted that this is contrary to the
standard which says that t should be inserted immediately after p.

I wrote a little program to test this out (see below). I then obtained
the following results, which suggest that the SGI docs are in fact
correct and that the insert should occur immediately before the hint.

Using gcc 3.2.2 with -O3:

start - no hint
finish: 14 seconds
start - insert after hint
finish: 14 seconds
start - insert before hint
finish: 3 seconds

Using Sun CC 7.1 ith -O5:

start - no hint
finish: 54 seconds
start - insert after hint
finish: 43 seconds
start - insert before hint
finish: 17 seconds

Was I just confused before and the standard does say that the hint
should point after the location of the insert? How else can these
results be explained?

Thanks,
Mark
TEST CODE BELOW:

#include <set>
#include <iostream>
#include <ctime>

int main() {

const int MaxVal = 30000000;
time_t startTime, endTime;
std::set<int> is;
std::set<int>:: iterator isIter;

// no hint

std::cout << "start - no hint" << std::endl;
time(&startTime );

for (int i = 0; i < MaxVal; ++i)
is.insert(i);

time(&endTime);
std::cout << "finish: " << difftime(endTim e,startTime) << " seconds"
<< std::endl;

// insert after hint

is.clear();
std::cout << "start - insert after hint" << std::endl;
time(&startTime );

isIter = (is.insert(0)). first;
for (int i = 1; i < MaxVal; ++i)
isIter = is.insert(isIte r,i);

time(&endTime);
std::cout << "finish: " << difftime(endTim e,startTime) << " seconds"
<< std::endl;

// insert before hint

is.clear();
std::cout << "start - insert before hint" << std::endl;
time(&startTime );

isIter = (is.insert(MaxV al)).first;
for (int i = MaxVal-1; i >= 0; --i)
isIter = is.insert(isIte r,i);

time(&endTime);
std::cout << "finish: " << difftime(endTim e,startTime) << " seconds"
<< std::endl;

}


Many 'set' implementations are backed by a red-black tree.
The performance of these trees degrades quickly when inserting
pre-sorted data (as you are doing here) due to all of the
tree re-balancing that must occur.

Also, many implementations have special handling for the
'hint' boundary conditions ('end()' and 'begin()').
In your last example (where you inserted data in
high-to-low order) the iterator returned by the 'insert()'
call is actually 'begin()'. Since 'insert()' has special
handling for this 'hint' the performance of the NEXT 'insert()'
call is high.

If you pass an iterator that is neither 'begin()' nor 'end()',
when 'insert()' increments that iterator and it does not
point to a valid item, insert() starts its search with
'begin()' (i.e. your 'hint' iterator is ignored).

If you are inserting data in low-to-high order this
approach will speed things up greatly:

isIter = (is.insert(0)). first;
for (int i = 1; i < MaxVal; ++i)
is.insert(is.en d(), i);

Regards,
Larry

--
Anti-spam address, change each 'X' to '.' to reply directly.
Jul 23 '05 #6
Larry I Smith wrote:
Many 'set' implementations are backed by a red-black tree.
The performance of these trees degrades quickly when inserting
pre-sorted data (as you are doing here) due to all of the
tree re-balancing that must occur.
An excellent point which I overlooked (based on the frequent use of "Rb"
in the internal names of the STL souce, I strongly suspect it is a
red-black tree underlying the set)

Also, many implementations have special handling for the
'hint' boundary conditions ('end()' and 'begin()').
In your last example (where you inserted data in
high-to-low order) the iterator returned by the 'insert()'
call is actually 'begin()'. Since 'insert()' has special
handling for this 'hint' the performance of the NEXT 'insert()'
call is high.

If you pass an iterator that is neither 'begin()' nor 'end()',
when 'insert()' increments that iterator and it does not
point to a valid item, insert() starts its search with
'begin()' (i.e. your 'hint' iterator is ignored).

If you are inserting data in low-to-high order this
approach will speed things up greatly:

isIter = (is.insert(0)). first;
for (int i = 1; i < MaxVal; ++i)
is.insert(is.en d(), i);


Thanks for the good advice. I'll give this a try.

-Mark
Jul 23 '05 #7

This thread has been closed and replies have been disabled. Please start a new discussion.

Similar topics

7
12107
by: iqbal | last post by:
Hi all, We have an application through which we are bulk inserting rows into a view. The definition of the view is such that it selects columns from a table on a remote server. I have added the servers using sp_addlinkedserver on both database servers. When I call the Commit API of oledb I get the following error: Error state: 1,...
18
2280
by: deancoo | last post by:
I have gotten into the habit of often using copy along with an insert iterator. There are scenarios where I process quite a lot of data this way. Can someone give me a general feel as to how much of a performance hit I'm taking using this technique versus using 'copy' to copy directly into a container with elements in place? Thanks, d
1
2420
by: Mark P | last post by:
Is an insert with hint (in say std::set) only beneficial if the insertion occurs immediately before the pointed to location of the hint? Suppose the insertion occurs right after the hint location? Or suppose the insertion is near to the hint location but not adjacent? Is there a performance penalty for an approximate hint, a partial gain, or...
3
3838
by: rudymoore | last post by:
I'm often find-ing the same string in a map<string,xxx> I want to pass a hint to map::find but was disappointed to find that only map::insert takes the hint. Why would map::insert take a hint but not map::find? Thanks, Rudy
3
2505
by: Howard Hinnant | last post by:
I recently asked for a survey of multimap insert with hint behavior, in support of a paper I'm writing concerning lwg issue 233. My sincere thanks to Beman Dawes, Raoul Gough, Russell Hind, Bronek Kozicki, Nicola Musatti, John Potter and Maxim Yegorushkin for helping with that survey. Since I started work on this paper at least two people I...
16
3860
by: robert | last post by:
been ruminating on the question (mostly in a 390/v7 context) of whether, and if so when, a row update becomes an insert/delete. i assume that there is a threshold on the number of columns of the table, or perhaps bytes, being updated where the engine just decides, screw it, i'll just make a new one. surfed this group and google, but...
5
5819
by: asdf | last post by:
I have a program that reads sorted data from a database and inserts it element by element into a set. If the data is sorted is there a faster way to insert ? Meaning is there a way to tell the std::set to check the end of the tree before doing the insert as a fast check on where the key is going to end up ? Thanks.
6
9010
by: reppisch | last post by:
Hi Ng, I have a multiset for keeping elements sorted in a container but key values may also be equal. Is there any guaranteed traversal order within the duplicate keys of a multimap? When inserting duplicate keys to the multiset i need output order to be in insert order.
0
1388
by: subramanian100in | last post by:
consider the following program: #include <cstdlib> #include <iostream> #include <map> #include <utility> #include <string> using namespace std;
0
7831
marktang
by: marktang | last post by:
ONU (Optical Network Unit) is one of the key components for providing high-speed Internet services. Its primary function is to act as an endpoint device located at the user's premises. However, people are often confused as to whether an ONU can Work As a Router. In this blog post, we’ll explore What is ONU, What Is Router, ONU & Router’s main...
0
7746
by: Hystou | last post by:
Most computers default to English, but sometimes we require a different language, especially when relocating. Forgot to request a specific language before your computer shipped? No problem! You can effortlessly switch the default language on Windows 10 without reinstalling. I'll walk you through it. First, let's disable language...
0
8262
jinu1996
by: jinu1996 | last post by:
In today's digital age, having a compelling online presence is paramount for businesses aiming to thrive in a competitive landscape. At the heart of this digital strategy lies an intricately woven tapestry of website design and digital marketing. It's not merely about having a website; it's about crafting an immersive digital experience that...
0
8129
tracyyun
by: tracyyun | last post by:
Dear forum friends, With the development of smart home technology, a variety of wireless communication protocols have appeared on the market, such as Zigbee, Z-Wave, Wi-Fi, Bluetooth, etc. Each protocol has its own unique characteristics and advantages, but as a user who is planning to build a smart home system, I am a bit confused by the...
1
5641
isladogs
by: isladogs | last post by:
The next Access Europe User Group meeting will be on Wednesday 1 May 2024 starting at 18:00 UK time (6PM UTC+1) and finishing by 19:30 (7.30PM). In this session, we are pleased to welcome a new presenter, Adolph Dupré who will be discussing some powerful techniques for using class modules. He will explain when you may want to use classes...
0
5331
by: conductexam | last post by:
I have .net C# application in which I am extracting data from word file and save it in database particularly. To store word all data as it is I am converting the whole word file firstly in HTML and then checking html paragraph one by one. At the time of converting from word file to html my equations which are in the word document file was convert...
0
3768
by: TSSRALBI | last post by:
Hello I'm a network technician in training and I need your help. I am currently learning how to create and manage the different types of VPNs and I have a question about LAN-to-LAN VPNs. The last exercise I practiced was to create a LAN-to-LAN VPN between two Pfsense firewalls, by using IPSEC protocols. I succeeded, with both firewalls in...
0
3785
by: adsilva | last post by:
A Windows Forms form does not have the event Unload, like VB6. What one acts like?
1
2258
by: 6302768590 | last post by:
Hai team i want code for transfer the data from one system to another through IP address by using C# our system has to for every 5mins then we have to update the data what the data is updated we have to send another system

By using Bytes.com and it's services, you agree to our Privacy Policy and Terms of Use.

To disable or enable advertisements and analytics tracking please visit the manage ads & tracking page.