The reason why >2.6GHz Phenom won't be out soon

alg7_munif

Supreme [H]ardness
Joined
Oct 9, 2006
Messages
5,862
One of the reason is the latest 2.6GHz Phenom has already carried the number 9950. So what is next? 10050 X4? ;)
 

MrWizard6600

Supreme [H]ardness
Joined
Jan 15, 2006
Messages
5,779
To keep up with Nehalem (or even core 2) deneb will have to crank to 3.0 GHz in an SKU. AMD's in trouble and if their shrink doesn't yeild some seriously higher clocks K10 wont stand any kind of chance against Core 2. I expect you'll see a 3.0GHz Phenom SKU soon.
 

harpoon

Gawd
Joined
Jul 14, 2006
Messages
829
I thought the 140W TDP had something to do with it... you gotta draw the line somewhere.

MrWizard, I don't think even a 3GHz Phenom will stand much chance 'keeping up' with Nehalem. 4GHz maybe...
 

QuickS

Bad Trader
Joined
Jan 31, 2008
Messages
218
I thought the 140W TDP had something to do with it... you gotta draw the line somewhere.

MrWizard, I don't think even a 3GHz Phenom will stand much chance 'keeping up' with Nehalem. 4GHz maybe...

Agreed, theres no way that a 45nm Phenom K10.5 could keep up with Nehalem let alone core2.
 

xdivenx

[H]ard|Gawd
Joined
Mar 4, 2008
Messages
1,941
Glad to see thisn dark side of the forum has come alive, wait until some of the NV fainboi admins come and start deleting our topics. =D
 

oldpablo

Supreme [H]ardness
Joined
May 31, 2003
Messages
6,352
Heh yeah I've honestly wondered what the next chip will carry for a model. Maybe it'll be 9999 or they'll start with decimals like 9950.1. :p
 

alg7_munif

Supreme [H]ardness
Joined
Oct 9, 2006
Messages
5,862
I think that AMD will drop the Phenom name for the future products or maybe they will use something like Phenom 45 X4 just like their Athlon 64 naming scheme.
 

harpoon

Gawd
Joined
Jul 14, 2006
Messages
829
I think the Phenom name will stay, though the numbering system would obviously have to change with Deneb.
 

RamonGTP

Supreme [H]ardness
Joined
Nov 9, 2005
Messages
8,150
There is one good thing about Phenom's "flop" and that is that the name Athlon still has some respect. I'd ditch the Phenom name all together for their next microarchitecture and call it the Athlon X4, but ONLY if it's competitive, otherwise they tarnish that name as well.
 

Lost!

Limp Gawd
Joined
Jun 9, 2008
Messages
190
Deneb should in average be 10% faster than current Phenoms.

So it will be VERY close to Core2, even outpacing it when FSB runs out of steam.

Newest info seems to say that Deneb clocks to 3.4Ghz nicely and with superPi.
 

pxc

Extremely [H]
Joined
Oct 22, 2000
Messages
33,064
Deneb should in average be 10% faster than current Phenoms.
Probably not. AMD's optimistic "20% increase" includes frequency scaling, which implies low single digit per clock performance improvements. It also implies that the targeted frequency is not very high. Some might not find that very impressive. :p

AMD's problem is that it is still struggling with the Q6600, which was released 18 months ago, and all the "problems" it has: no integrated memory controller, "slow" FSB and not "native" quad core. AMD's 45nm should be able to surpass that almost 2 years too late. Intel's current slowest 45nm desktop quad core is now even faster (2.5GHz Q9300), and also has per clock performance improvements in the Yorkfield core tweak (anand found a 7.4% average increase). Even that Q9300 might be discontinued before Deneb ships in late Q4'08 or early Q1'09 in favor of a faster model leaving AMD's best in the same place it currently is: below Intel's slowest. And it looks like Nehalem will be out before Deneb too, with real performance improvements over Yorkfield due to triple channel ODMC, QPI and "native" quad core. Oops.

I think your wildly optimistic view will lead to some wild disappointment come 5-6 months. ;) Is that you, Duby?
 

alg7_munif

Supreme [H]ardness
Joined
Oct 9, 2006
Messages
5,862

I thought that this thread is about the reason why >2.6GHz Phenom won't come out soon. What does Intel CPU have to do with this? Why do you like to post about how great is an Intel CPU inside the AMD subforum?
 

alg7_munif

Supreme [H]ardness
Joined
Oct 9, 2006
Messages
5,862
Um, because the subject was brought up by the poster I quoted.

Why don't you jump on him for going off topic? :rolleyes:

Well, he was talking about Deneb and how it could be a >2.6GHz Phenom. You on the other hand were talking about Intel CPUs :rolleyes:
 

pxc

Extremely [H]
Joined
Oct 22, 2000
Messages
33,064
Well, he was talking about Deneb and how it could be a >2.6GHz Phenom.
FAIL

Deneb should in average be 10% faster than current Phenoms.

So it will be VERY close to Core2, even outpacing it when FSB runs out of steam.
Gee, it looks like he was talking about Intel Core 2 CPU performance, which I quoted and responsed to. :rolleyes:
 

mpcamer1220

[H]ard|Gawd
Joined
Oct 2, 2003
Messages
1,717
Just a heads up - AMD's Desktop Brand Marketing Manager has told me a few times not to extrapolate AMD"s numbering/naming scheme for future products. ;)

I've asked him this very question a few times - once late last year and again in late Q1/early Q2.
 

Lost!

Limp Gawd
Joined
Jun 9, 2008
Messages
190
Probably not. AMD's optimistic "20% increase" includes frequency scaling, which implies low single digit per clock performance improvements. It also implies that the targeted frequency is not very high. Some might not find that very impressive. :p

AMD's problem is that it is still struggling with the Q6600, which was released 18 months ago, and all the "problems" it has: no integrated memory controller, "slow" FSB and not "native" quad core. AMD's 45nm should be able to surpass that almost 2 years too late. Intel's current slowest 45nm desktop quad core is now even faster (2.5GHz Q9300), and also has per clock performance improvements in the Yorkfield core tweak (anand found a 7.4% average increase). Even that Q9300 might be discontinued before Deneb ships in late Q4'08 or early Q1'09 in favor of a faster model leaving AMD's best in the same place it currently is: below Intel's slowest. And it looks like Nehalem will be out before Deneb too, with real performance improvements over Yorkfield due to triple channel ODMC, QPI and "native" quad core. Oops.

I think your wildly optimistic view will lead to some wild disappointment come 5-6 months. ;) Is that you, Duby?


Wildly optimistic? Considering it's 12% faster in SuperPi WITHOUT FREQUENCY SCALING (they are both Phenoms), possibly the WORST AMD benchmark, you know they've improved the chip.

Oh, and happy waiting for your nonexistant mainstream Nehalem that will come 1 year, not 3 months after.
 

LstBrunnenG

Supreme [H]ardness
Joined
Jun 3, 2003
Messages
6,677
Wildly optimistic? Considering it's 12% faster in SuperPi WITHOUT FREQUENCY SCALING (they are both Phenoms), possibly the WORST AMD benchmark, you know they've improved the chip.
Heh, didn't stop Opteron buyers from using it to compare overclocks back in the day. Only reason Intel has such an advantage is because the benchmark can fit in the cache of most Core 2s.
Oh, and happy waiting for your nonexistant mainstream Nehalem that will come 1 year, not 3 months after.
Fail.

pxc, this isn't Duby. His grammars are too good.
 

pxc

Extremely [H]
Joined
Oct 22, 2000
Messages
33,064
Wildly optimistic? Considering it's 12% faster in SuperPi WITHOUT FREQUENCY SCALING (they are both Phenoms), possibly the WORST AMD benchmark, you know they've improved the chip.
These are the only comparable scores I can find for each:

1M SuperPi Mod
Phenom X4 9850BE @ 3.0GHz 25.937s http://www.hardware.info/en-US/usersys/cmtxYZ1ik2ud/view/

Phenom 45nm @ 3.0GHz 23.547s http://www.itocp.com/thread-11103-1-1.html

That's a pretty good 9% speed improvement. AMD must have (finally) fixed the dog slow L3 cache. ;) Since it appears Deneb is out there now, maybe we'll see if other programs show similar improvements. AMD did give a pretty grim scaling assessment though, so I wouldn't count on increases across the board.
 

duby229

2[H]4U
Joined
May 1, 2005
Messages
3,014
I think your wildly optimistic view will lead to some wild disappointment come 5-6 months. ;) Is that you, Duby?

Nope and you can please stop slandering my good name.

I see your still getting paid well. Boy what I wouldnt do to be a paid shill. To even have the pleasure of having it advertised.. (pick your own..... you deserve it) Yeah, you really do deserve it.... :D :D
 

LstBrunnenG

Supreme [H]ardness
Joined
Jun 3, 2003
Messages
6,677
Nope and you can please stop slandering my good name.

I see your still getting paid well. Boy what I wouldnt do to be a paid shill. To even have the pleasure of having it advertised.. (pick your own..... you deserve it) Yeah, you really do deserve it.... :D :D
zomg he's back!
 

RamonGTP

Supreme [H]ardness
Joined
Nov 9, 2005
Messages
8,150
I think it's probably more accurate to say you still troll around these forums than to say "I heard mention"

Whatever reputation you've got around here, you earned it.
 

pxc

Extremely [H]
Joined
Oct 22, 2000
Messages
33,064
Nope and you can please stop slandering my good name.
It would possibly be libel, not "slander," since slander is used for a spoken and libel for written defamation. Believe me, I never mention you IRL. :p But pointing out how someone is unrealistically optimistic probably doesn't count as libel anyways. You may want to consult an attorney about that if it really concerns you.

Anyways, welcome back. We have some things to follow up on soon. Remember those predictions we made last year? :D Here we are in Q3 and you're already dead wrong on a couple of items, as usual.
 

duby229

2[H]4U
Joined
May 1, 2005
Messages
3,014
I think it's probably more accurate to say you still troll around these forums than to say "I heard mention"

Whatever reputation you've got around here, you earned it.

Thats a very trollish thing to say isnt it now? I'm not allowed to browse my favorite forums? Clearly not because high and mighty said so...
 
Top