HKEPC: AMD's 65nm at 5th December

[qcmadness]

http://www.hkepc.com/bbs/itnews.php?tid=699767

Hope it is not the rumour or "gossip at the Inq".

 

[dasickninja]

It sounds reasonable. I see no reason to doubt it.

 

[BaronMatrix]

http://www.hkepc.com/bbs/itnews.php?tid=699767

Hope it is not the rumour or "gossip at the Inq".

I doubt it since HKEPC is in Taiwan where a lot of packaging is done. Some is done in Malaysia, but HKEPC is a reputable source. I guess AMD wants that XMas money.

 

[qcmadness]

http://www.hkepc.com/bbs/itnews.php?tid=699767

Hope it is not the rumour or "gossip at the Inq".

I doubt it since HKEPC is in Taiwan where a lot of packaging is done. Some is done in Malaysia, but HKEPC is a reputable source. I guess AMD wants that XMas money.

HKEPC is a Hong Kong website. 8O

 

[Frozen_Fallout]

http://www.hkepc.com/bbs/itnews.php?tid=699767

Hope it is not the rumour or "gossip at the Inq".

Would be good to see AMD get something new out

 

[Ycon]

Cant see the 65nm parts being better than the 90nm processors in the end.

 

[BaronMatrix]

http://www.hkepc.com/bbs/itnews.php?tid=699767

Hope it is not the rumour or "gossip at the Inq".

I doubt it since HKEPC is in Taiwan where a lot of packaging is done. Some is done in Malaysia, but HKEPC is a reputable source. I guess AMD wants that XMas money.

HKEPC is a Hong Kong website. 8O

I meant "in" Taiwan, not located there. But at any rate they are reputable.

 

[qcmadness]

Cant see the 65nm parts being better than the 90nm processors in the end.

You have insider information????

 

[gOJDO]

Cant see the 65nm parts being better than the 90nm processors in the end.

i think there may be minor improvements.
Like?

 

[qcmadness]

Cant see the 65nm parts being better than the 90nm processors in the end.

i think there may be minor improvements.
Like?

Lower power consumption.
Better overclocking (spectulation).

 

[qcmadness]

higher core clocks better memory handling,more efficient core.

i could go on about what id like to see but i am hopeful it will have at least a 5% perf boost.

I don't think so.

 

[brainysmurf]

Not trying to pick a fight, but I would like to point out there were noticeable improvements in performance between the AMD 64 3000+ Winchester core and the AMD 64 3000+ Venice cores that came out later.
Both were 90nm chips. The Venice core was just a later revision with AMD's usual minor nips, tucks, adjustments, added extensions, etc.

I don't have any hard data handy, but I saw similarly configured machines running (same MB, same brand and model memory, same video card) and I would say the difference between the Venice and Winchester core chips (same mhz, same cache) "felt" like about 5 percent.

How many months have passed between the release of the AM2 socket chips and the planned Dec. 5 release at 65nm? I would argue that's enough time for AMD to have also included a few minor revisions -- no major changes to the architecture, just the usual minor improvements introduced with a new stepping.

 

[r0ck]

http://www.tgdaily.com/2006/11/17/65nm_transition_at_amd/
TGDaily also confirms December the 5th. Same 4x4 debut? Perhaps they'll surprise us with 65nm 4x4 :lol:

 

[gOJDO]

Cant see the 65nm parts being better than the 90nm processors in the end.

i think there may be minor improvements.
Like?

Lower power consumption.
Better overclocking (spectulation).
I agree.

higher core clocks

All available data today says that the highest clocked K8 will be the 90nm CPUs.

better memory handling

I don't think that AMD have changed anything spectacular on the 65nm die shrink. If they could just improve the intercore communication by skiping the L2 cache to RAM writeback, that might boost the multithreading perfromance a bit in some cases.

more efficient core.

Yes, if you are talking about the energy efficiency.

i could go on about what id like to see but i am hopeful it will have at least a 5% perf boost.

It is the same architecture with same the features as Rev F. How do you expect any perfrormance boost?

 

[m25]

Good, however already late since rumours spead for a pre-December release and there are not benchmarked ESs yet :roll:

 

[ikjadoon]

Are we going to get any worthwhile overclocking increases? Or is this going to be like those energy-efficient models, who almost overclocked worse than their stock counterparts?

~Ibrahim~

 

[ikjadoon]

NewEgg was selling a AMD X2 3800+ Brisbane (65nm) a few days ago, but that has disappeared. Was at a premium too, at $175.

~Ibrahim~

 

[qcmadness]

Not trying to pick a fight, but I would like to point out there were noticeable improvements in performance between the AMD 64 3000+ Winchester core and the AMD 64 3000+ Venice cores that came out later.
Both were 90nm chips. The Venice core was just a later revision with AMD's usual minor nips, tucks, adjustments, added extensions, etc.

I don't have any hard data handy, but I saw similarly configured machines running (same MB, same brand and model memory, same video card) and I would say the difference between the Venice and Winchester core chips (same mhz, same cache) "felt" like about 5 percent.

How many months have passed between the release of the AM2 socket chips and the planned Dec. 5 release at 65nm? I would argue that's enough time for AMD to have also included a few minor revisions -- no major changes to the architecture, just the usual minor improvements introduced with a new stepping.

I have to point out that the 130nm => 90nm transition for Athlon64 is from Newcastle to Winchester.

X-Bit Labs have published a report that the performance stayed the same for the transition.

http://www.xbitlabs.com/articles/cpu/display/athlon64-90nm_12.html

 

[qcmadness]

Are we going to get any worthwhile overclocking increases? Or is this going to be like those energy-efficient models, who almost overclocked worse than their stock counterparts?

~Ibrahim~

I think so......

The energy-efficient ones are the low-power bins. But there is a positive relation between power consumption and overclock potential.

 

[ikjadoon]

Yeah...I mean, I have not seen conclusive evidence to say they overclock worse, but the reviews I have read are leaning towards it.

~Ibrahim~

 

[brainysmurf]

I have to point out that the 130nm => 90nm transition for Athlon64 is from Newcastle to Winchester.

X-Bit Labs have published a report that the performance stayed the same for the transition.

http://www.xbitlabs.com/articles/cpu/display/athlon64-90nm_12.html

Yeah, but the 90nm Winchester chips got the same performance at lower temps than the 130nm. Right? I'm going on memory here, but I seem to recall that's why I paid about $45 more for my 3000+ Winchester instead of a Newcastle core a the same speed rating. Because I wanted to overclock and the 90nm process gave a bit more overhead.
That was my first overclock and, not really knowing what I was doing back then and following the advice of others on this post, I very quickly and easily had my 1.8 ghz Winchester running at 2.2 ghz. And it stayed well under 40C at load.

Again, there won't be major speed increases with the change in manufacturing process, but I expect some very minor tweaks and will be disappointed if they are not there.

 

[brainysmurf]

Well, crap.
I just went back and carefully reread the xbit link posted by qcmadness. He's right. The evidence shows that in 2004, when AMD made the 90 nm transition, the new process amounted to an overall average performance improvment of less than 1 percent.

So I acquiesce and admit my anticipation of "about 5 percent better performance" from 65 nm is unrealistically hopeful.

I remain hopeful, however, about overclocking potential.

 

[toasty2]

Hopefully they'll be cheaper.

 

[m25]

thats what i was getting at .

if its cost effective we will see a slight boost.thats just common sense.
in marketin 65nm doesnt really mean much ,but when you do a minor improvement over the same socket design,and give a slight boost,it becomes more marketable.

am2 wasnt shouted about for its slight bump with ddr2 800,as a matter of fact if you werent an enthusiast youd be asking am what?am2 oh.
the chances of amd wanting to provide a slight boost are massive;the only drawback is the probability factor.and apparently these chips are on the doorstep already so with no engineering samples coming out for benches the right assumption is any boost will be next to nothing or there is no boost at all.

they can do 5% they are not stupid.but,time versus money;it may be wiser to wait for the next revision of am2.for them.

Yep, I followed you.... Clock speeds are not going to go up very quickly with process technology revisions with simple shrinks from here on out.... it will take something radical like new materials, different semicondctors, or non-planar CMOS transistors.

I just feel sorry for the AMD die-hards who are thinking AMD's 65 nm process will come in clock to high heaven and save the day... it simply will not. A lot of people are going to be very disappointed.

Not because AMD is not good at process development, but physics is simply reaching the limits.
Yes, most of the improvements will be just from the smaller process (at least at the beginning), however, if 90nm to 65 nm brought such small frequency increase in the P4 architecture, the overclocking potential grew a lot, especially for the Celeron Ds, which lacked a 2M bronto-cache, so I am curious just about how reliable AMDs 65nm SOI will be on this point.

 

[evilr00t]

Er...

I don't know about better overclocking, when the first intel 65nm came out they were horrbile and just as bad at overclocking as the netburst cpus.

when the first intel 65nm came out they were netburst cpus

Ofcourse AMD is a more reputible company and will do something to the architexture.

WTF are you smoking mate? Intel has held the "reputation" crown for quite some time, though AMD has held the "performance" crown for some time