Collection of Conroe Data. (Core 2 Duo and Core 2 Extreme!)

Page 14 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Intel could then still launch the Extreme Edition at 2.93Ghz (which I don't really like, but it's the best option). Later, when AMD's 4x4 solution launchs, Intel will have the hardware in place to counter with a 3.33Ghz 1333MHz Extreme Edition. Intel will then have support through their own i975Xs and also through ATI's RD600. The 1333Mhz FSB will also be available to Kentsfield when it ships. I'm hoping AMD's 4x4 approach with the 2 IMCs will force Kentsfield and Cloverton to ship with a 1333MHz FSB. Otherwise, no matter how good Core 2 is, with 4 cores and 2 L2 caches on the same 1066Mhz bus, you will be disappointed.
Click to expand...

Well that is just it. Intel publically stated they are forced to use a 1066FSB for Kentsfield due to electrical constraints on the Front Side Bus. While it is likely there will be bottlenecking, I think it's also a possibility that it might be enough? Considering that so far Conroe can OC up to 5ghz with a perfectly linear performance increase to match, although I'm not sure how much of that is due to the OCed FSB. But a stock Kentsfield should have enough headroom to work with, and OCing would only help more than hurt I believe?

One thing I don't understand about AMD's "4x4" gambit, is the practicality of such a thing. Not only would you have to buy two high-end expensive FX processors, but you'd also have to buy two sets of DDR2 memory to match them. Considering Intel will be pwning AMD in terms of pricing, it seems like they are just making a bad situation worse... then again, unless I missed something I suppose it's all marketing PR/bragging rights anyway...

My thoughts about the 985X chipset, is that Intel will be forced to release one next year when their current 965PE chipset is replaced with a 1,333FSB capable 965PE chipset refresh. Unless Bearlake itself is the replacement for 975X? Bearlake in '07
 
Oh, well whether or not Intel embraces a "1337"FSB I sure will be with a RD600 motherboard, unless the benchmarks disappoint. If Conroe's 1066FSB is fast enough that the chip doesn't bottleneck, then it's fast enough for me. Having a lower FSB will just let me get away with more OCing headroom. :twisted:
 
I wonder if the RD 600 will really do all the things that the OC list is promissing. An average user will not be able to take full advantage. 😳

promising the advertisement sounds however when it is stated that three PCI-Express x16 slots are for three graphics cards.

DDR2-1066 memory! ...When will that be available?
 
One more thing - Intel's new 965 chipset also supposts Quad-Core
no-kidding

check this out
http://www.giga-byte.com/Products/Motherboard/Products_Overview.aspx?ClassValue=Motherboard&ProductID=2295&ProductName=GA-965P-DQ6
Click to expand...

This is true. Kentsfield should be a drop in replacement for Conroe on most of the higher end boards.
 
SFS (Stepless Frequency Selection) allows
- FSB tuning from 100MHz up to 400MHz at 1MHz increment
- Memory tuning from 533MHz up to 1066MHz at 1MHz increment
- PCI Express frequency tuning from 100MHz up to 150MHz at 1MHz increment

Sorry - I missed this paste in the last post. I found the stepless memory tuning interesting. It appears that the memory controler in the 965 is very tuneable. We appear to be rid of fixed FSB/Memory ratios.
 
It's for energy saving/performance profiling. Kinda like a "Super-speedstep".

Systems will be able to downclock and overclock themselves depending on processor/system usage, and how many times the CST has to step in and correct bitmisses in the cache.
 
From the new thread:

Let's remove AMD FX Series from the equation for the moment, and just compare Conroe against Intel's current Top fo the line Pentium 4 965 Extreme Edition, in the hopes of a more mature discussion. Here we have the Intel Core 2 Duo E6700 (2.67GHz) against bit-tech's optimized 3.73GHz Pentium 965 XE.

Credit to bit-tech.net for the new data.

F.E.A.R.: http://www.bit-tech.net/hardware/2006/06/04/intel_conroe_performance_preview/5.html

CPU Limited low Resolution:
lq-fear.png


Realistic Resolution and detail levels:
hq-fear.png



Half Life 2: http://www.bit-tech.net/hardware/2006/06/04/intel_conroe_performance_preview/5.html

CPU Limited low Resolution:
lq-lostcoast.png


Realistic Resolution and detail levels:
hq-lostcoast.png



Now let's get crazy.

Far Cry: http://www.bit-tech.net/hardware/2006/06/04/intel_conroe_performance_preview/6.html

lq-farcry-volcano.png


hq-farcry-research.png



To Quote the bit-tech article: "We think these results speak for themselves... would you have guessed that Conroe is going to be a kick-ass gaming CPU by now? The words 'completely' and 'obliterated' spring to mind... "
Click to expand...
 
It's for energy saving/performance profiling. Kinda like a "Super-speedstep".

Systems will be able to downclock and overclock themselves depending on processor/system usage, and how many times the CST has to step in and correct bitmisses in the cache.
Click to expand...

8O wOw! 8)
 
By the way, I'm not really sure where else to put this so here:

http://www.xtremesystems.org/forums/showthread.php?t=101926

It appears that another Stepping of Conroe is starting to hit the scene. FCG is hitting 3.8GHZ air cooled, at a 1.378 Vcore. That is incredible to say the least. As FCG says: "This is extremely good news for the aircooling an low-end watercooling guys. Its obvious to me that the silicon is getting better....crazy to imagine what it will be like when it comes time to release Conroe for retail. "
 
Honestly, my take -- Intel should embrace the ATI chipset solution and run with it... get a 1333 chip out there.
Click to expand...

Intel has more stringent and conservative methods so what other companies like ATI and Nvidia thinks its fine running 1333FSB Intel may not.
 
Update By TAM

A slightly improved modPI 1M 10.140s @5013.1Mhz this weekend.

Time: 10.140s (Super PI / mod1.4)
Clock: 5013.1MHz (fsb417.8MHz x12)
CPU: Core 2 Extreme X6800 : (2.93G/266x11/FSB 1066/L2 4MB)
M/B: Intel D975XBXLKR-rev302 modded (bios:1073)
CPU Cooling: LN2 (Pro4 + V-TEC ARC Bed Rev.3.4)
Memory: Corsair PC2-5400UL(v1.3) 512MB x2
VGA: ATI rageXL (PCI)
HDD: IBM DTLA-307020
Power Supply：Zippy-460w
Vcore: 1.74v, Vdimm: 2.55v, Vio: 3.65V
OS: Windows Server 2003
MEM: 4:5, CL4-3-3-2

X6800-5013_10s140.gif


CPU-Z_X6800-5018.gif
 
What do you guys think of the poor showing in multithreaded tests in PCMark05? something wrong with the software, ie not optimized, or something
to do with the chip? has anyone posted any ps for quake 4 with conroe and MT?
 
I am not disagreeing with you, you are right -- Intel puts much more conservative specs against what they are capable of doing... this has been historical practice.

What I am saying is that Conroe (Core 2) obviously can take the FSB clocked extra-ordinarily high so this is not the FSB limit -- the question I would like to answer then, what is? Is it the chipset does not come up to snuff? Is it the actual FSB lanes and cross talk? If ATI is willing to ship out a chipset that is 1333 and as high as 1600, then why not partner up and take advantage.... it won't happen, I am pretty sure..but I am just speculating...
Click to expand...

The reality though is, if Intel sees its the limit they'll accept, it is the limit. The amount of validation they take their products before shipping is one of the highest(if not the highest) in the industry. This is why it makes overclocked results really tell little about what Intel is gonna produce, cause they don't care about that, they care about how stable it is(in their own terms).

Pentium D, which is basically two Prescott 1M cores in one dcm package, took one year from beginning to shipping, mostly which would be validation process.
 
Not sure what to make of it, it rears itself in the compression test which in other benches is not the huge strenth in the Core 2 line -- it could be optimizations used by the compressor in terms of threading, it could be a problem with the cache (thrashing?)... was the Presler using hyperthreading? If presler was had hyperthreading turned on, then this would give it more of an advantage if the application is capable of spawning more than 2 threads.

The other two tests are within 1-3%, eh, not much. The question becomes is there a different threading approach between multi-core and hyperthreading, at this point it is difficult to be sure. It is cause for debate though, will the L2 cache sharing be a problem in the future?
Click to expand...

Performance impact of cache thrashing(which seems to be BS from people against shared caches say when in fact all high performance processors will have shared cache), compared to how Core 2 outpaces other CPU, is insignificant. In the days where having a new CPU that's faster than older CPU was all people could have wished for, couple % loss for having poor caches was significant. If anything, shared caches are better for multiple cores than dedicated, due to being able to have dynamic allocation and such.

I assume the problem has to do with Core's hunger for memory bandwidth. The higher performing the CPU is, more memory bandwidth its gonna require it to be taking the advantage.

If you head over to Xtremesystems and read their tests, one tests show that Core 2's advantage over Core Duo increases as clock is increased, which in this case is done by overclocking. As memory bandwidth increases by being overclocked, Core 2 takes better advantage over Core Duo with the extra memory bandwidth.

This thing has been said by Charlie at the Inquirer, where he said the CPU will be limited by the platform. He thinks Kentsfield/Clovertown, the quad core variant will have almost no advantages over Woodcrest because of the lack of memory bandwidth, or the increase will be so small, that it won't be worth it.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom