Wednesday, August 27th 2008

Phenom FX in the Works, AMD to take Another shot at...Kentsfield

Aug 27th, 2008 13:13 Discuss (294 Comments)

The transition of the K10 architecture by AMD to the 45nm silicon fabrication process is stirring up interesting revelations these days. First, it was about surprisingly low power consumption of the quad-core Phenom parts, and then about the overclocking headroom those 45nm parts provided, at least the engineering samples did so far. And now, news coming in that AMD could be resurrecting the "FX" series of extreme performance products. Over the past three or so years, the performance trail AMD products had over Intel's made it close to impossible for AMD to sell parts that provide performance tuning advantages such as unlocked FSB multiplier settings for a premium, like it did back when K8 reigned the performance segment. "Black Edition" chips made up for that deficit by providing consumers overclocking advantages while not charging a significant premium and at the same time, safeguarding the "FX" title, not letting it dilute.

Come AMD Deneb core and lot seems to be on offer. To begin with, unlike the Windsor core that had a maximum FSB multiplier of 16.0x, initial reports suggest the Deneb to sport a maximum 25.0x multiplier, 200 MHz x 25 = 5.00 GHz, with the FSB left to play with. Considering at 2.30 GHz the Deneb draws in 57.3 W (according to findings), it should still leave enough room for AMD to sell premium products clocked at high frequencies.

From Reviewage's findings, there seem to be two Phenom FX processors in the making. The numbering seems to take off where it last left at the Athlon64 FX 74. The two chips, Phenom FX 80 and Phenom FX 82 could be clocked at 4.00 GHz and 4.40 GHz respectively (stock speeds). An interesting statement is that at 4.00 GHz, the Phenom FX 80 should outperform an Intel Kentsfield core clocked at 5.00 GHz, implies it has to be faster than the Kentsfield on a clock-to-clock basis. This opens up an interesting debate on how these parts compare to the succeeding Yorkfield chips. This should also open gates for several models to enter the market at various clock speeds.

Source: Reviewage

Add your own comment

294 Comments on Phenom FX in the Works, AMD to take Another shot at...Kentsfield

#226

X1REME

Wile EHT3.1 isn't going to do squat for performance. The current HT isn't even close to maxed out, neither is PCIe2. Red Herrings to take away from the fact that nothing significant has changed in the architecture besides the die shrink.

And even if 45nm do OC to 4GHz and beyond, Phenom is still slower clock for clock vs current Intel Quads. Tweaks and Die shrinks aren't going to change that fact. They need a whole new architecture to pull that off. To add to all of this, by the time Deneb releases, i7 will be out, and be even faster still.

i7 is only 12% faster in some apps and about 15/20 approx in others compared to current, but is a squat faster in games lol and all for $999 = 3ghz (you will need to spend more on 3x ddr3 sticks-x58 mobo-psu etc)

Without doubt, HyperTransport 3.1 will be used as a communication interface between CPU and GPU and a bandwidth of 51.6 GB/s may open a whole new world of possibilities and an opportunity to be more competitive with Intel in terms of overall performance.
www.tomshardware.com/news/Hypertransport-AMD-Fusion,6179.html

#227

X1REME

BvB123Bullshit :)

K10 = 2MB L3 Cache

2,4Ghz Phenom 9750 = TDP 95W
Deneb 2,3Ghz TDP ? no one know it.;)

AMD cache structure if they have stuck to their K8 implementation is pretty good though. They are able to optimize the accesses and the level used to store code/data. AMD in the K8 days used to be able allocate some code/data in one level and other code/data at another and the CPU could access either level directly. It's like having a big cache that has various amounts of latency. I'm sure K10 will have this as well since it's beneficial for a lot of applications out there including virtualization technology.

#228

Wile E

Power User

X1REMEi7 is only 12% faster in some apps and about 15/20 approx in others compared to current, but is a squat faster in games lol and all for $999 = 3ghz (you will need to spend more on 3x ddr3 sticks-x58 mobo-psu etc)

Without doubt, HyperTransport 3.1 will be used as a communication interface between CPU and GPU and a bandwidth of 51.6 GB/s may open a whole new world of possibilities and an opportunity to be more competitive with Intel in terms of overall performance.
www.tomshardware.com/news/Hypertransport-AMD-Fusion,6179.html

I'm not talking games. It doesn't take a quad core for gaming. I'm talking cpu performance in general.

And HT3.1 isn't going to help performance at all at this stage. HT is not a bottleneck as it is. HT 3.1 can only help in multi-socketed server setups. The desktop market has no need for the bandwidth.

And besides, when did we bring the roadmap into this? We are talking Deneb vs Kentsfield/ Yorkfield/i7.

The fact remains, AMD is slower clock for clock. And I'm willing to bet that is still gonna be the case after the 45nm chips release, especially after i7 hits the market. I have a feeling that those that think or hope otherwise are just working themselves into another letdown.

I will never get my hopes up over an AMD cpu again after the Phenom letdown. I'll wait for benchmarks.

#229

Melvis

Wile EThey need a whole new architecture to pull that off. To add to all of this, by the time Deneb releases, i7 will be out, and be even faster still.

LOL thats funny, the new i7 is going to be very similar to the AMD architecture anyway, and by the looks of things isnt going to be faster (maybe around the same). So a whole new architecture? i don't think so.

#230

Wile E

Power User

MelvisLOL thats funny, the new i7 is going to be very similar to the AMD architecture anyway, and by the looks of things isnt going to be faster (maybe around the same). So a whole new architecture? i don't think so.

Well, considering their current arch can't keep up clock for clock, I'd say yes. At best, the shrink will be an incremental upgrade, which might get it to Kentsfield, just as the title of this thread suggests. That means they are still behind Yorkfield, and since i7 is faster than Yorkfield, AMD is still behind even that.

They are getting nowhere fast on k10, which is still based on K8. They need to make a bigger change than a shrink to 45nm to keep up with Intel right now. I'll say it again, a die shrink, and a couple of small architecture tweaks will not provide the performance difference necessary to catch Intel clock for clock. It will take major tweaks to pull that off. You only have to look at the history of die shrinks and minor revisions of ANY cpu to figure that out.

#231

Melvis

Wile EWell, considering their current arch can't keep up clock for clock, I'd say yes. At best, the shrink will be an incremental upgrade, which might get it to Kentsfield, just as the title of this thread suggests. That means they are still behind Yorkfield, and since i7 is faster than Yorkfield, AMD is still behind even that.

They are getting nowhere fast on k10, which is still based on K8. They need to make a bigger change than a shrink to 45nm to keep up with Intel right now. I'll say it again, a die shrink, and a couple of small architecture tweaks will not provide the performance difference necessary to catch Intel clock for clock. It will take major tweaks to pull that off. You only have to look at the history of die shrinks and minor revisions of ANY cpu to figure that out.

Yea but have you noticed the difference between the two at clock to clock? and do you see why? there is a difference in performance clock to clock? its the Cache sizes, it allways has been, all you have to do is look back to the old P4 days and see that the CPU's that had more L2 cache was quicker, and of course also the OBMC. Clock to clock means stuff all, i think you should realize that from the past, but now since they are getting very close to been the same, then it is a factor more so now, but not back then. Once again, wait for the increase of the Cache sizes then test, that will be the benchmark that will tell all.

Get two CPU's that have very similar configuration then see, at the moment, you just cant. end of story.

#232

Wile E

Power User

MelvisYea but have you noticed the difference between the two at clock to clock? and do you see why? there is a difference in performance clock to clock? its the Cache sizes, it allways has been, all you have to do is look back to the old P4 days and see that the CPU's that had more L2 cache was quicker, and of course also the OBMC. Clock to clock means stuff all, i think you should realize that from the past, but now since they are getting very close to been the same, then it is a factor more so now, but not back then. Once again, wait for the increase of the Cache sizes then test, that will be the benchmark that will tell all.

Get two CPU's that have very similar configuration then see, at the moment, you just cant. end of story.

What does it matter? AMD doesn't put more cache on their chip, so that's a pointless argument, and doesn't change the fact that Intel is faster clock for clock.

#233

Melvis

Wile EWhat does it matter? AMD doesn't put more cache on their chip, so that's a pointless argument, and doesn't change the fact that Intel is faster clock for clock.

It makes all the difference, why do you think the C2D is so fast? can you tell me the main reason why? from its old P4 days? if you look its got a F load more Cache?and FSB and you know why they also dont put more cache on there AM2 CPU's?, because it god dam expensive to do. Its not pointless, because its the main and most obvious reason why, look back, and see, its all there, anyone can see it. Now since AMD is going to up there Cache sizes, and i wonder why , like derr, we might now see some real close results.

Its just so obvious its in everyone's faces, and everyone thinks its not, like hello, its the biggest change of all for any intel CPU, apart from the next gen, with a OBMC, thats even bigger. its all there, just look back and you will see the facts.

#234

farlex85

MelvisIt makes all the difference, why do you think the C2D is so fast? can you tell me the main reason why? from its old P4 days? if you look its got a F load more Cache?and FSB and you know why they also dont put more cache on there AM2 CPU's?, because it god dam expensive to do. Its not pointless, because its the main and most obvious reason why, look back, and see, its all there, anyone can see it. Now since AMD is going to up there Cache sizes, and i wonder why , like derr, we might now see some real close results.

Its just so obvious its in everyone's faces, and everyone thinks its not, like hello, its the biggest change of all for any intel CPU, apart from the next gen, with a OBMC, thats even bigger. its all there, just look back and you will see the facts.

Cache is only part of the reason core 2 succeeds, and amount isn't everything. They improved the efficiency of it, added instruction sets, more efficient execution codes, ect. The entire architecture is just far more efficient. Even a lowly allendale w/ 1mb or 2mb of cache mops the floor w/ a p4, and at much lower clock speeds. It's just more efficient. Cache is part of it, amount of cache is part of it, but not by a long shot all of it. It's just the easiest thing to point out, there's much more going on under the hood so to speak.

#235

fullinfusion

Vanguard Beta Tester

Wile EWhat does it matter? AMD doesn't put more cache on their chip, so that's a pointless argument, and doesn't change the fact that Intel is faster clock for clock.

at the moment it is..;)

#236

Melvis

farlex85Cache is only part of the reason core 2 succeeds, and amount isn't everything. They improved the efficiency of it, added instruction sets, more efficient execution codes, ect. The entire architecture is just far more efficient. Even a lowly allendale w/ 1mb or 2mb of cache mops the floor w/ a p4, and at much lower clock speeds. It's just more efficient. Cache is part of it, amount of cache is part of it, but not by a long shot all of it. It's just the easiest thing to point out, there's much more going on under the hood so to speak.

well yea thats true to, i can understand that, i didn't say it was all of it but a major and most stand out one of them all. AMD is using a old architecture in some area's, the K8 has been around for ages, and the AM2 from 939 wasn't anything different at all, apart from the DDR2, but for something that is getting out dated now, its done a dam good job of giving intel a good run for there money, and thats were it all come's down to, your wallet.

#237

cdawall

where the hell are my stars

thing is current AMD chips can hit 4ghz it just takes work my 5000BE would post @4ghz under water (and at ungodly high volts) which is old school K8 there is no reason with some work AMD cant make that a 24/7 clock if they drop the volts on the chips and make them more efficient which is what K10.5 will do with 45nm. the 6mb cache which IS more efficient than intel's cache structure WILL improve the performance of K8(10.5) significantly look at a phenom vs a AX2 ~20% better performance clock for clock my phenom @2.84ghz BEAT my AX2 @3.35ghz this is same mobo same VGA same ram etc. now think if the triple the cache and double the clock speed what that will do to performance esp. when you consider that a phenom scales with clock very very well.

#238

X1REME

Wile EI'm not talking games. It doesn't take a quad core for gaming. I'm talking cpu performance in general.

And HT3.1 isn't going to help performance at all at this stage. HT is not a bottleneck as it is. HT 3.1 can only help in multi-socketed server setups. The desktop market has no need for the bandwidth.

And besides, when did we bring the roadmap into this? We are talking Deneb vs Kentsfield/ Yorkfield/i7.

The fact remains, AMD is slower clock for clock. And I'm willing to bet that is still gonna be the case after the 45nm chips release, especially after i7 hits the market. I have a feeling that those that think or hope otherwise are just working themselves into another letdown.

I will never get my hopes up over an AMD cpu again after the Phenom letdown. I'll wait for benchmarks.

actually games matter very much as it stimulates the small enthusiast crowd, which in return effects normal users choice in buying. (that's a FACT)

well its funny how Intel can use there (qpi etc) to communicate between there chipsets (server and desktop) so why is it not possible for amd who actually invented it.

i never thought it was between you and me and the road map was for info, to say they are not sticking with it.

certainly it has not been the case in the past e.g. AMD Athlon, opteron etc

you make it sound as if Intel has always been on top form. Intel has amd to thank for their i7 design or should i say Opteron. amd has always been the best designers and they did mess up on k10 ..hardware bug—known as an erratum—affected the clock speeds of AMD's quad-core processors, but that's not to say that's how it will stay

#239

X1REME

Wile EWhat does it matter? AMD doesn't put more cache on their chip, so that's a pointless argument, and doesn't change the fact that Intel is faster clock for clock.

L2 cache is very good for games (games demand it lol) as it show the most potential.

the funny thing is! amd will build on K8/K9/K10/K10.5/K10.5 Rev-D e.g. deneb, shanghai, bulldozer, fusion etc.
so why would amd do that if k10 wasn't any good :confused:

AMD is the best in design wins to this very day

#240

trt740

man where , where you guys when I needed you back in the AMD days eh WilE remember the beat down I took. The AMD chips are really a fantastic design to be able to stand the test of time and do aswell as they currently do. The 6000+/6400+ still to this day have plenty to offer while gaming and if their dies were shrunk they could easily do the extra 400+ mghz over their true current air overclocking 3.6 ghz, to hit 4.0ghz. Amd should have revisted that idea along time ago. However it appears they are trying.

#241

Tatty_Two

Gone Fishing

MelvisYea but have you noticed the difference between the two at clock to clock? and do you see why? there is a difference in performance clock to clock? its the Cache sizes, it allways has been, all you have to do is look back to the old P4 days and see that the CPU's that had more L2 cache was quicker, and of course also the OBMC. Clock to clock means stuff all, i think you should realize that from the past, but now since they are getting very close to been the same, then it is a factor more so now, but not back then. Once again, wait for the increase of the Cache sizes then test, that will be the benchmark that will tell all.

Get two CPU's that have very similar configuration then see, at the moment, you just cant. end of story.

Lol, it's not the cache size, it's the basic architecture, Intels is just more efficient....simple as that, did you not read the link I posted?, Bang on anothe 4MB of L2 on a K10 and it will make little or no performance difference and whilst the article was a little old, quads for instance have seperate dedicated L2 tied to each core so that makes little difference.

#242

PP Mguire

there is a difference in performance clock to clock? its the Cache sizes, it allways has been, all you have to do is look back to the old P4 days and see that the CPU's that had more L2 cache was quicker, and of course also the OBMC.

Wrong. Comparing those day P4s together, i have a 3.0ghz Northwood 512k L2 that stomps its 1mb Prescott compettitor. With that being said, my 3000+ @ 2.4ghz stomped even the 3.0ghz Northwood. When i clocked my 3000+ up to 2.8 (its a Venice core) it was even worse of a stomping. Back in those days cache didnt mean squat. AMD had the better archietecture. Nowadays C2D is pwning the crap outa AMD with their upped cache and FSB. Where as the current argument is being said about cache and quads, we'll just have to wait and see if the 6mb L3 is gonna help at all. No point in arguing over something that isnt out yet. Thats like arguing whether my shits gonna be light brown or dark brown later today based on what i had to eat yesterday.

#243

SANEagent

Evo85If these ring true in the end, I just found my next CPU! :toast:

Same here!

#244

PP Mguire

Same here!

+3 but i was gonna get a 9950 anyways :/

#245

TheGuruStud

PP Mguire+3 but i was gonna get a 9950 anyways :/

The 140W 9950 is 177 with free shipping on newegg

Do you think it's worth the buy or wait? B/c that's pretty cheap.

#246

PP Mguire

Well ill make the decision when i have some money to spend :) I have to buy a new board and new PSU as well (i have good DDR2 waiting)

Knowing me ill probably wait to see benchies on the new 45nm, and see if the FX rumour is true. If it is, then see how much they are. Then probably buy a super cheap 9950 still rofl.

#247

Melvis

PP MguireNowadays C2D is pwning the crap outa AMD with their upped cache and FSB .

Ummmm thanks, do i need to say much more? LOL ^

Ok maybe i used a bad eg, the P4's were woeful anyway no matter what Cache sizes they had, enough said there. But on a AMD it made a big difference, thats why you had the FX CPU's, with higher amount of Cache, and also the lower 3700, 4000, etc, and with this increase of Cache size alone, made them perform better in mainly gaming. This factor alone made us decide to buy a 3700+ over the 3800+, because it would be better in games.

Im not saying that Cache alone is the be all and end all, but im saying its what stands out the most and also makes them perform better in CERTAIN apps. you cant tell me it doesn't help? then why would they even bother putting that much on? If it doesn't make a difference then WTF? I bet a lower Cache C2D vs a normal rated C2D would loose in a lot of areas, not all, but alot. But we cant =/, so we will have to wait and see for these new AMD's then compare with C2D.

#248

TheGuruStud

C2D with small cache is like trying to breed a dog without nuts. It just doesn't work very well haha.

#249

PP Mguire

Ok maybe i used a bad eg, the P4's were woeful anyway no matter what Cache sizes they had, enough said there. But on a AMD it made a big difference, thats why you had the FX CPU's, with higher amount of Cache, and also the lower 3700, 4000, etc, and with this increase of Cache size alone, made them perform better in mainly gaming. This factor alone made us decide to buy a 3700+ over the 3800+, because it would be better in games.

Idk why but 3000+ @ 2.8ghz > 4000+ @ 2.8ghz for me. The Sandy had 1mb where as the Venice had 512k but i ALWAYS got a better cpu score with the Venice. With the cooling i had both walled at 2.8ghz, but the Venice ran cooler and used less volts. So even AMD vs AMD sometimes more cache isnt better. But like i said, really no point in debating chips that arent out yet. 45nm Deneb will have 6mb L3 so we will see if it makes a vast improvement against its bigger competitor i7.

#250

Abu Assar

Deneb 4 GHz is FAKE

hi all

regarding the original thread , it is an old screen shot of an oc'd deneb
and the real voltage is 1.475 not 1.168 as cpu-z did't read it correctly .

here is the original screen shot which the reviewage site faked shamefully :

and for more info refer to this topic in amd forums:
forums.amd.com/forum/messageview.cfm?catid=318&threadid=99307&enterthread=y

so , please btarunr state this facts on the first post , because this hurts AMD greatly
as over hyping their product will only means that it has a very little chance to meet the expectations .

and let's play the wait and see game

no more deneb rumors and over hyping fake news please .

Add your own comment

Phenom FX in the Works, AMD to take Another shot at...Kentsfield

294 Comments on Phenom FX in the Works, AMD to take Another shot at...Kentsfield

Latest GPU Drivers

New Forum Posts

Popular Reviews

Controversial News Posts

Phenom FX in the Works, AMD to take Another shot at...Kentsfield

Related News

294 Comments on Phenom FX in the Works, AMD to take Another shot at...Kentsfield

Latest GPU Drivers

New Forum Posts

Popular Reviews

Controversial News Posts