• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

AMD-K9

A

AMDrulZ

Member
Well Intels new architecture has been talked a lot lately. But don't forget about the guy's in fab 30 and east fishkill, and what they have been working on. this will mostly be speculations but we do know some small detail's,so let's get talking about the (K-9)

🙂
 
Last AMD Roadmap

If we see the main enhancements are 667 DDR2 memory controller and bus...which I assume must mean 333HTT speeds...I dont know...I domnt know what 667 DDR2 really does....

fact is that we know from testing that the AMD 64's now have no measurable advantage going from say 9x200 to a 6x300 with memory being the same speed...system bus speed is pretty neglible....We aslo have seen that in most things bandwidth does not mean much and the AMD currently doesn't use much of wHat it haves...Dont get me wrong ot will be helpful in certain things and AT showed some mild increase are their but what kind of hit will we take running cas3 DDR2??? It may negate any bandiwdth advantages we may have gotten....

Other thing is 2 bidirectional HTT link....Up to this point this usually only occurred with opteron cpus. Does this hint at possibly dual memory controller or separate dual channle memory banks like in opteron boards??? This could be nice, but then again we go back to how much will it take advantage of this....


Thirdly we have presidio security technology and Pacifica Virtualization....yeah they may be soe nice features for a select few but I dont see them enhancing much for me....



EDIT: K9??? It is the quad core right???? ooppps sorry....

As for quad core it really is going to show the bottleneck we have with software so without a real conserted effort by programmers quad core is going to a V8 with half the cylinders shoot down on the highway like a Hemi....
 
Yea they say it won't be called K-9 but AMD has used the K designation for year's and a lot of people call AMD chip's buy there K designation's so it wouldn't be smart on there part to drop the K designation.
 
AMD must have something that they have been working on for the past 3 years, i all way's thought that once they get one product on they market they start designing the next generation almost the next day!!! So there must be a new architecture that amd is working on or is finished and is just trying to get all they can out of the K8!!! and not just M2 and the opteron socket-F (1207). They are hiding something, because sun micro systems said not to long ago that they will eventually help design opteron with amd and ibm is helping amd on (R&D) and fab processes.
 
K9 isnt ddr2 and all the good stuff we'll be getting in the couple years but a whole new architecture. meaning the chips are going to have more pipes and new design implimentations and do things differently. up until now and including the whole switch to ddr2/new socket is just the same old k8 with enhancements... atleast from what i understand.

it seems far to early to talk about K9.

im putting money on us seeing a more IBM-like cpu design(or approach atleast) then an AMD one.
 
Originally posted by: Duvie
Here are some charts I found....

http://www.heise.de/ct/05/09/022/bild.jpg

looks like AMD is going to get rather power hungry and the roles reverse....
Click to expand...

Anyway, I was under the impression that AMD was calling current dual core K8s the K9, so I'll be referring to them as such.

Notice how the "K9" quad core has the same power envelope as the Opteron Dual-core 2.8GHz in that chart? That's probably the TDP for the first few members of the "K9" quad-core family. Assuming that this TDP is overstated (like AMD's TPD always is) then we can expect it to be around (just a guess, by the way) 100-120 watts. Not too shabby for something that appears to be a 65nm shrink of two current K9 dual-cores. Whitefield's TDP, on the other hand, is probably understated (unless Intel chooses to change its TDP measuring practices) so these two CPUs might end up matching each other in TDP when they actually come out. That is, of course, at full load. I do believe that Intel's CPUs will have better power management at idle or close to idle (since they'll probably inherit all that from the pentium M). I would really like to know where these people got the info in this chart from (I'd guess from motherboard/server manufacturers), it seems quite forward-looking.

EDIT: ack, spelling and a bit of grammar.
 
Originally posted by: Furen
Originally posted by: Duvie
Here are some charts I found....

http://www.heise.de/ct/05/09/022/bild.jpg

looks like AMD is going to get rather power hungry and the roles reverse....
Click to expand...

Anyway, I was under the impression that AMD was calling current dual core K8s the K9, so I'll be referring to them as such.

Notice how the "K9" quad core has the same power envelope as the Opteron Dual-core 2.8GHz in that chart? That's probably the TDP for the first few members of the "K9" quad-core family. Assuming that this TDP is overstated (like AMD's TPD always is) then we can expect it to be around (just a guess, by the way) 100-120 watts. Not too shabby for something that appears to be a 65nm shrink of of current K9 dual-cores. Whithefield's TDP, on the other hand, is probably understated (unless Intel chooses to change it's TDP measuring practices) so these two CPUs might end up matching each other in TDP when they actually come out. That is, of course, at full load. I do believe that Intel's CPUs will have better power management at idle or close to idle (since they'll probably inherit all that from the pentium M). I would really like to know where these people got the info in this chart from (I'd guess from motherboard/server manufacturers), it seems quite forward-looking.
Click to expand...



I think you are right there as well...at least that is what I found mentioned in 2 different places.....K-10 is more likely to be a 65nm quad cores....so perhaps the M2 DDR2 X2's I was first talking about are truly the K9's...
 
Well from what i have heard AMD's next generation chip has not been announced. In their 1.5 year roadmap, all they have are upgrades to the current processor (DDR2, DV).

As for DDR2-667. Would give an increase in bandwidth while sacrificing some latency from what i understand. Therefore unless AMD goes straight to DDR-800 we probably wont see any difference. Another problem with going to DDR-2 is power consumption. While the chips actually consume a lower power, the logic required to operate these chips is moer complex, hence the raise in max TDP on the socket M2's.

-Kevin
 
Originally posted by: Gamingphreek
Well from what i have heard AMD's next generation chip has not been announced. In their 1.5 year roadmap, all they have are upgrades to the current processor (DDR2, DV).

As for DDR2-667. Would give an increase in bandwidth while sacrificing some latency from what i understand. Therefore unless AMD goes straight to DDR-800 we probably wont see any difference. Another problem with going to DDR-2 is power consumption. While the chips actually consume a lower power, the logic required to operate these chips is moer complex, hence the raise in max TDP on the socket M2's.

-Kevin
Click to expand...
Higher consumption doesn't have to do with the DDR2 adoption. That doesn't make any sense. Fact is that AMD hasn't published anything else than TDP numbers so far. As you might know, these numbers can't be applied on a chip-by-chip basis. The only thing they really tell us is what the maximum power conumption of any chip in the series is going to be. The reason for higher TDP of the DDR2 Athlon64 parts is therefore more likely connected to the fact that AMD will push the clock frequencies higher and therefore increase power consumption. The lower frequency chips will still produce very similar power output.
 
I beleive you are all missing the fact that recent AT tests show that more bandwith ACTUALLY helps the X2 processor. I assume that will be the main reason to switch to DDR2.
 
This is true. As additional cores are added bandwidth begins to become a premium. So the evolution to DDR2 (and, eventually, DDR3) is very logical.
 
That is correct. The only new chips from AMD in 2006 will bring a new socket for DD2 support and the new virtualizatin technology.
 
Originally posted by: Duvie
fact is that we know from testing that the AMD 64's now have no measurable advantage going from say 9x200 to a 6x300 with memory being the same speed...
Click to expand...

Wouldn't memory bandwidth start being more important with dual core? I mean, once people start doing that level of multitasking or popular apps come out that efficiently uses both cores.

Yea they say it won't be called K-9
Click to expand...
Yeah, AMD doesn't want anyone to call their CPU a dog. :laugh:
 
Im not sure if someone mention this but isnt k-9 suppose tobe the x-2 chips, while k10 will be the new cpu down the road?
 
For the guy who said that the K9 will most likely be an IBM design instead of an AMD one. THAT will most likely NOT be the case. AMD has alway's prided itself for designing a better architecture than intel ever since the introduction of the Athlon. so it will be an AMD designed architecture not an IBM one it think that IBM and AMD's relationship is mostly fab processes and not architecure design. i think if any thing they are helping each other in the area's they are strongest that way they both have good architecure's for there next generation cpu's, to rival or surpass intel's.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top