Why I'm staying with AMD

Xsorovan

Senior member
Oct 14, 2002
320
0
0
All the hype around Intel's C2D chip is, quite honestly, making me feel a little sick. It seems every reviewer on the internet is completely drooling over these processors and driving a frenzy of "best CPU ever" from people who have yet to play with the processor. In a world where computer technology changes so quickly it seems very short sighted of reviewers to put all their stock into this one chip.

My main reason for staying with AMD right now, as well as when I buy a new machine in a couple of weeks, is the nanometers. AMD is still running at 90nm. Intel gained a 20% increase from going to 65nm. Of course, this is to be expected, everything is that much closer together. What happens in January when AMD drops their 65nm chip? Intel goes back to being behind as AMD generally does better work with their chips. So while AMD might be behind now, I think it's short sighted to say that Intel is worth investing in as a long term winner over AMD. Granted it's going to give you a speed boost now, but what about in January when all the benchmarks come in on the AMD 65nm and it beats Intel?

I understand the hype, this is the first available 65nm chip available to everyone, and there are obviously speed increases that come with that, but just because Intel beats AMD to market doesn't mean that Intel will have a faster chip than AMD's 65nm. I say we hold judgement until we can race them side by side. I'll buy my AM2 mobo and slap a 4600+ into it now, knowing that the C2D is faster, but also knowing that when AMD drops their 65nm chip I won't have to buy a whole new machine when suddenly its "AMD's newest chip is soooo fast!! I must have one!"
 

thilanliyan

Lifer
Jun 21, 2005
12,053
2,271
126
I think you assuming that 65nm is gonna suddenly make AMD top dawg is a bit optimistic...no one knows how well the 65nm AMDs will do.

Also, the C2D is not just a process shrink...it's a completely different architecture...unless AMD actually changes the architecture, do you think they'll get like a 20-30% improvement??

Even going from 130nm to 90nm for AMD, did that give upwards of 20% improvements?? I think the larger effect was on power consumption. Just because you move to 65nm doesn't mean you can scale the GHz up at will...the basic architecture also has a say in CPU speed, not just the die process.
 

RichUK

Lifer
Feb 14, 2005
10,341
678
126
AMD need a new Architecture to compete with Conroe not a die shrink. AMD will only be able to reduce the heat output form a die shrink, not improve or gain any tangible performance from migrating to 65nm.

K8L is the answer, but when and what would be the performance.
 

n19htmare

Senior member
Jan 12, 2005
275
0
0
Originally posted by: Xsorovan
Intel gained a 20% increase from going to 65nm.

HAHA... that's the best comment i've heard in a while....

I don't know man, some Pentium -D are 65nm... they didnt get any boost from transition. Whats up with that?

i'm sure all AMD needs is a transition to 65nm to beat conroe.

thanks for the laugh, i was getting really bored at work.
 

dexvx

Diamond Member
Feb 2, 2000
3,899
0
0
Originally posted by: RichUK
AMD need a new Architecture to compete with Conroe not a die shrink. AMD will only be able to reduce the heat output form a die shrink, not improve or gain any tangible performance from migrating to 65nm.

No, die shrinks help if done properly. Judging from 130nm -> 90nm, the K8 went from a 2.6Ghz (FX-55) to a whopping 2.8ghz (FX-57). Even assuming a 65nm shrink will get the K8 to 3.6Ghz, which is rather generous, that would still barely match the slated 3.33 Conroe XE at the end of the year.

And contrary to popular opinion, Intel at 90nm matched K8 at 90nm performance and wattage: the 90nm Pentium-M Dothan. The only problem was that it was marketted and sold as a mobile solution instead of a desktop one. If Intel slapped 2x Dothans on a die instead of 2x Prescotts, it would've made an infinitely better Dual Core solution.
 

RichUK

Lifer
Feb 14, 2005
10,341
678
126
Originally posted by: dexvx
Originally posted by: RichUK
AMD need a new Architecture to compete with Conroe not a die shrink. AMD will only be able to reduce the heat output form a die shrink, not improve or gain any tangible performance from migrating to 65nm.

No, die shrinks help if done properly. Judging from 130nm -> 90nm, the K8 went from a 2.6Ghz (FX-55) to a whopping 2.8ghz (FX-57). Even assuming a 65nm shrink will get the K8 to 3.6Ghz, which is rather generous, that would still barely match the slated 3.33 Conroe XE at the end of the year.

And contrary to popular opinion, Intel at 90nm matched K8 at 90nm performance and wattage: the 90nm Pentium-M Dothan. The only problem was that it was marketted and sold as a mobile solution instead of a desktop one. If Intel slapped 2x Dothans on a die instead of 2x Prescotts, it would've made an infinitely better Dual Core solution.

Yes i am aware this. My main statement was that AMd require a new archtecxture.

Also the record for AMD's highest clocked chip was done on a 130nm FX55, not a 90nm FX57.
 

dexvx

Diamond Member
Feb 2, 2000
3,899
0
0
130nm FX-55 Clawhammer was 2.6Ghz, fastest 130nm
90nm FX-57 San Diego was 2.8ghz, fastest 90nm (still to date)
 

thilanliyan

Lifer
Jun 21, 2005
12,053
2,271
126
Originally posted by: dexvx
130nm FX-55 Clawhammer was 2.6Ghz, fastest 130nm
90nm FX-57 San Diego was 2.8ghz, fastest 90nm (still to date)


I think he meant world record for fastest K8...not stock from AMD.

Wasn't it something like 4GHz??
 

Canterwood

Golden Member
May 25, 2003
1,138
0
0
I think a lot of people are going crazy cos Intel hasn't had anything worthwile for 3-4 years.
Finally they have something which can compete, and thats good for consumers with all the price cuts that are going to happen.

Hopefully AMD will recover from this and keep the cpu wars going, or else all we have to look forward to is very high cpu prices and much less innovation.
 

JPH1121

Member
Mar 11, 2006
80
0
0
Isn't it a single core FX55 @ 130nm to a dual core FX62 @ 90nm?
200 ghz more and twice the cores in the same TDP if I'm not mistaken...

I'd call that pretty significant...

If AMD doesn't double the cores, we can assume a power drop @ the same clock speed and higher clocks...

We'll see...it certainly COULD give AMD the ppw crown again...if they retrieve the performance crown is a different deal...
 

Keysplayr

Elite Member
Jan 16, 2003
21,219
54
91
And a die shrink improves performance per clock, exactly how? Are you hoping that AMD can reach a 4GHz o/c on the 65nm process? That would be terrific if so, but still not enough as it seems Core 2 Duo can o/c quite well itself. I don't see how you can think a die shrink can improve performance, do you? If you mean power consumption and heat reductions, most certainly. But both of those things go right up again as it is overclocked just like any other processor. You're on the wrong track bro. Don't know where you got your thinking cap from today. RMA it.

AMD needs architectural overhaul to compete or take back the "crown". They are going to be forced to play the MHz game as Intel did with P4.
 

cmrmrc

Senior member
Jun 27, 2005
334
0
0
AMD will regain some ground with 65nm, i'd say max 5%...switching to 65nm is not just a die shrink, they can improve the architecture just a little bit...i remember when AMD switched from 130nm to 90nm, there were something like -2 to 5% increase in performance.
 

RichUK

Lifer
Feb 14, 2005
10,341
678
126
Originally posted by: dexvx
130nm FX-55 Clawhammer was 2.6Ghz, fastest 130nm
90nm FX-57 San Diego was 2.8ghz, fastest 90nm (still to date)

Im talking about the OC record, of 4Ghz+ for the Clawhammer, against about 3.8/9Ghz record for the 90nm San Diego core.

I'm aware that die shrinks, bring a better yeild in clock frequency, they probably could have had a number of 2.8Ghz Clawhammer chips, but the yield would have been low. But if i remember correctly AMD used different technologies on its FX57 San Diego silicon against the Fx55 Clawhammer Silicon. One of which aided in faster switching transistors, i think it was the strained process used on the San Diego core, however i believe this process also lead to more leakage too.
 

RichUK

Lifer
Feb 14, 2005
10,341
678
126
Originally posted by: cmrmrc
AMD will regain some ground with 65nm, i'd say max 5%...switching to 65nm is not just a die shrink, they can improve the architecture just a little bit...i remember when AMD switched from 130nm to 90nm, there were something like -2 to 5% increase in performance.

Yeah bit what part of the K8 Core did they imporve, nothing.

They used different manufacturing process?s like Strained SOI, and added SSE3 support, along with a few memory controller tweaks.
 

Keysplayr

Elite Member
Jan 16, 2003
21,219
54
91
Originally posted by: broly8877
wow... eat a cyanide capsule, Xsorovan.

That's a tad bit harsh dude. Xsorovan just seems a little misguided at the moment.

 

DrMrLordX

Lifer
Apr 27, 2000
22,894
12,951
136
Originally posted by: keysplayr2003

That's a tad bit harsh dude. Xsorovan just seems a little misguided at the moment.

While the OP doesn't need to eat any cyanide, he should do himself a favor and read Anandtech's Conroe articles. It seems to me that he either ignored/skimmed the articles or is in denial.

 

GuitarDaddy

Lifer
Nov 9, 2004
11,465
1
0
There are valid reasons that can be argued for staying with AMD, unfortunenately the OP's 65nm argument is not one of them
 

JPH1121

Member
Mar 11, 2006
80
0
0
Don't neglect that AMD has touted a 40% improvement in transistor performance iwth thier 65nm process.

I have no idea how that translates to overall CPU performance...considering that CPUs are comprised of transistors...it sounds good to me :p
 

coldpower27

Golden Member
Jul 18, 2004
1,676
0
76
Originally posted by: JPH1121
Don't neglect that AMD has touted a 40% improvement in transistor performance iwth thier 65nm process.

I have no idea how that translates to overall CPU performance...considering that CPUs are comprised of transistors...it sounds good to me :p


Doesn't mean much, in certainly won't mean a 40% increase in clockspeed.

 

Keysplayr

Elite Member
Jan 16, 2003
21,219
54
91
Originally posted by: DrMrLordX
Originally posted by: keysplayr2003

That's a tad bit harsh dude. Xsorovan just seems a little misguided at the moment.

While the OP doesn't need to eat any cyanide, he should do himself a favor and read Anandtech's Conroe articles. It seems to me that he either ignored/skimmed the articles or is in denial.

Sure seems that way.

 

CaiNaM

Diamond Member
Oct 26, 2000
3,718
0
0
Originally posted by: GuitarDaddy
There are valid reasons that can be argued for staying with AMD, unfortunenately the OP's 65nm argument is not one of them

:light: