*OFFICIAL* Pi-to-1M Calculating Contest. How fast can YOU be?

  • Thread starter Evolution.
  • 203 comments
  • 17,486 views
8908pi.gif


Compaq Presario V4075EA with 1GB of ram (standard is 512). Bare in mind I have 85 processes running, though...

9784cpu.gif
 
opendriver19a
Not to shabby for a dell:
Click to expand...

Check Out this Dell then!

:D Thought I'd weigh in with my score... I had to use a different version of Super Pi, as the link in the first post doesn't work. If you want me to run this test again with the same version, let me know.

19.812s (XPS700: E6700,2gb DDR2 533mhz ram,590sli chipset). I could of probably got this faster still, if it wasn't for the slow ram. :sly:

pipl8.jpg
 
42 seconds.

Athlon 64 3500+, 2Gb RAM, can't remember my motherboard model (it's a Gigabyte).

I bet memory makes little or no difference in a test like this - when it starts it says it's using 8Mb or so and I can't see why hundreds of megs of RAM would a million consecutive simple routines. More like a measurement of the raw number crunching ability of your PC.
pisx8.jpg



KM.
 
Keep in mind, KM, it's quality, not quantity. Faster RAM makes a difference.

Also, KM, you need to use a Super Pi mod that shows up to 1/1000th of a second to be eligible for the leaderboard (another use currently has a 42s time also).
 
*Wanders in*

1mpiby2.jpg


:)

CPUID's:

Core 1 Core 2

I know it'll be beaten (until I can upgrade my cooling at least :mischievous: ), but it's a start. 💡

Specs so you don't have to copy it from CPUID:

Intel Core2 Duo E6600 @ 2.7GHz
Gigabyte GA-965P-DS3
Corsair Twin2x2048-6400
 
magburner
Check Out this Dell then!

19.812s (XPS700: E6700,2gb DDR2 533mhz ram,590sli chipset). I could of probably got this faster still, if it wasn't for the slow ram.
Click to expand...
Can you post a screen shot of your calculation again, with a shot of CPU-Z showing your processor??? The reason I ask is, your time is only slightly slower than that of people running some moderately overclocked C2D chips. I'm sure your Dell is pretty fast. But, I've seen people hit 18s times (like jammyozzy, btw)with the same chip as you and with faster RAM running at a much higher speed. I just want to see what your Dell is running at. You can see the screen shot of what I mean. That being my computer here at the shop, by the way. New low time. But, with a 900mhz Athlon with a whopping 128MB of RAM, its just not SuperPi worthy.

Hilg

http://www.cpuid.com/index.php

superpiry0.png
 
Sweet, I made the list. And dead last, right where this computer deserves. I'll try and remember to do this when I get home on one of my more respectable machines. Thats just despicable. HA HA

Hilg
 
What did you do for overclocking the E6600 Jammy? We're running the same stuff minus the RAM. :)
 
MachOne
Keep in mind, KM, it's quality, not quantity. Faster RAM makes a difference.
Click to expand...
It depends on the application. Having the latest greatest 667Mhz dual-channel DDR2 RAM mightn't be that useful in a simple situation like this where your CPU is crunching numbers far faster than your snazzy new RAM can deliver it to the CPU. It took my PC 42 seconds to calculate a million digits of pi. I wonder how much data could theoretically be sent down my CPU-RAM pipe in that time. In my case it's 42 seconds x 16-bytes at a time (dual channel) x 400Mhz = a couple of hundred gigabytes of data. I don't think RAM speed or RAM capacity is relevant in this benchmark, simply because it's so simple and doesn't eat up more than a tiny fraction of the memory capacity of a modern PC.

Also, KM, you need to use a Super Pi mod that shows up to 1/1000th of a second to be eligible for the leaderboard (another use currently has a 42s time also).
Click to expand...
The link in the OP wasn't working earlier today, but I already had a version of Super Pi downloaded from a previous thread on some forum somewhere. So I just used it.


KM.
 
KieranMurphy
It depends on the application. Having the latest greatest 667Mhz dual-channel DDR2 RAM mightn't be that useful in a simple situation like this where your CPU is crunching numbers far faster than your snazzy new RAM can deliver it to the CPU. It took my PC 42 seconds to calculate a million digits of pi. I wonder how much data could theoretically be sent down my CPU-RAM pipe in that time. In my case it's 42 seconds x 16-bytes at a time (dual channel) x 400Mhz = a couple of hundred gigabytes of data. I don't think RAM speed or RAM capacity is relevant in this benchmark, simply because it's so simple and doesn't eat up more than a tiny fraction of the memory capacity of a modern PC.
Click to expand...
Quite wrong.

I dare you to run into your BIOS, change your CAS latency on your memory a tick higher, and run the same benchmark. $20 says it'll be noticeably slower.

You see, even though DDR667 RAM could theoretically push x gigabytes of data back and forth in any given amount of time, in this particular test, the importance is on how quickly it can store and retrieve small bits of information, not how much information it can store (think of Hard Drives; with a standard rotational speed of 7200.7rpm, a hard drive on average can seek random bits of information in, say, 8ms. However, on the other hand, it can do a sustained read of 60mbps – That’s still not going to help you do a random seek on a thousand small bits of information stored in random places on your hard drive).. You're right in that huge amounts of memory bandwidth not playing a role in this particular test, however what I'm getting at are memory timings, which regardless of how much or little data you're reading or writing, it still has to store and access it - In the case of this test, very quickly.

To put it into a practical example, my personal computer:
I recently built it from some more-or-less inexpensive parts with intent on overclocking, in order to squeeze the most out of the parts I attained. For example, my Pentium-D 915 cost me $120 (it's since gone down a few dollars as well), and at its current clock of 3.65Ghz, is almost as fast as the $1000 Extreme Edition Pentium D. Theoretically, once I get it to 3.8+, it'll be there (personal goal is 4Ghz+, to join the club). That aside, memory timings.. Like I said, I recently put the computer together and begin overclocking, and as a result, it' still in that sort-of-stable-but-not-really-and-still-tweaking stage. When I first ran the test, I got some ridiculous time of 57 seconds or something. However, when I adjusted my RAM timings (they were set very loose, in order to make sure that any crash my PC may encounter was due to the CPU, so I knew what to fix), I got the time of 36 seconds or whatever it was. The only thing I changed were RAM timings.

RAM plays more of a role in more things than people might think. :)
 
Burnout
Quite wrong.

I dare you to run into your BIOS, change your CAS latency on your memory a tick higher, and run the same benchmark. $20 says it'll be noticeably slower.
Click to expand...
Hmm, I changed the CAS latency from 2.5 to 3 and ran the test again and got the same score - 42 seconds.

I got the time of 36 seconds or whatever it was. The only thing I changed were RAM timings. RAM plays more of a role in more things than people might think. :)
Click to expand...
Strange, changing the latency made no difference in my case.


KM.
 
From what I've read (and I don't claim to be anything close to an expert), the Core 2 Duo CPUs are all the same, the only difference is the clock speed. Is the Extreme edition of the new Intel chip really just a higher clocked version of the same chip that's available for a fraction of the price, i.e. same core, cache, frontside bus, etc.? Just a different clock speed? It seems wildly expensve vs. the low-end Core Duo chips, i.e. something like $900 vs. $200 for just a 50% difference in clock speed?


KM
 
KieranMurphy
Is the Extreme edition of the new Intel chip really just a higher clocked version of the same chip that's available for a fraction of the price, i.e. same core, cache, frontside bus, etc.? Just a different clock speed?
Click to expand...
Well, first off, there are 2 versions of the new Core 2 chips. Those with 4MB of shared L2 cache, and those with 2MB of shared L2 cache. The E6600 on up have 4MB, everything else below is 2MB.

But, between the E6600 and X6800, yes, they are the same. Same everything, except for one critical factor to overclockers. The XE chips come from Intel with an unlocked multiplier. That gives MUCH more options with regards to overclocking. The standard chips, just as with Intels older "Net Burst" chips, could only be adjusted on FSB, no multiplier changes. But now, the XE chips are coming unlocked, so not only can you adjust the FSB, you can adjust the multiplier to help with overclocking.

Hilg
 
I should add, when I said the XE chips were unlocked, I meant unlocked both UP and DOWN.

The X6800 chip comes standard at 266mhz FSB with a stock multiplier of 11x to get its 2.93ghz speed. If you want, and your system components are up to it, you can up that to a 12x multiplier for 3.2ghz, or 13x for 3.47ghz, and so on. But, you can also go lower on the multiplier, but higher on the FSB. So, if your system will overclock to a 330mhz FSB lets say, you can drop that down to a 10x multiplier for better stability, while still keeping a 3.3ghz speed.

The standard Core 2 chips, the E6700 and lower are also available to adjust multiplier, but they can only go lower. The E6700 runs a standard 10x multi, the E6600 a 9x, and so on. The only option at that point for overclocking is to adjust the FSB higher and try a lower ratio multiplier.

That is what gives both of the different chips a certain appeal to different people. The X6800 chips have the benefit of basically "free" overclocking. If your motherboard will do it, bumping up the multiplier only will give you and instant boost. This usually isn't possible without more core voltage, but that is easilly bumped up, and the type of user the X6800 is being used by is well aware of this.

But, on the other hand, the FSB, instead of the ratio, is what generally means more. So, going lower on the ratio, and higher on the FSB will net you more "real world" speed. That is where the other chips benefit. Say you take the stock 266mhz FSB speed on an E6700 up to 300mhz, while dropping the ratio from 9x down to 8x. This will net you the same 2.4ghz speed as the E6600. But, because of the increase in FSB, it will be quite a bit faster. That is why you see a lot of people go for the cheaper E6600 or E6700 instead of the X6800.

Both chips have their intended users. And while both are the same chips with some regards, they are adjusted for speed in much different ways.

Hilg
 
I think that this test is all about clock speed and L2 Cache. Once you've cleared certain factors, such as the need for 8MB system RAM, and more than 512KB Level 2 Cache, then results are driven by clock speed more than anything. I recently replaced my 2.42GHz Athlon 3800+ with a 2.45GHz Athlon x2 4800+, and added 2GB system RAM, and my reduction in time was less than a tenth of a second. The guys with the Intels > 3GHz are killing the slow-clock Athlons.
 

Latest Posts

Back