• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

New Superpi!

S

Shadowwarrior

n00b
Joined
Feb 28, 2005
Messages
9
Hi everybody, superpi has been for years de benchmark that we use for teste our CPU performance,
so I think that it's time to change, I found a new program, that calculates the
Euler Constant (2,7182818284590452353602) instead of the PI (3,14), and I'd like you test it
to see what you think. To test you need to do this:

1- Download the program here
2- Execute it
3- Go to Calculate -> Calculate -> Choose the number of digits (256k - 262144) and then click OK.
4- Wait
5- The program will show the time that the calculation took and a security number to avoid cheats, don't click ok yet.
6- Run CPU-Z and take a screenshot that show the result and your CPU.
7- Post the screenshot on the topic.

Here it is an screenshot example:

ecalc.JPG
 
why is it that it SCARES me when a Noobie asks me to download and try something on his first and only post :eek:
 
hulksterjoe said:
why is it that it SCARES me when a Noobie asks me to download and try something on his first and only post :eek:
Click to expand...
because you're older and wiser now.
 
Thanks, I definably feel safer getting the same file from you. :rolleyes:

Just another day in noobie hell
 
Gee, thanks for the warm welcome! :rolleyes: I'm hardly a noobie...maybe to this forum, I am, but not in general...but damn, that really makes me wanna stay!
 
Welcome to the Forum, but what would you expect in this thread..

The first post is from a noobie saying HEY DOWNLOAD THIS.. on his first post ever..

So we call him/her on it and bust some chops..

and then you walk right in with the exact same file name saying this program is better and its your first post. How is what your saying different.

To clarify: I personally think that anyone who would ever download a file from any new forum member on their first post is just asking for a world of hurt.
 
No offense, but you dont have nearly enough credibility for me to risk downloading that and getting a virus on my computer.

I, like many here, are paranoid tin foil panty wearing freaks when it comes to security. I dont download anything unless its from a reputable site and has a history of being adware/virus clean.
 
He's a MOD on EOCF for crying out loud... Run a virus scanner on these files if you guys are so paranoid! Jebus. :rolleyes:
 
OK, My post count isnt that high, but I've been around a while and I can vouch for this program from CodeWeasel...

There ARE other forums out there folks.... EOCF forums is probobly in the top5 (If not the number 1) of all the Overclocking forums and Codeweasel and I are established members there.... Just trying to spread the word for this new benchmark program..... If you dont want to try it out..... Then dont.... Two HUGE advantages of this program over superpi is that it calculates to the thousanth of a second..... No more 8462 people with the same superpi score... And a Vericode to make sure the results are acurate....
 
CodeWeasel said:
I suppose that's understandable...but damn, dude, harsh words! ;)

Perhaps I should provide some background?
http://forums.extremeoverclocking.com/showthread.php?t=158991
http://forums.extremeoverclocking.com/showthread.php?t=158997

The CodeWeasel in those threads are me, I am a mod of programming and off-topic at EOCF.com and wrote this benchmark utility for the members there.

If those threads don't provide proof that I'm not trying to screw anybody, I don't know what will... :(
Click to expand...
Impaqt said:
OK, My post count isnt that high, but I've been around a while and I can vouch for this program from CodeWeasel...

There ARE other forums out there folks.... EOCF forums is probobly in the top5 (If not the number 1) of all the Overclocking forums and Codeweasel and I are established members there.... Just trying to spread the word for this new benchmark program..... If you dont want to try it out..... Then dont.... Two HUGE advantages of this program over superpi is that it calculates to the thousanth of a second..... No more 8462 people with the same superpi score... And a Vericode to make sure the results are acurate....
Click to expand...
'm I a n000b too? :confused:

I can vouch for my fellow EOC members.

Here are my 4 Stickies:

http://forums.extremeoverclocking.com/showthread.php?t=136602

http://forums.extremeoverclocking.com/showthread.php?t=140905

http://forums.extremeoverclocking.com/showthread.php?t=157519

http://forums.extremeoverclocking.com/showthread.php?t=152446

Just because someone is new to your forum, does not make them a n000b!

Dave ;)
 
Erm, first of all, a virus scan won't always catch the very latest new stuff. He just announced it was a new version after all d-: That said, I must admit that I too wasn't exactly jumping on board with the idea of immediately downloading and installing the thing. I am sorry, but, you just never know.

Anyway, looks like everything checks out. I'm giving the program a shot right now. Sorry, I'm too lazy to give a screenshot, but, my signature has all the info you could need about my system surely (probably more accurate than CPU-Z since detection processes aren't always 100% accurate with non-standard settings.) This is the copy/paste result using version 1.2:

eCalc 256K test said:
---------------Results----------------
Total number of digits: 256000
Elapsed time of calculation: 0d 0h 0m 6s 652ms

Calculation start time: 14:03:37
Calculation stop time: 14:03:43

VeriCode: 4738812
Click to expand...

EDIT: Just for the sheer fun value:
eCalc 32M test said:
---------------Results----------------
Total number of digits: 32000000
Elapsed time of calculation: 0d 0h 13m 39s 394ms

Calculation start time: 14:09:44
Calculation stop time: 14:23:23

VeriCode: 15536300
Click to expand...

Just curious, but, what's the point of using one over the other with this vs SuperPi? Is one algorithm a better test of the CPU than the other? Seems to me like they both basically do the same kind of thing, no? No offense mind, I kind of like this program better than SuperPi. It feels a tad bit more professional quality to me somehow (don't know what it is about it exactly.) I like the idea of this VeriCode thing, but, if ordinary people don't have the algorithm to determine if it's valid or not it does us little good -- yet, if we do have it, then it becomes possible to fake it anyway.
 
What a warm welcome this forum alaways gives to new members. Just here to back up CodeWeasel, and davidhammock200.
 
Nazo said:
I like the idea of this VeriCode thing, but, if ordinary people don't have the algorithm to determine if it's valid or not it does us little good -- yet, if we do have it, then it becomes possible to fake it anyway.
Click to expand...
Ah, but you DO have the ability to check anyone's results without knowing the algorithm. I thought of everything. ;)

Under the eCalc menu is the VeriCode Verification System, you just enter the time they say they got, and the code they posted with that time.

As for why this over SuperPi... I like the logging method of this one better, to tell you the truth. With this, you get a running log of all of your calculations. Also, this one counts to the milliseconds, so like mentioned before, no more uberlong member lists under one time. The VeriCode system's a plus, too, and will only get better. Lastly, this program is compiled with optimizations for the most popular CPUs, something I'm not sure SuperPi has...I could be wrong, though. But I do know eCalc provides a pretty decent forshadowing of future performance, from my experiments here in the dorm.
 
There are versions of SuperPi with optimizations, though it does seem to me like your program is using better optimizations. (I think there was only unoptomized vs P4 optomized. Don't recall if there was an Athlon optomized version.) Imo, it is more realistic to use whatever the cpu can actually do for a more real world situation.

Anyway, the VeriCode thing has me sold. Sounds like a good idea to me. I didn't notice the option to verify it.
 
Nazo said:
There are versions of SuperPi with optimizations, though it does seem to me like your program is using better optimizations. (I think there was only unoptomized vs P4 optomized. Don't recall if there was an Athlon optomized version.) Imo, it is more realistic to use whatever the cpu can actually do for a more real world situation.

Anyway, the VeriCode thing has me sold. Sounds like a good idea to me. I didn't notice the option to verify it.
Click to expand...
Glad you like it.

If anyone has any questions/comments/suggestions, I'm all ears.
 
hulksterjoe said:
Welcome to the Forum, but what would you expect in this thread..

The first post is from a noobie saying HEY DOWNLOAD THIS.. on his first post ever..

So we call him/her on it and bust some chops..

and then you walk right in with the exact same file name saying this program is better and its your first post. How is what your saying different.

To clarify: I personally think that anyone who would ever download a file from any new forum member on their first post is just asking for a world of hurt.
Click to expand...

lol, I didn't now that people was so scared about virus... no one knows an anti-virus? And if you are still scared, I've posted this same topic on a lot of forums, but with another nickname because mine was already been used and no one complained:

http://www.nforcershq.com/forum/viewtopic.php?t=57786&postdays=0&postorder=asc&start=0

http://www.xbitlabs.com/forum/viewtopic.php?t=6627

As you can see, the other forums members downloaded the program and tested and nothing happened. I can't make no one test the program, I just suggested it.
 
Shadowwarrior said:
lol, I didn't now that people was so scared about virus... no one knows an anti-virus?
Click to expand...

Nazo said:
Erm, first of all, a virus scan won't always catch the very latest new stuff. He just announced it was a new version after all d-:
Click to expand...

Seriously. Do you think the anti-virus software just magically knows the intentions of the programmer? You can't always determine if something is a virus or not. In fact, I have some ideas of how one could make an absolutely deadly virus that may well end up ignored by those programs. Don't bother asking, lol, I fear if I ever say it someone will make it. The point is, with just this idea alone anyone who knows more programming than me at all could probably make one that wouldn't be detected. It's quite easy for these things to slip right past an anti-virus software and you wouldn't know until it's too late. Sometimes even if you run them in a controlled environment (virtual system) you won't know something is wrong until much later. (Of course, with software like this that would defeat the entire purpose.)
 
Dude, shadow, stop spreading that one! 4 minutes to calc 262 thousand digits? wtf is that using, factorials?

EDIT: ROFL that one's title is "E CALC FAST"
 
Lol, yeah, I did have to kind of laugh when I took a look at mine after seeing his. I didn't think a ~2.2GHZ Athlon XP would do THAT much better than a ~2GHz Athlon 64... 4 minutes on his, 6 seconds on mine? ^_^ Ok, his was a bit more digits, but, still, a matter of 1 second or so difference at the most.
 
hulksterjoe said:
Welcome to the Forum, but what would you expect in this thread..

The first post is from a noobie saying HEY DOWNLOAD THIS.. on his first post ever..

So we call him/her on it and bust some chops..

and then you walk right in with the exact same file name saying this program is better and its your first post. How is what your saying different.

To clarify: I personally think that anyone who would ever download a file from any new forum member on their first post is just asking for a world of hurt.
Click to expand...

And your crybaby, whiney-ass opinion would be...? Your a joke, you dont evan know CodeWeasel. Atleast try to welcome new forum members with a smile on your face and try to establish a thougt for the person first before saying crap about his post



:mad:
 
Maddawg said:
And your crybaby, whiney-ass opinion would be...? Your a joke, you dont evan know CodeWeasel. Atleast try to welcome new forum members with a smile on your face and try to establish a thougt for the person first before saying crap about his post
Click to expand...
Whoa there, man, relax, everyone's past this. I wouldn't expect anyone here to know me, and they make a valid point of being wary of, hell, ANY file on the internet.

But maddawg also makes a good point in the last part of his post. I thought it was a little rude the way Shadow and I were greeted, to be honest.
 
CodeWeasel said:
Dude, shadow, stop spreading that one! 4 minutes to calc 262 thousand digits? wtf is that using, factorials?

EDIT: ROFL that one's title is "E CALC FAST"
Click to expand...

Damn, I forgot to update the screenshot. The first post one was an old version of the program. Here it is a screen of the new one (the available for download)

http://img171.exs.cx/my.php?loc=img171&image=ecalc7yb.jpg

And the name of the program is e-calc FAST because the previous version was VERY slow!
 
Shadowwarrior said:
the previous version was VERY slow!
Click to expand...
And so is this one. Dude, it shouldn't take 2 minutes to calculate 262144 digits of e, especially when the Euler Constant's algorithm can (and should) be coded to be much more efficient.

I have the sneaking suspicion that this program uses factorials, probably because that's all the creator new from school or something...blech
 
Yeah. If I can calculate 256K in about 6 1/2 seconds, you need to be able to calculate 256Ki (note my use of multipliers here! This is important because we aren't talking about bytes/etc, but a real number...) in a very near area on an Athlon 64... Get any more than, say, 12 seconds and something's up. (Don't know how much difference there will be for this running at ~ 2.2GHz and his running at ~ 2.0, but, if it's hit 12+ seconds, something is horribly wrong based on what I'm getting with CodeWeasle's version.)

I do note that my point about detection processes being inaccurate was just proven in that screenshot. Lol, how in the heck did it get the idea you have a NEGATIVE HTT bus anyway... ^_^ And that's some pretty low latency memory apparently, considering it runs on 0s.... Lol, new "no-latency" memory eh? What's that, Corsair UberXMS? ^_^ Lol. No, I'm not picking on you, I'm picking on CPU-Z for not having a backup to detect that some of those things aren't possible.
 
Nazo said:
I do note that my point about detection processes being inaccurate was just proven in that screenshot. Lol, how in the heck did it get the idea you have a NEGATIVE HTT bus anyway... ^_^ And that's some pretty low latency memory apparently, considering it runs on 0s.... Lol, new "no-latency" memory eh? What's that, Corsair UberXMS? ^_^ Lol. No, I'm not picking on you, I'm picking on CPU-Z for not having a backup to detect that some of those things aren't possible.
Click to expand...


Yeah! I thing that negative HTT bus funny too, but I think that is only with me, because I asked a friend to test and on his machine the HTT is normal:



This CPUID that you can see on the image, is one that create a security code, which check if the image is true and avoid cheating. You can test it here:

CPUID
 
Disclaimer: I am the creator of the program. My friend Shadowwarrir asked me to respond to the criticism presented here.

And so is this one. Dude, it shouldn't take 2 minutes to calculate 262144 digits of e, especially when the Euler Constant's algorithm can (and should) be coded to be much more efficient.

I have the sneaking suspicion that this program uses factorials,
Click to expand...

Yes it does, just like EVERY OTHER PROGRAM. PiFast uses factorials.(And it's bloody fast, it calculates 1Million digits im 3 seconds her) However, it does it in a more efficient manner, as explained here .

ECALC FAST (ROFL)! is an experiment. The first version took about 10 minutes (mid-range machine) to calculate 10k digits. THEN came ECALC FAST, that's much, much faster than the first version (but still slow)

I'm sorry I'm not as smart as the PiFast guys, but anyway, the purpose is to have fun and learn, not to calculate a bazillion digits in a millisecond...

Anyway, it's SUPPOSED to be not that fast, for benchmarking purposes...

And if you are afraid of viruses, etc, I recommend you run it under a restricted XP account (with proper security in place, like NTFS , etc)

Or just buy VirtualPC, have Windows installed and run! Yeah, it will be pretty slow, but..
 
That's the thing... Codeweasle's eCalc takes about 6 1/2 SECONDS for me to calculate 256K digits. Why does yours take me 2 minutes, 22.60 seconds (verification code d355be3e9d1f69926e032daf36483785fe78f997dbedac2eb2baf368) to calculate 256Ki digits? It's not that many more digits...
 
Nazo said:
That's the thing... Codeweasle's eCalc takes about 6 1/2 SECONDS for me to calculate 256K digits. Why does yours take me 2 minutes, 22.60 seconds (verification code d355be3e9d1f69926e032daf36483785fe78f997dbedac2eb2baf368) to calculate 256Ki digits? It's not that many more digits...
Click to expand...

What's the difference between K and Ki??

Anyway, because they were built differently. PiFast would take 3 seconds to calculate 1Million digits... it's 2 times faster for 4x the digits.

And the calculation follows the (pretty obvious path) of factorials. NO, it does not calculate 1000!, but it does calculate 1/1000!

It takes approximatelly 3,8 times longer for calculating 2*x digits than to calculate x digits.

It's aking to asking why a Ferrari is faster than a Ford Fiesta, or something like that...
 
Shadowwarrior said:
Yes it does, just like EVERY OTHER PROGRAM.
Click to expand...
Sorry, let me clarify. I meant does it use a direct implementation of the Taylor series? As in, did you write a factorial function and execute it over and over again? Because a good computation algorithm does not do this. For instance, PiFast uses binary splitting. If you look at the algorithm I used, you will not see the popular recursive "factorial( int num )" function anywhere.
Code: 
int factorial( int num )
{
    if ( num > 1 )
        return num + factorial( num - 1 );
    return 1;
}

Recursion may make a solution look elegant, but it can also be terribly inefficient, such as in the case above.
 
CodeWeasel said:
Sorry, let me clarify. I meant does it use a direct implementation of the Taylor series?
Click to expand...

Yes! Pretty much. Its logic is similar to this program: E classic , with the exception of some MMX and SSE optimization.

[QUOTE/As in, did you write a factorial function and execute it over and over again? [/QUOTE}

No, that would be awfully slow (even MORE than it is today).

I know what you mean: I'll need 2! first, then when I need 3! I'll calculate it AGAIN (1*2*3), not using 2! as a starting point. It does not work like this...

Instead, I calculate 1, then 1/2, then 1/3! (by dividing 1/2 by 3) and so on (and adding the results)

Because a good computation algorithm does not do this. For instance, PiFast uses binary splitting. If you look at the algorithm I used, you will not see the popular recursive "factorial( int num )" function anywhere.
Click to expand...

I know. And anyway, 1000! would probably overflow the stak or something.
 
Dude, if you're using the algorithm from that E classic file, no wonder it's so slow! That thing's nasty...it pretty much calculates the factorial all over again on each iteration like I had though, though not as directly as I had thought...plus, it wastes a LOT of cycles just outputting the numbers. And you're wasting cycles outputting every iteration of your loop.
 
Shadowwarrior said:
What's the difference between K and Ki??
Click to expand...
K means Kilo, Ki means Kibi. A kilo is 1000, a kibi is 1024. ECALC FAST is basically using 256*1024 while eCalc uses 256*1000. This is referring to the number of digits. A number of digits is a real number, so doesn't have to follow the computer system rules where you end up having to use a basis of 8,16,32,etc so much. Mind you, I don't know how the algorithms themselves are doing all this, perhaps it's actually easier to make it do

Anyway, because they were built differently. PiFast would take 3 seconds to calculate 1Million digits... it's 2 times faster for 4x the digits.
Click to expand...
Er... Are you saying it actually does larger numbers faster than smaller numbers? Is that even physically possible? Well, I figured, better just to look and see if it does or not myself. I ran E CALC FAST and selected the highest number of digits it does (8Mi) and then compared to eCalc at a comparable number of digits (8M.) To make this as accurate as possible, I ran different sized tests on eCalc and figure they should be added to the 8M to make up about 8Mi. Here's the 8M test:
---------------Results----------------
Total number of digits: 8000000
Elapsed time of calculation: 0d 0h 3m 25s 853ms

Calculation start time: 18:22:35
Calculation stop time: 18:26:01

VeriCode: 21835476
Click to expand...
That's 3m, 25s, 853ms. You have my earlier test for 256K weighting in at 6s, 652ms.
Together that's up to 8256000. Next up I run 128K making it as close as I can get it reasonably (8384000.) Not 100% accurate, but, the difference is 4608 digits.
---------------Results----------------
Total number of digits: 128000
Elapsed time of calculation: 0d 0h 0m 3s 287ms

Calculation start time: 18:32:08
Calculation stop time: 18:32:11

VeriCode: 1269840
Click to expand...

That's (3m, 25s, 853m) + (6s, 652ms) + (3s, 287ms)
Grand total for eCode with 8284000 digits (assuming I did my math with time correctly): 4m, 0s, 866ms

Ok, so I fired up E CALC FAST and select 838608 digits to test this. Well, it's been over 5 minutes for certain (I think at least 10) and I'm still waiting on the results (I'll add an edit when it's done with the verification code and all.)

I'm not criticizing you or anything, these things aren't easy to make, so I can easily accept one program may do something with a method that is more efficient than another. All I'm saying is that we might just as well use the one that's more efficient (eg exponentially faster) and has a built in verification system that the average person can use without having to contact the coder or whatever.

EDIT1: Still running. >20 minutes already. Bear in mind I got 32M digits in eCalc in roughly 13 (or was it closer to 13 1/2) minutes. Something's up here. But, hey, don't take my word for it (mind you, this is why the verification codes are there.) Test it yourself.
BTW, if you DO want us to continue working with it, could you give us an english localized version? I know it's fairly simple, but, kind of annoying when your menus say things like "Arquivo" rather than File (or whatever.) I don't know how many people in this forum would prefer that, but, it seems like since the forum assumes

EDIT2: Sorry, I hit an hour and got tired of not being able to use my PC. I'll start it up again when I go to sleep.

PS. I posted all this from my laptop with no access to the PC. The pc I just leave sitting there without even moving the mouse.

EDIT3: Started it up again. Let it run while I was gone for several hours. When I came back, it was up to 481628 digits. I closed it and gave up. Sorry, if you want an exact number you'll have to do it yourself. Frankly, considering how much faster it is to do about 3.8x as many digits in eCalc, I'm inclined to say something is very fishy here. I started it at about 7am, closed it just now (about 11:15 apx.) I suppose this program might make a better test of the effects of processor cache and will probably show itself much better on systems with larger amounts (like the new 2MB processors) than on those of us not blessed with that much "cache" (read cash -- that's right, I made a VERY bad joke. *shrugs*) This latest test was with the E CALC FAST English translation that you linked to.
 
Shadowwarrior, you need to try testing those two and comparing the results yourself. Something's fishy here. IMO, it makes a great learning experience if CodeWeasle's is opensource. (Is it?) Whatever algorithm he is using is massively more efficient.
 
Nazo said:
Shadowwarrior, you need to try testing those two and comparing the results yourself. Something's fishy here. IMO, it makes a great learning experience if CodeWeasle's is opensource. (Is it?) Whatever algorithm he is using is massively more efficient.
Click to expand...

The entire program is not open source. However, a simple C implementation of the algorithm I used comes in a text file with the program.
 
[QUOTE}Anyway, because they were built differently. PiFast would take 3 seconds to calculate 1Million digits... it's 2 times faster for 4x the digits.[/QUOTE]

Er... Are you saying it actually does larger numbers faster than smaller numbers? Is that even physically possible?
Click to expand...

For different programs, YES! I mentioned PiFast... PiFast is faster than ECALC CodeWeasel (and of course, faster than ECALC FAST )

Ok, so I fired up E CALC FAST and select 838608 digits to test this. Well, it's been over 5 minutes for certain (I think at least 10) and I'm still waiting on the results (I'll add an edit when it's done with the verification code and all.)
Why does yours take me 2 minutes, 22.60 seconds to calculate 256Ki digits?
Click to expand...

Let's do some math here... if it takes 2m22.60s to go to 256Ki then it shoud take about:

2059s for 1Mi digits -> 34m19s
8h2m29s for 8Mi digits...

Yes, it sucks! But it's my first shot at it...
Click to expand...
Click to expand...
 
You must log in or register to reply here.
Back
Top