an honest question about the Conroe vs. FX-60 tests...?

[AthlonAlien]

How were the CPU 'overclocks' chosen? I mean, the Conroe was running at a 23.07% overlock, whereas, the FX-60 was running at a 7.69% overclock?

Let's put aside some of the responses like:

1. The FX's multi was raised, so no HTT overclocking was done, hence making the FX slightly less efficient.

2. AMD runs slightly faster on Nvidia boards (and SLI) than ATI boards (and CrossFire), thus giving a small performance hit.

3. Wait until AM2 comes out... it may not be a MAJOR performance increase, but it will still increase performance some.

I am just talking about the CPU overclocks right now... Are all Conroe's supposed to overclock this high on air or was this just a cherry-picked chip? It just seems odd that a 2.167GHz part was used and overclocked to 2.667GHz for the test. Are there any faster 'stock clocked' Conroe's in the works? I just see this as a little confusing... I mean, if a 2.53GHz part was used and then overclocked, it would make more sense to me. I am just wondering if the 2.667GHz overclock was a "hand picked" chip to make the comparison look a little better (since we know several FX-60's can do more than 2.8GHz, a lot can do 3.0GHz+ on air)?

 

[Cooler]

Originally posted by: AthlonAlien
How were the CPU 'overclocks' chosen? I mean, the Conroe was running at a 23.07% overlock, whereas, the FX-60 was running at a 7.69% overclock?

Let's put aside some of the responses like:

1. The FX's multi was raised, so no HTT overclocking was done, hence making the FX slightly less efficient.

2. AMD runs slightly faster on Nvidia boards (and SLI) than ATI boards (and CrossFire), thus giving a small performance hit.

3. Wait until AM2 comes out... it may not be a MAJOR performance increase, but it will still increase performance some.

I am just talking about the CPU overclocks right now... Are all Conroe's supposed to overclock this high on air or was this just a cherry-picked chip? It just seems odd that a 2.167GHz part was used and overclocked to 2.667GHz for the test. Are there any faster 'stock clocked' Conroe's in the works? I just see this as a little confusing... I mean, if a 2.53GHz part was used and then overclocked, it would make more sense to me. I am just wondering if the 2.667GHz overclock was a "hand picked" chip to make the comparison look a little better (since we know several FX-60's can do more than 2.8GHz, a lot can do 3.0GHz+ on air)?

First they were both running ATI cards so this should have almost no impact on performance and the latest ATI mobo are just as good or better then nfroce(for both ATI systems and Intel).

Second not all FX-60 can even hit 3.0 GHz on air. The Intel Chip was OCed to the Speed it will be when launched at price of $530

Third they are comparing a $999 to $530 chip
Intel $999 chip will be at 3.33 GHz with 1333 FSB
Most Hight end intel mobos today can get close to that fsb so low end OCing should be good.


I would say Intel did a fair comparison with AMD.

Remeber Intels R&D Buget Is much bigger then AMD'S .

 

[bunnyfubbles]

Intel $999 chip will be at 3.33 GHz with 1333 FSB

Not any time soon that's for sure. They might have a 2.8 or 3.0 chip, and it will most likely be a lot more than $1000 based on past EE pricing. Besides, such a chip (if the numbers are honost) would have the performance justifying an 'extreme' price. But there's no way they'd have an EE @ 3.33GHz with their highest non EE being a mere 2.67GHz.

 

[Marmion]

Umm, they have Woodcrest at 3Ghz using a 1333Mhz FSB. It is very much possible that the Conroe EE will be greater than or equal to the top Woodcrest server chip. I think though that the EE will come in Q4. The rest of the Conroe line comes in Q3. I didn't know they overclocked the Conroe, but that is the top of the mainstream parts (2.67Ghz, 1066Mhz FSB)

 

[Cooler]

Originally posted by: bunnyfubbles

Intel $999 chip will be at 3.33 GHz with 1333 FSB

Not any time soon that's for sure. They might have a 2.8 or 3.0 chip, and it will most likely be a lot more than $1000 based on past EE pricing. Besides, such a chip (if the numbers are honost) would have the performance justifying an 'extreme' price. But there's no way they'd have an EE @ 3.33GHz with their highest non EE being a mere 2.67GHz.

They will have a 2.93 Ghz chip coming for about $650 in Q3/Q4.


Im not sure if it will be on a 1333 FSB or not.

 

[swtethan]

Originally posted by: Cooler

Originally posted by: bunnyfubbles

Intel $999 chip will be at 3.33 GHz with 1333 FSB

Not any time soon that's for sure. They might have a 2.8 or 3.0 chip, and it will most likely be a lot more than $1000 based on past EE pricing. Besides, such a chip (if the numbers are honost) would have the performance justifying an 'extreme' price. But there's no way they'd have an EE @ 3.33GHz with their highest non EE being a mere 2.67GHz.

They will have a 2.93 Ghz chip coming for about $650 in Q3/Q4.


Im not sure if it will be on a 1333 FSB or not.

its 1337 fsb

 

[bjc112]

Reguardless of the minor differences in hardware.. Conroe looks to smack AMD around for a Quarter or Two.

 

[Skott]

Personally I like both company's products. Its my opinion though that Intel with its bigger resources/budgets and longer experience should be able to make a chip better than AMD's. Nor should it be suprising to anyone when they do make a better product.

 

[Markfw]

People !!!! Its not even out yet, and nothing but Intel provided systems to bench. When it really comes out, and 5 reputable sites review it THEN we can talk. Until then its all guesswork and conjecture.

KNOCK OFF THE GUESSING AND ARGUMENTS !!!!!

 

[BrownTown]

where do you get the 23.07% overclock number?, this is the stock speed of the processor, its a 0% overclock. And talking about overclocking a processor that isnt even out yet makes no sense anyways becasue the stock speeds arent even set in stone to begin with. There will be Core processors running at 3G at the very least by the end of the year, and likely the 3.3G XE. Intel showed off 3G Woodcrests at IDF, so clearly they can make chipsat that speed, (and this is 4 months before launch, so yields will only go up).

The test was simple, an AMD processor running at its stock spped for Q3 2006 vs a Conroe running at stock speed for Q3 2006. You can argue the BIOS issues, DDR2, or XFire performance, but the clockspeeds are exactly what AMD and Inte'ls roadmaps show for Conroe launch.

 

[PentiumIV]

From my understanding of the company the situation is as follows:

1. There were absolutely no intention to inflate Merom (this is how the design is called internally, Conroe is a desktop version) numbers vs. AMD. Apparently, ATI chipset for AMD was chosen to allow the same graphics cards on both systems.

2. A reasonable effort was done to optimize AMD system vs. the Conroe. The AMD system used about the fastest DRAM possible (CAS2 1T DDR-400) vs. quite ordinary DDR2-667 memory on the Conroe side, thus giving AMD a latency advantage.
As for the latest BIOS, etc.... we didn't have guys on-site to squeeze the last FPS from the AMD machine. The same can be said about Conroe.

3. To get the 20% performance boost one should do a very serious redesign of the core.
It would take at least 2 years to do so (design, circuit work, layout, validation, etc.) + 1 year of silicon debug. If AMD would do so, we shoud now hear of the new silicon in the AMD labs. So, far, this doesn't happen.

 

[Sunbird]

I'm gonna like this, cause AMD will have to come down in price a lot and we can have $70 overclocking monsters again

Welcome back good old days.

 

[Sable]

I'm thinking of a river in Egypt. Guess which one.

 

[Vegitto]

Originally posted by: Sable
I'm thinking of a river in Egypt. Guess which one.

The Nile?

 

[Duvie]

Originally posted by: PentiumIV
From my understanding of the company the situation is as follows:

1. There were absolutely no intention to inflate Merom (this is how the design is called internally, Conroe is a desktop version) numbers vs. AMD. Apparently, ATI chipset for AMD was chosen to allow the same graphics cards on both systems.

2. A reasonable effort was done to optimize AMD system vs. the Conroe. The AMD system used about the fastest DRAM possible (CAS2 1T DDR-400) vs. quite ordinary DDR2-667 memory on the Conroe side, thus giving AMD a latency advantage.
As for the latest BIOS, etc.... we didn't have guys on-site to squeeze the last FPS from the AMD machine. The same can be said about Conroe.

3. To get the 20% performance boost one should do a very serious redesign of the core.
It would take at least 2 years to do so (design, circuit work, layout, validation, etc.) + 1 year of silicon debug. If AMD would do so, we shoud now hear of the new silicon in the AMD labs. So, far, this doesn't happen.

1) I agree

2) Any moron who has not been a fanboy and played with both architectures of recent knows the AMDs are not bandwidth limited but much more sensitive to cas latency...also know that INtels are extremely bandwidth limited and do not take nearly as much of a hit with latency.

3)I agree possibly...However in the past many upgrades have come in with 5-10% increases with little or no pre warning...Dual core came much faster to the market then we thought...Most of the things I have listed (in anotheer thread as possibilities) have all been rumopred at one time or not over the last year...So who's to say in another 6-9 months it wont be ready???

Until I actually know what your credentials are taking your opinion is as worth as much to me as farting!!!!...The old "opinions are like a55holes"...you know the rest....


More and more my gut says 65nm part quad cores may be rushed up to launch at time of conroe and steal some of INtels thunder...IE like they tried to do with those sh^tty smithfields during AMD's dual core launch...That is a perfect example of rushing a make shift architecture to the market....

If you look at it Intels server chips have been lagging behind their desktops, whereas AMD has focused on the opposite...quad cores may become extremely bandwdith limited so DDR2 (667-800) or even DDR3 controllers will be a must to feed the cores....


I for one and my uses find a quad core more important then a faster single core or dual core...

 

[PentiumIV]

Originally posted by: Duvie
3)I agree possibly...However in the past many upgrades have come in with 5-10% increases with little or no pre warning...Dual core came much faster to the market then we thought...Most of the things I have listed (in anotheer thread as possibilities) have all been rumopred at one time or not over the last year...So who's to say in another 6-9 months it wont be ready???

Until I actually know what your credentials are taking your opinion is as worth as much to me as farting!!!!...The old "opinions are like a55holes"...you know the rest....


More and more my gut says 65nm part quad cores may be rushed up to launch at time of conroe and steal some of INtels thunder...IE like they tried to do with those sh^tty smithfields during AMD's dual core launch...That is a perfect example of rushing a make shift architecture to the market....

If you look at it Intels server chips have been lagging behind their desktops, whereas AMD has focused on the opposite...quad cores may become extremely bandwdith limited so DDR2 (667-800) or even DDR3 controllers will be a must to feed the cores....


I for one and my uses find a quad core more important then a faster single core or dual core...

First of all, I design microprocessors (part of them) for a living.I validated PentiumIV (hence the nick), and then switched to some more adventurous things So, I more-or-less understand, how the things are done ... To get +20% integer performance requires a very significant redesign. One can stick 2 cores on the same die without major redesign (if the cores are only connected via the external bus) on a quite short notice.

However, this will not get you +20% single-threaded integer performance advantage!
To do so you should do a serious internal redesign. Such redesign would require re-floorplanning, re-layout, re-circuit, re-validation (complete IA-32 re-validation !)
All this takes time!