No clockspeed wonders from 65 nm AMD [link:theInq]

[apoppin]

Updated:
i assumed this was not a letter . . . sorry



take it for what it is worth

[nothing]

===================================










So all you silly fanboys expecting clockspeed wonders from 65 nm AMD please keep it in your dillusioned dreams. Those expecting K8L to totally own Conroe do the same.

Ouch . . . for real?

quoted . . .in part

...But some fanboys are running around and keep saying that it will change when AMD intros 65 nm chips that will clock through the roof.

I AM SO TIRED of saying it over and over again. So I am going to say it once more. But here goes the explaination.

AMD's first generation processes have been pretty crappy for a while.

First generation 0.18 K7 Thunderbirds couldn't clock above 950 Mhz until they went with copper interconnects (second generation 0.18), which clocked to 1400 MHz. Third generation 0.18 (Palomino) went to 1.73 GHz.

First 0.13 K7's only went to 1.80 GHz (2200+). With the second 0.13 process we saw 2.2 GHz (3200+) and that was far from the limit as I am sure everyone remembers with Mobile chip overclocking. A shift to the K8 architecture and SOI took 0.13 all the way up to 2.6 GHz (FX-55).

First 0.09 K8's clocked LOWER than the aforementioned Clawhammers (the first actual drop in clockspeed in the above history). . . .

if we look at roadmaps for Quad Father, we see 90 nm FX'es kicking it up at clocks of up to 3.2 GHz well into next year. If AMD could produce 65 nm chips hitting those clocks, why on earth would they keep making these 90 nm chips which are more expensive, and produce more heat?

Here is what we will see again here. 65 nm first generation will clock worse than 90 nm cherry steppings. When second generation 65 nm comes around, we will see them finally clocking better than 90 nm (which will be 3.2 GHz by then for good ol' plain K8). Sounds like that will happen after or around the same time Intel moves to 45 nm.

. . . Looks like K8L loses the clock speed race here. By how much? Well, 3.4 GHz Conroe would be a piece of cake for Intel to make if K8L's IPC kicks serious butt. If 2.9 GHz K8L is the high end, to make up the clock speed deficit and MATCH Conroe (or Kentsfield at 4 cores) performance, K8L better have 20% higher IPC than Conroe. Do you see it happening? I don't.

So all you silly fanboys expecting clockspeed wonders from 65 nm AMD please keep it in your dillusioned dreams. Those expecting K8L to totally own Conroe do the same.

Note: I have been an almost exclusively AMD user for the last 6 years, but how I hate fanboys and like my Conroe. Grrrr.

Ivan Andreevich

i am just now starting to read your threads here . . . and am particularly looking to completely *makeover* my rig next Winter . . . i have watched Intel and AMD go head-to-head in technology with AMD keeping the performance lead for the last couple of years - until conroe. Is Intel really that far ahead?
:Q

 

[TanisHalfElven]

inquirer pffft!!!!!!!!!!!!!!!!

 

[apoppin]

Originally posted by: tanishalfelven
inquirer pffft!!!!!!!!!!!!!!!!

FIRST of all . . . it is reflected in the topic's title. TheInq has been 'on' [many times] . . . and 'off' [more often] . . . they look to be right about G80 [as confirmed bt PCWelt]

Secondly it is an editorial . . . he seems to be right about the 1st generation parts and o/c'ing . . . Even Intel . . . the first P4 was really [really] crappy compared even with the [soon-to-be-discontinued much cheaper] Tualatin Celerons
[boy, intel sure went in the wrong direction that time]

and i AM reading that the 65nm parts from AMD will be super o/c'ers

that is the POINT of the article . . . NOT to expect too much from the 1st gen parts.

edit:

related info
[yes . . . daamit . . . theinq]

AMD desktop market share about 27%

 

[TanisHalfElven]

i SAID

inquirer pffft!!!!!!!!!!!!!!!!

 

[ScythedBlade]

Hey ... they bad-mouthed Conroe too ... but then again, theoretically, the evidence for AMD's trend does hold more water than when they used evidence for Conroe ...

 

[lopri]

This is not even an Inquirer reporter's writing, but a letter from.. eh.. Ivan. And what is the point of transfering some sour fanboy's letter?

Yellow Journalism.

 

[apoppin]

Originally posted by: lopri
This is not even an Inquirer reporter's writing, but a letter from.. eh.. Ivan. And what is the point of transfering some sour fanboy's letter?

Yellow Journalism.

you are absolutely right . . . when i saw the supposed link to http://www.volesoft.com/default.aspx
i assumed wrongly that it was not a letter

sorry

nothing to see

move on . . .

 

[SparkyJJO]

I can see K8L providing some killer performance. If intel went from the P-D which was ok to the amazing C2D, why can't AMD go from the good X2 to something amazing as well? It CAN work both ways ya know

yeah yeah call me an AMD fanboy but after having the crown for 3 years I don't think that AMD will give it up too easily. I don't think intel will give it up easily either after getting it back, but we'll just have to wait and see.

 

[Regs]

AMD stated that their Barcelona will offer 150% increase in performance over K8. Though when you consider it's 4 cores you can easily split that percentage up in real world performance to be 37-38% which would make it a little over par with the Core 2. They also talk about increased parallelism. What that will translates into real world performance is yet to be seen.

"The most obvious change we will see will be the addition of a shared L3 cache. This new L3 cache, debuting at 2MB in Athlons and Opterons, will be used by all four cores. Each core will also have their own smaller L1 (64KB) and L2 (512KB) caches that can be shared by all the cores as well. This moves towards latency reduction in an attempt to lay the hardware groundwork for truly parallel software programming that can use all the cores simultaneously.

- http://enthusiast.hardocp.com/article.html?art=MTE3NCwxLCxoZW50aHVzaWFzdA==

AMD is improving their Interconnect for more than just performance. Even with a cross-trade tech agreement with Intel, their competition will have a lot of catch-up to do before they can even reach AMDs advancements.

--------------------
When talking to both Phil and Pat, we got some interesting answers about advancements in Instructions Per Clock with the Barcelona Pat Conway said that we will see a ?performance improvement of 150% over the next two years that will be within the same power envelope.? If you take into account that you can back off frequency and add more cores, some of this can be achieved now if you are counting IPC across all cores. (See below how simply pulling back on frequency by as little as 16% can have a huge impact on power draw.) Intel is bound by the same principles as well.


On IPC and AMD?s Barcelona Phil Hester had this to say: ?We want to stay focused on upgrade compatibility. We must realize more IPC per watt, and our next-generation architecture will show a 50% improvement of IPC per watt.? So I came away with a somewhat clouded view on IPC, but I did get the distinct impression that AMD?s next-generation architecture would be on par or better than what we have seen with Intel?s Core 2 Duo, which is the low benchmark AMD must shoot for.
-----------

 

[Regs]

"While PC games are still dominantly single-threaded applications, Phil sees the Microsoft Xbox 360 and the Sony Play Station 3 as the hardware devices that will bring gaming around to being true multithreaded applications. We?ll ignore the unreleased PS3 for now, but the Xbox 360 has enjoyed tremendous success. The Xbox 360 has three dual-threaded processors onboard, and Phil thinks that this has game developers moving down the multithreaded path.

"Undoubtedly game developers are aware that the frequency jumps they enjoyed in the past are going the way of the dinosaur, but the last Core 2 Duo gives a great bit of promise in the gaming world as it seems to scale very well. While GPUs seem to now be the biggest bottleneck for high-definition gaming, that will change soon and processors will once again have to pick up the slack. For that to be achieved, games are simply going to have to become multithreaded."