Jump to content
Sign in to follow this  
el kido

Linux and Super Pi

Recommended Posts

QUOTE(zeno0771 @ 6:44pm Thu Jan 10 2008) 1461119[/snapback]
10.744 seconds...don't know if Linux boxes can play in this forum but if it makes anyone feel any better I had a full desktop going complete with Compiz-Fusion and was web-browsing on my other monitor which I cropped out of the pic to save bandwidth. The other terminal window is my POV-ray score-

How did you get 10.744 at 3.6ghz??? I cant even do that at 4.2ghz

Share this post


Link to post
Share on other sites

How did you get 10.744 at 3.6ghz??? I cant even do that at 4.2ghz

 

Better version of Super PI and a better Operating System.....;)

 

:geezer:

Share this post


Link to post
Share on other sites

How did you get 10.744 at 3.6ghz??? I cant even do that at 4.2ghz

 

Never mind that, how did you get 4.2 on your Q? My homebrew watercooling keeps a pretty good lid on temps (22C idle/36C load, with the hottest core hitting the mid-60s) but I gotta push almost the same volts I needed to keep my E6300 stable at 423x7. This thing's recently deveoped a case of vdroop-itis; ~1.45v in BIOS/1.46 indicated, and it goes down to 1.39 if I load more than 2 cores...

4.2 GHz...man, I could drop into the 7's with that, easy... and this is with 4 GB RAM and an OEM chip! :mrgreen:

Share this post


Link to post
Share on other sites

Better version of Super PI and a better Operating System.....;)

 

:geezer:

 

I looked at the beginning of the thread, was there an older version I should have been using? Don't want to tee anyone off with only my 3rd post here :unsure:

Honestly, I was just happy to find more than one bench tool that ran on both Windoze and *Nix...Few forums let Linux boxes play (something about an "unfair advantage") and I can see if you don't run an X server and shut down anything that isn't being used you could seriously artificially lower your score but in the interest of fairness I run all my benches just as I would use my machine on a daily basis: Full GUI w/ Compiz (yes, the cube is cool) and an average of 200+ running processes. In fact I used 4 instances of Pi 32M to pronounce my rig stable at 3.6; after an hour of Prime on all 4 cores the temp on the hottest core started outrunning my w/c :angry:

Fact is, though, I was running XP back when I first started OCing; I started learning Linux about the same time and there was something like an 11 second difference between the two (in the 1M test) with no hardware changes (sorry, this was 2+ years ago, don't have any screenies)

Edited by zeno0771

Share this post


Link to post
Share on other sites

Never mind that, how did you get 4.2 on your Q? My homebrew watercooling keeps a pretty good lid on temps (22C idle/36C load, with the hottest core hitting the mid-60s) but I gotta push almost the same volts I needed to keep my E6300 stable at 423x7. This thing's recently deveoped a case of vdroop-itis; ~1.45v in BIOS/1.46 indicated, and it goes down to 1.39 if I load more than 2 cores...

4.2 GHz...man, I could drop into the 7's with that, easy... and this is with 4 GB RAM and an OEM chip! :mrgreen:

 

I have a good board with good ram and good cooling. :mrgreen:

 

are you using 4 x 1gb?

Share this post


Link to post
Share on other sites

I have a good board with good ram and good cooling. :mrgreen:

 

are you using 4 x 1gb?

 

Yep. Don't get me wrong, I'm not complaining about 3.6 with this setup. If I got greedy I could pull 2 GB out and get what, maybe another 200 MHz? Nah...this is an every-day machine, I'm good with it-

Share this post


Link to post
Share on other sites

Well, as of earlier today, I guess this is officially the only board that will allow a Linux score post; got an honorable mention (of sorts) at another one only to get denied by the mod. Oh well, at least I know I trumped the next highest score there by 2 seconds and they were running an E6850 jacked to 4.2 GHz, and needed 1.77 volts to do it.

Thanks for not being Nazi about my OS, guys. Next up: Gigabyte DS2R and 2x2GB in a Microfly case, on water...

Share this post


Link to post
Share on other sites

:IG:

 

You can play as long as you don't beat my 9.36. :P

 

Posted Image

 

j/k :wp:

Share this post


Link to post
Share on other sites

:IG:

 

You can play as long as you don't beat my 9.36. :P

 

Posted Image

 

j/k :wp:

 

Um, well...that's some loud MHz you have there...you have a 1.3 sec lead on me and needed 1.3 GHz to do it...

:unsure:

Yeah, I'm cool with a solid Top 5 spot til I get my Yorkie (after the price drop next year!!! :blink: )...I just got this Q a few weeks ago.

I hear Gigabyte's G33M DS2R's are evil clockers but I won't have that for another month so yeah...

Share this post


Link to post
Share on other sites

Um, well...that's some loud MHz you have there...you have a 1.3 sec lead on me and needed 1.3 GHz to do it...

:unsure:

 

Yeah, in my experience that's how it works. Linux is a lot faster than Windows. Higher MHz = lower PI times. What did I miss?

Share this post


Link to post
Share on other sites

If I set up a machine with windows and Linux, can the scores from Linux super pi and any other bench that will run in Linux be used, and then use the windows partition to run the applications that will only be run in windows?

 

It is after all one computer.

 

Just wondering because superpi runs about 4 seconds faster in Linux.

 

The test below is a completely stock q6600, not a single tweak to anything on the system.

 

Posted Image

 

Any benchmark that has is cross platform just runs better in Linux.

 

So why couldn't I do this?

 

Does this scare the windows only folks too much?

 

I have been tinkering with a few things and am considering entering, but before I do, I just want to know if I can do this.

Edited by Bruce

Share this post


Link to post
Share on other sites

If I set up a machine with windows and Linux, can the scores from Linux super pi and any other bench that will run in Linux be used, and then use the windows partition to run the applications that will only be run in windows?

 

It is after all one computer.

 

Just wondering because superpi runs about 4 seconds faster in Linux.

 

The test below is a completely stock q6600, not a single tweak to anything on the system.

 

Posted Image

 

Any benchmark that has is cross platform just runs better in Linux.

 

So why couldn't I do this?

 

Does this scare the windows only folks too much?

 

I have been tinkering with a few things and am considering entering, but before I do, I just want to know if I can do this.

 

As far as why you "couldn't do this", it really depends on the forum; some just want to keep the playing field as level as possible, others just have a chip on their shoulder about Linux' perceived "unfair advantage". I've seen some people get downright childish about it, and in those cases I just let the kids have the sandbox to themselves. There's another hwbot forum I subscribe to where I was told I could post the score but it wouldn't be entered in their top 10 list because of something having to do with hwbot not accepting non-Windows software or something (my score would've been right between 2 C2D's running @ 4.7GHz+, and the rest were all QX Yorkies).

What you propose actually sounds like a pretty good idea, if only from a baselining standpoint; some would split hairs over the validity of any comparison (especially windoze fanboys) but keeping any testing variables to a minimum is always a good idea if you want unbiased results.

 

EDIT: What are you using for your sysinfo? Is that SUSE-specific?

Edited by zeno0771

Share this post


Link to post
Share on other sites

If I set up a machine with windows and Linux, can the scores from Linux super pi and any other bench that will run in Linux be used, and then use the windows partition to run the applications that will only be run in windows? ...

That is an excellent idea and IMO is no different than using XP for 3DMark 01 and 03, Win2K for Aquamark, Vista for 05 and 06, etc.

 

Unfortunately, after a discussion last year about Linux and Super Pi, the rules to the contest were amended as follows:

 

The Rules......

One system per player, unless the forum is short on the 20 player target.

Then and second system by each player can be added to fill out the rosters (only 1 system can be on the A Team)

The 2nd Rig must have a different CPU and Reduced VGA.

Reduced VGA means if you use SLI, you cannot use the same set up on your 2nd Rig, you can only use 1 card.

But if you have a lesser set of VGA cards... thats OK!

Standard HDD performance tuning (like RAID) on ATA, SATA, SATA II are ok. No limit on # of HDD's.

Ram or Virtual Drives are not allowed.

Acceptable Operating Systems: Windows 2000, XP, Server 2003 or Vista and all variants of each (32-bit/64-bit, home/pro/ultimate, etc.)

Acceptable Drivers: ALL (No driver limitation)

Acceptable Hardware: All (except unreleased ES chips and similar)

From here: http://www.theraptorpit.com/forum/index.php?showtopic=1359

Share this post


Link to post
Share on other sites

EDIT: What are you using for your sysinfo? Is that SUSE-specific?

 

Sysinfo is a kio-slave created by Novell/SuSE, it is of course open source and last year folks started porting it to other distros. You can get the source code from kde-apps and compile it yourself.

 

http://www.kde-apps.org/content/show.php/K...B?content=58704

 

Or seeing as you are using Fedora you could use the one folks compiled for fedora.

 

http://www.kde-apps.org/content/show.php/F...1?content=63676

Share this post


Link to post
Share on other sites

That is an excellent idea and IMO is no different than using XP for 3DMark 01 and 02, Win2K for Aquamark, Vista for 05 and 06, etc.

 

Unfortunately, after a discussion last year about Linux and Super Pi, the rules to the contest were amended as follows:

From here: http://www.theraptorpit.com/forum/index.php?showtopic=1359

 

Thats sad really. Sad in that they don't limit the insane amounts of money people spend on specialized hardware, they don't limit what drivers or software can be used, they don't limit Bios hacks, they don't limit anything except one thing. You can't use the operating system of your choice.

 

I see huge loop holes that seem to have been created in the rules to allow certain people with a lot of cash to do things many others simply couldn't afford to do, yet they will not allow something that anyone can do legally and for free. :surrender:

Share this post


Link to post
Share on other sites

Thats sad really. Sad in that they don't limit the insane amounts of money people spend on specialized hardware, they don't limit what drivers or software can be used, they don't limit Bios hacks, they don't limit anything except one thing. You can't use the operating system of your choice.

 

I see huge loop holes that seem to have been created in the rules to allow certain people with a lot of cash to do things many others simply couldn't afford to do, yet they will not allow something that anyone can do legally and for free. :surrender:

 

Old hot-rodding rule..."Speed costs; how fast do you want to go?" :overclocking:

Seriously though, among all the other car-racing/PC modding analogies, one thing stays the same, and that's cash. Hot Rod and Car Craft each try every few years to come up with a "real street" competition (Hot Rod's is called the Pump-Gas Drags...unless they got rid of it already) and every time they do it ends up being a money-contest, no matter how many rules they come up with to make the cars qualify as street-legal.

It's all in how you play the game, I guess... :tux:

Share this post


Link to post
Share on other sites

The fact that speed costs is not the issue.

 

Thats a no-brainer.

 

The issue is that even in the world of auto racing they don't dictate what operating system can control the hardware in the race cars.

 

You see, what it comes down to is this.

 

These contests are not about who can get the most out of a computer sytstem or the hardware in it.

 

It comes down to who can make windows runs fast.

 

These contests should not be called hardware benchmarks/contests, simply put because they are not hardware contests.

 

They are contests to see who can keep windows running, and who can't. :lol:

Share this post


Link to post
Share on other sites

The fact that speed costs is not the issue.

 

Thats a no-brainer.

 

But it shouldn't necessarily lock anyone out of a competition (of any sort). IIRC that's the original idea behind OCing in the first place: To show up trust-fund babies with know-how. It could be argued that it also resulted in Linux appearing on home desktops...

The issue is that even in the world of auto racing they don't dictate what operating system can control the hardware in the race cars.

 

Not true; telemetry mods in pro racing are strictly regulated, and even if you discount that, the hardware is strictly regulated (i.e. NASCAR's 358 cubic-inch boat-anchor mandate for all competitors despite the fact that none of them have ever or will ever make one that size for any other purpose). Actually, that doesn't sound like a bad idea: How about classes (so you don't have QX9860s competing with E6300 C2Ds)? Without trolling or being arrogant about it, I've always liked the idea that my US$260 OEM Quad can run within a second of a $1200 QX hot-off-the-presses, but in the interest of fairness... :mrwinky:

You see, what it comes down to is this.

 

These contests are not about who can get the most out of a computer sytstem or the hardware in it.

 

It comes down to who can make windows runs fast.

 

These contests should not be called hardware benchmarks/contests, simply put because they are not hardware contests.

 

Well, if almost everyone is running the same 2 Asus boards and the same 2 types of RAM, what's the point? The trick is, with only one OS allowed, they have to be called hardware benchmarks, because a system is defined as both hardware and software; further, anyone who does testing for a living will tell you that to have an accurate comparison you need as few variables unaccounted for as possible. Unfortunately for the sterile world of the lab, the differences in scores are almost guaranteed to be because of the OS.

They are contests to see who can keep windows running, and who can't. :lol:

 

HEEEEYYYYY, that gives me another idea...few years back I saw a thread that was basically an uptime contest (mainly *Nix players there, surprise-surprise)...I want to see a Windows XP uptime thread...How long can you keep your Redmond calculator going at a stretch before you have to reboot. Now THAT would be funny...

In any event, most OCers are gamers, and that limits them (for now) to Windows; it's only fair that they would want to know how similar systems perform under similar (i.e. same OS) conditions. :hammer:

Incidentally, I wonder how many of those 8-second QXs out there can be considered stable? :unsure:

Share this post


Link to post
Share on other sites

The thing a lot of people forget when comparing benchmarks between Windows and Linux is that not only is Linux itself faster than Windows, but the executable files are usually compiled with different tools - and GCC usually generates more efficient code than the M$ compiler does - so in addition to the different operating system you're also comparing two different programs even though they appear to essentially do the same thing at the user level they can be quite different underneath. I also see some comments in that Linux version of Super Pi about "the C source program was manually optimized". It's entirely possible that the Windows version could also be sped up with some optimization.

 

In contrast for running 3dMark under XP instead of Vista for example you're still running the same code under either OS.

Share this post


Link to post
Share on other sites

The thing a lot of people forget when comparing benchmarks between Windows and Linux is that not only is Linux itself faster than Windows, but the executable files are usually compiled with different tools - and GCC usually generates more efficient code than the M$ compiler does - so in addition to the different operating system you're also comparing two different programs even though they appear to essentially do the same thing at the user level they can be quite different underneath. I also see some comments in that Linux version of Super Pi about "the C source program was manually optimized". It's entirely possible that the Windows version could also be sped up with some optimization.

 

In contrast for running 3dMark under XP instead of Vista for example you're still running the same code under either OS.

 

D'oh! And all this time I've been running Super_Pi just as I downloaded it! :cr@sh:

You're right though, and I don't foresee any windoze users going to the trouble...I just ran Super Pi mod 1.5 in Wine and it added a healthy 5 seconds to my 1M time, and THAT ain't no fun... :Tumbleweed: which kind of emphasizes Bruce's point: Are we testing just the CPU or the system as a whole? I mean I know that Pi is supposed to be strictly for CPU but a CPU by itself isn't going to do much...

Edited by zeno0771

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×
×
  • Create New...