K10 Barcelona 50% faster than Kentsfield ?

[jpeyton]

Originally posted by: Roy2001

Originally posted by: jpeyton
Looks like AMD will make a 1-2 punch and take the performance lead in both CPU and GPU markets this summer :thumbsup:

I suspect that won't happen. Clock to clock, Barcelona could be as good as or even beat C2Q, but Penry would crush it.

Penryn is launching this summer?

 

[shiznit]

this fall supposedly. as big a fan of amd as i am (my last intel was celeron 300a), i think it's going to be hard for me to buy a k10 knowing that 45nm penryn is right around the corner, probably with higher clock than k10. now if amd allows people to have a clear upgrade path from 65nm k10 to 45nm k10 on the same motherboard, and provided that k10 outperforms intel, then i might stick with amd. i dont wanna buy k10+mobo only to have to replace the mobo when 45nm k10 comes out.

 

[coldpower27]

Originally posted by: NoobyDoo
Fudzilla Link

We don't have many numbers but at least in specfp_rate2000 Barcelona ends up a bit less than 50 percent faster than Intel quad core codenamed Kentsfield.
We are quite sure that this is the best score but it definitely sounds impressive. AMD chaps are very confident that K10 marchitecture and native Quad core is the way to go and Intel will probably have to pay the price for duct taping its chips.

Title is misleading. Doesn't matter if Barcelona is faster then Kentsfield as these 2 don't compete, Bareclona being faster then Clovertown is the proper one and isn't as surprising as SpecFP does better the more memory bandwidth you have not to mention in FP calculations K8 is pretty much equal to Core clock for clock. So it's completely playing to AMD's strengths. K10 will further increase AMD's FP advantage as it adds even more FP execution resources.

 

[coldpower27]

Originally posted by: shiznit
this fall supposedly. as big a fan of amd as i am (my last intel was celeron 300a), i think it's going to be hard for me to buy a k10 knowing that 45nm penryn is right around the corner, probably with higher clock than k10. now if amd allows people to have a clear upgrade path from 65nm k10 to 45nm k10 on the same motherboard, and provided that k10 outperforms intel, then i might stick with amd. i dont wanna buy k10+mobo only to have to replace the mobo when 45nm k10 comes out.

Agena won't arrive till sometime in Q4 this year so were looking at roughly 1 Quarter advantage to AMD as the mainstream Yorkfield Quad's won't get here till Q1 2008 for desktops with the Mobile and Server DP launches happening in Q4.

 

[jpeyton]

Originally posted by: shiznit
this fall supposedly. as big a fan of amd as i am (my last intel was celeron 300a), i think it's going to be hard for me to buy a k10 knowing that 45nm penryn is right around the corner, probably with higher clock than k10. now if amd allows people to have a clear upgrade path from 65nm k10 to 45nm k10 on the same motherboard, and provided that k10 outperforms intel, then i might stick with amd. i dont wanna buy k10+mobo only to have to replace the mobo when 45nm k10 comes out.

If anything, Intel is far more notorious for requiring motherboard/chipset changes between CPU families.

AMD has thus far guaranteed that K10 will require a BIOS update in socket-compatible motherboards. Die-shrinks on the Athlon 64 X2 CPUs have been handled in the same manner; with BIOS updates on socket-compatible motherboards.

 

[rudder]

Originally posted by: destrekor
Dammit. This better not turn out to be true in real-world applications. I just made the switch to intel. Hopefully Intel redesigns the marchitecture for their quad-cores, Kentsfield is definitely not the right way to go about it. 2 chips sammiched together? boo
I don't plan on going quad-core till it's actually taken advantage of in applications, specifically in gaming, since hell, dual-core is only soon to be taken advantage of in gaming.

Well now that RAM is cheap it is easier to flip flop. When the core2duo came out I bit the bullet and got one even thought I had to pay a premium for DDR2 when my DDR was working fine for me.

 

[RichUK]

Lol. Specfp is a benchmark designed to test floating point performance. No wonder these results are in AMD?s favour.

The K10 is supposed to be an FP monster, hence the supposed superior performance relating to that article. This, however, doesn?t equate to processing performance in its entirety.

This will be about the only ?benchmark? that will be in AMD?s favour.

It?s meaningless.

Once Intel starts rolling out 45nm silicon, it will severely cripple AMD in the market.

You must understand how invaluable a die shrink actually is. It?ll breathe yet even more life into the already very successful Core 2 architecture. Intel is already touting possible level 2 caches in the region 6MB ? 12MB, although the extra Cache is relative in performance, it really shows how advanced Intel is becoming with its new manufacturing process and architectures.

I can?t say it?s looking at all good for AMD.

It looks like this AMD/ATI merger has had a real negative impact on business. The same thing is happening with R600, they just can?t keep up with the pace, and before they know it they?ll be too far behind.

 

[erwos]

Whomever puts out the fastest quad-core at the $300-$400 range is the one getting my money in Q4. I could care less who it is.

That said, 50% clock for clock faster seems impossible. Maybe in some very, very specialized cases, I suppose, but unless that specialized case is something like "H.264 encoding" or "database transactions", I doubt it'll matter.

 

[jpeyton]

Looking at the Penryn performance preview on Anandtech's main page, they concluded Penryn will provide a 5-10% performance clock-for-clock.

Hardly a bombshell.

 

[MotF Bane]

That article is definitely meaningless. How about AMD stops skulking about, and handing out a few vague wordings? Let's see some real benchmarks and concrete numbers to back this up. I'd still have to bet on Yorkfield beating out Barcelona by a decent margin, irregardless of being a non-native quad core versus a native quad core. Besides, Nehalem is coming in what? Two years? Enjoy your two years AMD.

However, ATI is going to beat out nVidia this summer. Welcome aboard, 300 stream processors, GDDR4, R600 80nm, and R650 65nm.

 

[allies]

Originally posted by: MotF Bane
That article is definitely meaningless. How about AMD stops skulking about, and handing out a few vague wordings? Let's see some real benchmarks and concrete numbers to back this up. I'd still have to bet on Yorkfield beating out Barcelona by a decent margin, irregardless of being a non-native quad core versus a native quad core. Besides, Nehalem is coming in what? Two years? Enjoy your two years AMD.

However, ATI is going to beat out nVidia this summer. Welcome aboard, 300 stream processors, GDDR4, R600 80nm, and R650 65nm.

You think AMD doesn't have plans for two years from now? Also, if they're ahead for 2 more years until Nehalem...that means out of the past 10ish years (two years from now) AMD will have been in front for the majority of them.

 

[MotF Bane]

...AMD is struggling for right now against Conroe, and they won't be ahead for the next two years against Penryn. Therefore, Nehalem takes away the last few edges that AMD will have had.

 

[coldpower27]

Originally posted by: jpeyton
Looking at the Penryn performance preview on Anandtech's main page, they concluded Penryn will provide a 5-10% performance clock-for-clock.

Hardly a bombshell.

No but combined with the increased FSB, larger LV2 cache and increased clockspeeds within the same thermal envelopes, made possible with the 45nm process, they are a nice welcome boost.

And unlike the SpecFP thing AMD keeps throwing around, these improvements are mroe universal,a and help in a wide variety of applications.

 

[SexyK]

Originally posted by: coldpower27

Originally posted by: jpeyton
Looking at the Penryn performance preview on Anandtech's main page, they concluded Penryn will provide a 5-10% performance clock-for-clock.

Hardly a bombshell.

No but combined with the increased FSB, larger LV2 cache and increased clockspeeds within the same thermal envelopes, made possible with the 45nm process, they are a nice welcome boost.

And unlike the SpecFP thing AMD keeps throwing around, these improvements are mroe universal,a and help in a wide variety of applications.

Agreed. People are quick to poo-poo a 10% clock for clock increase (actually the tests show much much more than that in some scenarios) while they conveniently ignore the higher clocks that will be possible on Penryn when compared to Conroe (and Barcelona). Even if Barcelona is 50% faster than Conroe clock for clock (which is most likely not even close to true across the board), when you add in the improvements in Penryn and the higher clock speed, that advantage disappears pretty quickly.

 

[coldpower27]

Originally posted by: allies

Originally posted by: MotF Bane
That article is definitely meaningless. How about AMD stops skulking about, and handing out a few vague wordings? Let's see some real benchmarks and concrete numbers to back this up. I'd still have to bet on Yorkfield beating out Barcelona by a decent margin, irregardless of being a non-native quad core versus a native quad core. Besides, Nehalem is coming in what? Two years? Enjoy your two years AMD.

However, ATI is going to beat out nVidia this summer. Welcome aboard, 300 stream processors, GDDR4, R600 80nm, and R650 65nm.

You think AMD doesn't have plans for two years from now? Also, if they're ahead for 2 more years until Nehalem...that means out of the past 10ish years (two years from now) AMD will have been in front for the majority of them.

There was a miscalculation there, it depends on what you mean by ahead. AMD will likely be good in the 2P and 4P Sectors where more of AMD's advantages come into play. As well where the 4P sector won't receive the 45nm refresh till sometime in 2k8

On desktops it's going to be much less clear,

Since when has AMD in front of the majority of the last 10 years? Or are you talking in front in terms of gaming?

Pentium 3 and Athlon were even blow for blow till Q4 2000 more or less, and since 10 years includes 1997 Intel was ahead when it was K6 vs Pentium 2/3. Athlon wasn't released till Mid 1999 and then things were more even between the 2. AMD could be argued to be ahead when you reach Willamette and Early Northwoods, but things even out with Northwood B and Northwood C took the lead for awhile. then AMD release Athlon 64's and thing were even again, and it wasn't till Athlon 64x2 that AMD takes the lead, which was then broken with Core 2 and will hold it till Q4 2007 when AMD releases desktop based K10's.

 

[Regs]

AMD "announced" Shanghia for 08' 45nm. So it's not like AMD is going to sit on their hands...

But then again who the heck knows anymore what AMD plans to do?

 

[allies]

Originally posted by: coldpower27
There was a miscalculation there, it depends on what you mean by ahead. AMD will likely be good in the 2P and 4P Sectors where more of AMD's advantages come into play. As well where the 4P sector won't receive the 45nm refresh till sometime in 2k8

On desktops it's going to be much less clear,

Since when has AMD in front of the majority of the last 10 years? Or are you talking in front in terms of gaming?

Pentium 3 and Athlon were even blow for blow till Q4 2000 more or less, and since 10 years includes 1997 Intel was ahead when it was K6 vs Pentium 2/3. Athlon wasn't released till Mid 1999 and then things were more even between the 2. AMD could be argued to be ahead when you reach Willamette and Early Northwoods, but things even out with Northwood B and Northwood C took the lead for awhile. then AMD release Athlon 64's and thing were even again, and it wasn't till Athlon 64x2 that AMD takes the lead, which was then broken with Core 2 and will hold it till Q4 2007 when AMD releases desktop based K10's.

No, my 10 years are assumed (from the other poster) that Nehalem will be the next intel chip that leapfrogs Barcelona. That means it goes to 1999. Athlon was neck and neck, if not faster since then, in the majority of benchmarks. P4 only held its own in media encoding.

 

[coldpower27]

Originally posted by: allies

Originally posted by: coldpower27
There was a miscalculation there, it depends on what you mean by ahead. AMD will likely be good in the 2P and 4P Sectors where more of AMD's advantages come into play. As well where the 4P sector won't receive the 45nm refresh till sometime in 2k8

On desktops it's going to be much less clear,

Since when has AMD in front of the majority of the last 10 years? Or are you talking in front in terms of gaming?

Pentium 3 and Athlon were even blow for blow till Q4 2000 more or less, and since 10 years includes 1997 Intel was ahead when it was K6 vs Pentium 2/3. Athlon wasn't released till Mid 1999 and then things were more even between the 2. AMD could be argued to be ahead when you reach Willamette and Early Northwoods, but things even out with Northwood B and Northwood C took the lead for awhile. then AMD release Athlon 64's and thing were even again, and it wasn't till Athlon 64x2 that AMD takes the lead, which was then broken with Core 2 and will hold it till Q4 2007 when AMD releases desktop based K10's.

No, my 10 years are assumed (from the other poster) that Nehalem will be the next intel chip that leapfrogs Barcelona. That means it goes to 1999. Athlon was neck and neck, if not faster since then, in the majority of benchmarks. P4 only held its own in media encoding.

You really should look up things before simplifying so much.

No The Northwood C's beat out the Athlon XP's across the board, not just media encoding Northwood B was even.

Since the Single Core Pentium 4's vs Athlon 64's remain a tie as each was suited to different tasks.

 

[dmens]

Originally posted by: jpeyton
Looking at the Penryn performance preview on Anandtech's main page, they concluded Penryn will provide a 5-10% performance clock-for-clock.

Hardly a bombshell.

that's funny. lots of architects would sacrifice their firstborn for a general 10% improvement.

you've been severely duped by marketing talk (ie bullshit). i guess those guys do deserve their pay, LOL.

 

[NoobyDoo]

AMD press release at techpowerup.com

Luckily, it?s much easier to find the results comparing AMD?s new Opteron 2222 processor and Intel?s 3.0GHz Xeon 5160. These show AMD enjoying greater performance by as much as 15% in some SPUCcpu2006 tests, which it credits to its Direct Connect Architecture and DDR2 memory.
...
AMD also disclosed updated performance projections for its upcoming native Quad-Core AMD Opteron? processors, code-named ?Barcelona.? The new Barcelona projections are based on the latest SPECcpu2006 benchmarks and show that AMD expects to have up to a 50 percent advantage in floating point performance and 20 percent in integer performance over the competition?s highest-performing quad-core processor at the same frequency.

 

[Cookie Monster]

...the competition?s highest-performing quad-core processor at the same frequency...

So they have to downclock the current X6800 quad just to show that 20% performance increase? (since rumours point to the highest clocked barcelona is ~2.7ghz i think it was, could be lower)

:roll:

Barcelona could be faster clock per clock, but doesnt necessarily show any advantages if the competition can clock their CPUs much higher than the competition can (if the upcoming intel CPUs clocking at 3.33ghz~), and the fact that intel has the 45nm process advantage. Unlike back in the past, intel is able to consistently up the clocks while putting the power consumption in check at "Acceptable" levels as well.

 

[MDme]

some rumors....(which could be facts)

1. Agena FX (Quad core) can OC to 3.05 Ghz
2. It's TDP is 95w
3. The QX6800 TDP is 130w (intel - typical TDP) or estimated by AMD at 150w (max TDP)

links:
http://www.fudzilla.com/index.php?option=com_content&task=view&id=663&Itemid=1
http://www.fudzilla.com/index.php?option=com_content&task=view&id=658&Itemid=1
http://www.fudzilla.com/index.php?option=com_content&task=view&id=649&Itemid=1
http://www.fudzilla.com/index.php?option=com_content&task=view&id=631&Itemid=1
http://www.fudzilla.com/index.php?option=com_content&task=view&id=470&Itemid=35

If the performance of Barcelona/Agena is really +50% in FP and +20% in integer then intel would need around 3+ Ghz of Conroe to match it (a little bit less if penryn is used for comparison)

The 50% FP and 20% int advantage means that it will outperform it clock-for-clock by at least 20% based on code used. 2.5Ghz Agena=3.0Ghz C2D in int and 3.7Ghz C2D in FP.

Don't forget that Kuma Dual cores (which will launch at 2.0-2.9Ghz) should scale to a higher frequency if the Agena is really able to OC to 3.05Ghz.

Just imagine a 3.2 Ghz Kuma DC....

*disclaimer: these statements assume that item #1-3 are accurate. THe poster does not assume responsibility for the content of the links.

 

[Regs]

Can you imagine if all these CPU's went to Dell instead of ...let's say...Newegg or ZipZoomFly?

Would you want to buy a 2000 dollar PC from Dell just to have a 200 dollar CPU?

I hope AMD doesn't make the same mistake twice.

 

[rexian96]

Originally posted by: MDme
some rumors....(which could be facts)

1. Agena FX (Quad core) can OC to 3.05 Ghz
2. It's TDP is 95w
3. The QX6800 TDP is 130w (intel - typical TDP) or estimated by AMD at 150w (max TDP)

links:
http://www.fudzilla.com/index.php?option=com_content&task=view&id=663&Itemid=1
http://www.fudzilla.com/index.php?option=com_content&task=view&id=658&Itemid=1
http://www.fudzilla.com/index.php?option=com_content&task=view&id=649&Itemid=1
http://www.fudzilla.com/index.php?option=com_content&task=view&id=631&Itemid=1
http://www.fudzilla.com/index.php?option=com_content&task=view&id=470&Itemid=35

If the performance of Barcelona/Agena is really +50% in FP and +20% in integer then intel would need around 3+ Ghz of Conroe to match it (a little bit less if penryn is used for comparison)

The 50% FP and 20% int advantage means that it will outperform it clock-for-clock by at least 20% based on code used. 2.5Ghz Agena=3.0Ghz C2D in int and 3.7Ghz C2D in FP.

Don't forget that Kuma Dual cores (which will launch at 2.0-2.9Ghz) should scale to a higher frequency if the Agena is really able to OC to 3.05Ghz.

Just imagine a 3.2 Ghz Kuma DC....

*disclaimer: these statements assume that item #1-3 are accurate. THe poster does not assume responsibility for the content of the links.

Sounds very interesting. I wonder if these AM2+ CPUs can be overclocked with an AM2 board ? I guess it's too early expect a definitive answer They didn't mention if they used a future board or one that's available in the market when they overclocked that 1.9GHz Agena.

 

[JackPack]

Originally posted by: MDme
some rumors....(which could be facts)

1. Agena FX (Quad core) can OC to 3.05 Ghz
2. It's TDP is 95w
3. The QX6800 TDP is 130w (intel - typical TDP) or estimated by AMD at 150w (max TDP)

1. The 3 GHz Agena OC screenshot has been proven to be fake. CPU-Z doesn't look for an L3 on the K8, according to the author of CPU-Z.

2. TDP of Agena FX is 120W. Agena might be 95W, but >2.5 GHz is 120W.

3. TDP of Intel QX series is 130W. Refer to reputable sites such as LostCircuits where they have already measured the QX6700 to be 115W.