Coolaler @ 5.26GHz Sub 10 second SuperPi_1M!!!

[charlie]

i wonder how many years untill sub 1 second...

I'd say 3 years...

[NiCKE^]

I'd say 3 years...

When is next intel architecture to be released? :P

[caater]

When is next intel architecture to be released? :P

2008 - nehalem

[tys90]

Amazing results...well done again Coolaler

[charlie]

I even think we'll see 7.999s by end of year.

[mursaat]

@charlie: seeing it doesn't scale linearly, for sub 8sec you'll need some insane mhz. 7ghz? more? I can't find the magnific table found in other threads, there are tons of Conroe posts

That, counting that Conroe and Kentsfield benefit more from disabling core(s) to get mhz than by calculating the 1M with 2 or 4 cores, right?.

[NiCKE^]

Maybe we need a multi core SuperPi

[charlie]

@charlie: seeing it doesn't scale linearly, for sub 8sec you'll need some insane mhz. 7ghz? more? I can't find the magnific table found in other threads, there are tons of Conroe posts

That, counting that Conroe and Kentsfield benefit more from disabling core(s) to get mhz than by calculating the 1M with 2 or 4 cores, right?.

I bet we'll see 6gHz LN2 Conroe/Kentsfield by the end of the year.

[TASOS]

Congrats
Amazing results.

[realsmasher]

I'd say 3 years...

3 years ?

sub 1 would be 10 times faster.

if we look when the last cpu was 10 times slower then today, it would be like 8 or 10 years.

[coolaler]

[kyosen]

Congratulations new 2M WR, Coolaler
...Now you are trying sub 10minutes for 32M, aren't you?

My latest 2M time is 26.813s
http://www.oohashi.jp/c-board/file/2...E6700_4582.png
I'll try it again with XMS2-6400C3

[Langley]

Congratulation!!!!!!!

[mursaat]

3 years ?

sub 1 would be 10 times faster.

if we look when the last cpu was 10 times slower then today, it would be like 8 or 10 years.

Hmm maybe in mhz, but in computing power is much less. In 1996 we had Pentium I 200mhz, P2 was released on May 7, 1997. I think Conroe is many more times than 10 "faster" than a Pentium 200. I would say it's 10 times more powerful, for not use "faster" than a Coppermine 1000? It was released on May 2000. Hard to tell, as each app/game will show different benefits from new features.

[coolaler]

Congratulations new 2M WR, Coolaler
...Now you are trying sub 10minutes for 32M, aren't you?

My latest 2M time is 26.813s
http://www.oohashi.jp/c-board/file/2...E6700_4582.png
I'll try it again with XMS2-6400C3

u r awosome@32mpi
but that's mission impossiable to me
my bad module need higher clocks

[Jonny5isalivetm]

Very Impressive CPU thanks for sharing Coolaler.




J5

[***Deimos***]

I even think we'll see 7.999s by end of year.

LOL Charlie.. you crack me up... There is a Conroe SuperPi1M scaling graph around somewhere... try and extrapolate where 8s would be (hint: its way off the scale).

And dont think for a second that the almighty 45nm will magically effortlessly double frequencies.

Technology has gone quite far, hasn't it?

A 1,000,000 digit calculation in under 9 seconds.. by a home desktop processor!

I know it's clocked -- but hell, as long as it's possible I stay amazed, good on ya coolaler

technically, its sub 10 sec. And calling it a "home desktop processor" is misleading... considering all the LN2 and mad overclocking skills that I'm sure went into this project.

I think 1M SuperPi should be kept. Later in the future it'll be funny to see 3~ 5 sec 1M. The superPi has three decimal digits so it'll be fine.

Btw, Coolaler, I gave you Five stars for this thread.

SuperPI 1M has long outlived it usefullness. Windows and low-level timers have finite resolution, and with <10s results you get "grainy" results. Besides, its hardly more indicative of stability than a CPUZ screenshot. I think we should skip ahead to SuperPI 4M... nice ~1min times with Conroe. Also a bit more relevant, since with ~33MB of memory used, its not just a test of Conroe's L2 cache.

Congrats Coolaler, REALLY nice. It was just a matter of time and a matter of having godlike OCing skils



Where did P4s begin? Did we go to 7 GHz when they got out? Nope, started with 5 GHz, then with LGA came 6GHz+ and at the end with kyosen's 'help' we got 7,6 GHz. It's the same as sub-10s PI run. Just a matter of time. Maybe we'll need a new stepping but i'm confident it will be done.

well that's obvious. Ofcourse its just a matter of time for Conroe architecture to scale to higher frequencies. But, I wouldn't count on a new stepping performing miracles.

[***Deimos***]

I'm still in shock...the fact that he broke into the 9's is amazing enough, but how far he broke into it is even more impressive. I was expecting the first sub 10s to be like a 9.99x, or atleast for sure in the 9.9xx, but to run a 9.828 is just crazy nice!

That is EXACTLY what impressed me the most too! He didn't just sneak close to or barely under 10s. Taking into account the exponential nature of the score, he absolutely demolished it with 9.828s. And what REALLY REALLY impressed me is how many runs he got, and how quickly he follows up with 3D benchmarks...

spectacular

somehow I'm reminded of:

"Once again I stand atop the broken bodies of my enemies... Victorious but not unscathed. The Earth-borne Directorate has been destroyed, and the Overmind lies dead and trampled beneath the ashes of Char. As for my unlikely allies, I think that I shall allow them a reprieve, for in time I will seek to test their resolve, and their strengths. They will all be mine in the end, for I am the Queen of Blades. None shall ever dispute my rule again." - Infested Kerrigan on Char

[coolaler]

5050mhz@spi2m 25.828s



5000mhz@spi4m 1m02s406

[railer]

Damn you never stop man That silicon you got is very nice. Some AM3 with 5.2G ?

[Punisher!]

LOL your 2MB is still faster than my 1!

[MTM]

I couldn't find that other Conroe scaling graph that someone made, but this is a quick/rough version I just made to give some idea of what is needed:

[***Deimos***]

I even think we'll see 7.999s by end of year.

Look at the graph above Charlie.
For <8s you would need >6.5Ghz (assuming FSB and DDR2 dont hold you back). Surely its going to happen EVENTUALLY..but I wouldn't pin my hopes on 2006.

Afterall, all these overclocking records and Intel still hasn't officially released Conroe to the public!

Coolaler, thanks for the SP4M screenie.. that 1.02.406 really hit the spot.. ever so close

[coolaler]

spi 1m result@today just tie the WR@9.828s(5266mhz last time)
but this time is 5200mhz@9.828s



last time@9.828s

[rustynator]

You really are producing some awesome OC results coolaler! With those extra 66MHz maybe you could push that WR to 9.6s??