Conroe 2.4Ghz on 965G mobo, brief test...

[s7e9h3n]

Placing X2 benchmarks in a Conroe thread makes it harder for people to find out how an X2 performs...

Hmmmm........

BTW, this thread won't be updated by Victor until the ~21st of this month anyways as he's away on business.....

[savantu]

Where did you get that info from?

I had a dream last night...

The stuff is like this :

Intel Israel designed Banias( 1999 ) ,then Dothan ( 2001) and started Yonah.
Intel Santa Clara took over Yonah and Intel Israel was given a new job : Merom ( all this happened in early 2002 )

Intel Oregon ( which designed the P4 ) had its Nehalem project based on Netburst killed in late 2003 , together with Tejas.In early 2004 they started a new Nehalem , a brand new microarch which is set to arrive in late 2007/early 2008 and replace Merom 2 ( 45nm )

In late 2005 Intel Israel started to work on another new microarch called Gilo which will arrive around late 2009-2010 timeframe and supplant Nehalem.

Basically Intel will have 2 design bureaus fighting each other and bring new stuff every 2 years.I hope AMD can sustain the pace , altough they look very confused right now ( killed K9 , K10 , stopgap K8L, new microarch in 2007 which I hope it is not K8L ) not knowing which to target : Merom 2 on 45nm or Nehalem ?

[Iconyu]

No pons intended.
What I dont understand is-why Intel giving conroe such a short life.
According to latest Intel's news on roadmap, it appears that 1Q 0f 2007 we will see yet another change. according to this article a newer architecture will be introduced in early 2007 leaving "core architecture" behind. The newer architecture will have 2m L2 instead of 4m as is conroe. This is how I read it: conroe appears to have short life if this roadmap is true.
But why? "core architecture has'nt even hit the market.
http://www.tgdaily.com/2006/04/14/intel_quad_core/

If this new architechture has an internal memory controller then it won't need 4mb L2. L2 cache makes up for high external memory latencies, and I think Conroe still hurts when it hits system ram, the fsb bottle necks it.
But 4mb shared should be more than enough to overcome this problem in the short term, and the transistors are cheap for Intel.

[savantu]

...L2 cache makes up for high external memory latencies..

More cache always helps , especially in commercial aplication and for servers where you need scaling.

How much it helps when you increase its size depends on how well you implemented it.

For example , it is possible that AMD's poor results with DDR2 are a combination of the modified IMC* ( 2x8 wide data buses vs 1x16 on 939 ) and its L2 line size.

*It isn't yet clear they took the 2x8 way like Intel 9xx, but it looks like it.

[Iconyu]

I came to this conclusion, after witnessing AMD's tactics with their switch to socket 939. Their 512k's did indeed perform close to the 1mb's on the 754 socket. The performance gap was very almost identical to 512k vs 1mb on the 754 at the same clocks.
I'm not saying I'm right, after all I joined this forum to learn more, and it would be foolish of me not to take each post with a open mind. There's a great depth of knowledge around here that I'd like dip into.

[Rovtar]

this thread is 99% spammed...

why the are you talking about amd in intel section if noone didn't ask for it..

open new thread "flame one amd vs intel" and post there

[P8baller07]

Look at the voltage on that beast, 3volts.

[savantu]

Look at the voltage on that beast, 3volts.

Yeah , and it pulls 119Amps.Can you believe how hot it must be ?

[Iconyu]

1.2 volts, not 3 volts.

[Haltech]

this thread is 99% spammed...

why the are you talking about amd in intel section if noone didn't ask for it..

open new thread "flame one amd vs intel" and post there

And your purpose for even posting this bs was what again? You are grouped into the small few that have ruined this thread.

Seriously mods, erase the bs posts in here or just close it.

[njkid32]

Ok guys this thread has gone waaay off topic.. So now its closed..