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

Quote:

---

Originally Posted by jermaink

Apparently the weakness of the Intel platform is it's FSB (or whatever the Hypertransport counterpart is), and it'll constrain the performance of the otherwise impressive Conroe chip.

---

LOL, I'd like to know in what way is the Conroe constrained?? It has faster scores across the board at all levels than any A64/X2/FX/P4/PD/P4EE/etc.. Even with A64's clocked 500Mhz faster (& P4's a 5Ghz) its putting up similar scores.

Yet, it seems to run cooler & more efficiently than any of its competition. And all of this on a PRE-PRE-PRE-Production motherboard, with an ES chip, unsupported BIOS, etc.. only 512mb of ram @ DDR2 533 (not 667, 800, etc..)

Wait until a motherboard comes out than can actual overclock one of these, and they're hitting 3+Ghz at 1333FSB+!

Quote:

---

Originally Posted by chrisf6969

LOL, I'd like to know in what way is the Conroe constrained?? It has faster scores across the board at all levels than any A64/X2/FX/P4/PD/P4EE/etc.. Even with A64's clocked 500Mhz faster (& P4's a 5Ghz) its putting up similar scores.

Yet, it seems to run cooler & more efficiently than any of its competition. And all of this on a PRE-PRE-PRE-Production motherboard, with an ES chip, unsupported BIOS, etc.. only 512mb of ram @ DDR2 533 (not 667, 800, etc..)

Wait until a motherboard comes out than can actual overclock one of these, and they're hitting 3+Ghz at 1333FSB+!

---

well said...

The "OC to 4+" is simply smoke and mirrors. The main focus of Conroe is better performance per watt (something like that but you get the idea). Therefore, if a 2.44 Conroe could only OC to 2.80 (for the sake of argument) at a substantial boost to performance then what difference does it make if it reaches 4+ GHz or not? :slapass:

What we as enthusiasts should be concerned with is not how much it can over clock BUT, the percentage gain vs the percentage over clocked. If we OC correctly by %10 does the performance gained only 10%, 20% or was it only 5%? This is what we must know as it would provide potentially valuable data as to the true benefits of having this CPU (with the right PSU, ram, VC, MB, etc) when increasing it's clock.

For example:
2.44GHz OC successfully by a max on air by 15% therefore, 2.806 or 2.81GHZ now the question is how much performance is gained by a 15% over clock? Let say (for the sake of argument) that 3dmark05 at stock did produced a score of 10500 Marks and Super PI reported 22 second 1 meg test, again all at stock settings. (with all things being equal with 1 gig of ram, 9x5 MB, etc, etc).

Now @ 2.81GHz will the 3dMark05 increase to 12075 Marks or better? Will Super PI decrease it’s 1MB time? These are the questions we need to ask. Remember, we are no longer dealing with P4 where every bit of performance gain in OC counted. And, it wasn't considered to worry about having the same % in performance as the same % OC. We should expect much better out of Conroe and these are the avenues that must be explored to understand Conroe's true potential.

Also, don't forget it still remains to be seen if the final production of Conroe will be locked or unlocked. If it's possible to believe that Intel will allow these beast to remain unlocked as final production CPU then we can quantify with certainty and with accuracy the potentional gains when OC the CPU.

If the CPU is locked...this will not provide the same level of data only because a locked CPU will generate constraints when trying to OC it.

Quote:

---

Originally Posted by Eastcoasthandle

What we as enthusiasts should be concerned with is not how much it can over clock BUT, the percentage gain vs the percentage over clocked.

---

nah mate

what you want is highest MHz in your class..................FX57 CPUs are reach 4Ghz benchable (well kingpins one only) and that's what counts......you don't want a San Diego 3700+ that does 3.8GHz with much higher percentage OC but lower overall MHz level....you want the FX @4GHz

Quote:

---

Originally Posted by chrisf6969

LOL, I'd like to know in what way is the Conroe constrained?? It has faster scores across the board at all levels than any A64/X2/FX/P4/PD/P4EE/etc.. Even with A64's clocked 500Mhz faster (& P4's a 5Ghz) its putting up similar scores.

Yet, it seems to run cooler & more efficiently than any of its competition. And all of this on a PRE-PRE-PRE-Production motherboard, with an ES chip, unsupported BIOS, etc.. only 512mb of ram @ DDR2 533 (not 667, 800, etc..)

Wait until a motherboard comes out than can actual overclock one of these, and they're hitting 3+Ghz at 1333FSB+!

---

If Conroe wasn't FSB limited, why would Intel bump it up to 1333MHz on the XE? If Conroe wasn't FSB limited, why would Intel be working on the new CSI bus?

All your arguement proves is that Conroe rocks so damn much that it is able to outperform the competition even with a cripled, outdated bus. Just imagine what it could do with an integrated mem controller and low latency, higher bandwidth bus. :)

Quote:

---

Originally Posted by dinos22

nah mate

what you want is highest MHz in your class..................FX57 CPUs are reach 4Ghz benchable (well kingpins one only) and that's what counts......you don't want a San Diego 3700+ that does 3.8GHz with much higher percentage OC but lower overall MHz level....you want the FX @4GHz

---

Those who have the money can buy FX chips but normal users want something that is affordable and OC well like 3700+.

Quote:

---

Originally Posted by Nazu

The fsb is the weak link in Conroe as it is the same in Netburst architecture, which cannot handle the full bandwidth of dual-channel DDR2 SDRAM.

---

A Pentium with 266MHz FSB (1066 Quad Pumped) has a theorectical bandwidth of 8,53GB/s. Dual Channel DDR2 @ DDR2-1066 has a theorectical bandwidth of 17GB/s.
Do you wonder why Sandra Sissoft does not go beyond 8GB/s even though using so fast memory and Athlon64 with "slower" memory does?
Don't you see the bottleneck in the text above?

When you pair up a 266MHz FSB (1066QDR) Pentium with dual channel DDR2-533 the memory and the CPU memory interface has the same theorectical bandwidth (8,53GB/s) - now a Sandra Sissoft bandwidth bench shows that 75-80% of the theorectical bandwidth is being used while it proberly show 80-85% with the DDR2-1066.

So taking advantage of DDR2 is a matter of running synchronously FSB and memory and if Conroe is to have 266MHz FSB as default and a moderate overclock is 25% then 333MHz FSB (10,6GB/s) will go "mainstream" while paired with DDR2-667 it would have an effective bandwidth of ~8000MB/s.

Xtreme people with Conroe XE (333MHz FSB default) would only have to overclock 50% to reach the 500MHz FSB that the current fastest DDR2 memory will be able to run 1:1.
But Intel's northbridge/memory controller will proberly limit the FSB before such highst are reached.

But I see why you think netburst is in-effective with DDR2 - Pentium 4 800MHz bus + DDR2-1067 does not give the result one would expect in Sandra Sissoft, but the CPU is a bottleneck at 6,4B/s although the memory would very much like to go above 15GB/s.

Now this chip is sweet, i want to see its overclocking abilities, hopefully will be a bit better than amd's....

Quote:

---

Originally Posted by 0range

Now this chip is sweet, i want to see its overclocking abilities, hopefully will be a bit better than amd's....

---

obviously it would be way better than AMDs...:slap: :p:

You guys need to stop comparing Conroe performance vs. current generation AMD cpu's. It simply isn't fair. Wait till AM2 arrives.....I think all of you will be in for quite the surprise......

Quote:

---

Originally Posted by Nazu

Those who have the money can buy FX chips but normal users want something that is affordable and OC well like 3700+.

---

i agree BUT i wasn't talking about affordability ;)

damn does that mean i'll have to get am2, ddr2, and am2 mobo now :(. if what you said about am2 is true, then it's time for me to save a dollar a day for it :). on another note, you folks should check this out. looks like the chinese got something very interesting in the work here. watch the videos and be amazed.

http://www.thelenovotapes.com/

i want one of these babies :).

Quote:

---

Originally Posted by s7e9h3n

You guys need to stop comparing Conroe performance vs. current generation AMD cpu's. It simply isn't fair. Wait till AM2 arrives.....I think all of you will be in for quite the surprise......

---

true.....but it's very hard for these guys to do that when there is nothing right now that gives even a hint of awesome performance coming from AM2.....as you know current reviews are less than flattering so it's a matter of waiting for them to show us exactly what it is they've got.......you get your ES chips yet :p:

Quote:

---

Originally Posted by dng

http://www.thelenovotapes.com/

i want one of these babies :).

---

it's fake

Quote:

---

Originally Posted by SluRRR

I purchased (past tense) this mobo in march following the IDF benchmarks that were completed on the D975XBX. My point is that the lack of definitive information regarding mobos that support Conroe is not good, and it appears that current owners of this board are out of luck.

---

This is a little dated but,

"Documents from Intel mentioned that the next generation Intel desktop processor, Conroe, will require a platform refresh on existing chipsets. The upcoming Intel G965 and P965 chipsets will support Conroe, but any other Intel chipsets from 975X on down will require a modified VRM and BIOS updates. The VRM, or voltage regulator module, is the component on the motherboard that adjusts the voltages to the CPU. Even if today's chipsets are physically compatible with Conroe, no motherboard today will support the CPU if manufacturers followed the VRM guidelines set forth by Intel when those chipsets were designed".

http://www.dailytech.com/article.aspx?newsid=789

Quote:

---

Originally Posted by frostedflakes

If Conroe wasn't FSB limited, why would Intel bump it up to 1333MHz on the XE? If Conroe wasn't FSB limited, why would Intel be working on the new CSI bus?

All your arguement proves is that Conroe rocks so damn much that it is able to outperform the competition even with a cripled, outdated bus. Just imagine what it could do with an integrated mem controller and low latency, higher bandwidth bus. :)

---

Conroe has FSB of 1033MHz.

Hypertransport rev 1.05 has bandwidth of 200-1400 MHz.

Big difference? Nope.

It doesnt make any difference in single or dual CPU platforms. Where Hypertransport shines is in 4-way, 8-way or more platforms.

Quote:

---

Originally Posted by s7e9h3n

You guys need to stop comparing Conroe performance vs. current generation AMD cpu's. It simply isn't fair. Wait till AM2 arrives.....I think all of you will be in for quite the surprise......

---

AM2 CPUs has been tested on HKEPC and does not even outperform 939 CPUs.

AMD will seriously lag behind Intel in single/dual CPU platforms until AMD can successfully produce 65nm parts in quantity. Until then, sucks to be them. :(

Quote:

---

Originally Posted by vitaminc

Conroe has FSB of 1033MHz.

Hypertransport rev 1.05 has bandwidth of 200-1400 MHz.

Big difference? Nope.

It doesnt make any difference in single or dual CPU platforms. Where Hypertransport shines is in 4-way, 8-way or more platforms.

---

Yeah that is true, the only sector where Intel may have a problem is in the Xeon MP sector, as it will be awhile till Core MicroArchitecture migrates to that space.

Quote:

---

Originally Posted by vitaminc

Conroe has FSB of 1033MHz.

Hypertransport rev 1.05 has bandwidth of 200-1400 MHz.

Big difference? Nope.

It doesnt make any difference in single or dual CPU platforms. Where Hypertransport shines is in 4-way, 8-way or more platforms.

---

Conroe is 1066/1333 (norm/ee) ;)

Quote:

---

Originally Posted by vitaminc

Conroe has FSB of 1033MHz.

Hypertransport rev 1.05 has bandwidth of 200-1400 MHz.

Big difference? Nope.

It doesnt make any difference in single or dual CPU platforms. Where Hypertransport shines is in 4-way, 8-way or more platforms.

---

just a note, conroe's bus speed is 1066MHz effective (or 1333MHz for the EE's)

Hypertransport is meaningless in memory bandwidth discussions. there is NO HT link (serial) between the cpu and ram. its the same kind of connection (parallel) that exists between intel's MCH and the DRAM. ie, for AM2 this is an effective 1.6GHz connection (dual channel DDR2 800)

in cpu to cpu communication as you have said it can make a difference in scaling ;)

not to whine or anything, but have you noticed fugger hasn't updated his conroe progress in like 5 days? Strange stuff.

overclock that pig lets see some 3ghz benches

Quote:

---

Originally Posted by ORCBEAST

Oh well..Intel guyz will be sitting with a faster cheaper system in the end of the day so whine all you like.

ORCBEAST

---

at the moment Intel isn't cheaper the mobo that seems to be the choice is over 300.00 which offsets the CPU price.


just an observance

Quote:

---

Originally Posted by vitaminc

AM2 CPUs has been tested on HKEPC and does not even outperform 939 CPUs.

AMD will seriously lag behind Intel in single/dual CPU platforms until AMD can successfully produce 65nm parts in quantity. Until then, sucks to be them. :(

---

HKEPC's cpu was either a crippled early sample or a fake. The numbers aren't representative of true AM2 performance.

Quote:

---

Originally Posted by iboomalot

at the moment Intel isn't cheaper the mobo that seems to be the choice is over 300.00 which offsets the CPU price.


just an observance

---

I965 Series will launch with Conroe, and should allow you to overclock as well.