Haswell refresh price list leaked

almighty15 · 3 Mar 2014 at 19:27

Martini1991 said:
Sorry but a minute ago it was all the die size.
I'm aware of the transistor design (Was waiting for that rebuttal tbh)

As far as I see it, I don't believe it can't be done, I just think it's done on purpose, but I guess we'll find it in years to come.
Neither of us knows 100%, so lets not pretend we know more than we do.

http://iweb.tms.org/PbF/JOM-0606-67.pdf

Have a quick look....

Kaapstad · 3 Mar 2014 at 19:30

almighty15 said:
The TM is fine for its purpose... I.e... Stock speeds

almighty15 said:
But they don't tell people to water cool them and overclock the **** off them...

I think you need to read the thread properly starting with your own posts.

IIRC the 4770k has the "K" on the end to say it is for overclocking and you also get to pay extra for it to boot.

Martini1991 · 3 Mar 2014 at 19:32

almighty15 said:
http://iweb.tms.org/PbF/JOM-0606-67.pdf

Have a quick look....

I'm not going to try and digest that

But I say how AMD have bare face lied before, I don't think Intel are above it either (I don't even think that's Intels document, and it's from 2006.......)

We'll have to wait and see what the future holds, that's my standpoint.

OLDPHART · 3 Mar 2014 at 19:33

almighty15 said:
You finished reading the tread yet?

What makes you think they can't solder a smaller die size, have Intel really come out and said it no, where's the prove.

This is a overclocking forum not a stock forum.

Kaapstad · 3 Mar 2014 at 19:36

Martini1991 said:
I'm not going to try and digest that

But I say how AMD have bare face lied before, I don't think Intel are above it either (I don't even think that's Intels document, and it's from 2006.......)

We'll have to wait and see what the future holds, that's my standpoint.

The only bit of that article you have to digest is the date

June 2006 lol

This predates SB, SB-E, IB, IB-E and Haswell. Things have moved on since 2006.

almighty15 · 3 Mar 2014 at 19:45

Kaapstad said:
The only bit of that article you have to digest is the date

June 2006 lol

This predates SB, SB-E, IB, IB-E and Haswell. Things have moved on since 2006.

AMD, Intel and Nvidia all do research and planning years in advance...

Can't actually find the proper Intel one :/

Martini1991 · 3 Mar 2014 at 19:46

almighty15 said:
AMD, Intel and Nvidia all do research and planning years in advance...

Can't actually find the proper Intel one :/

Theory and practical are radically different, hence netburst.

Planning in advance only gets so far, they physically wouldn't have had anything near 22nm, let alone 32nm back in 2006!.

almighty15 · 3 Mar 2014 at 19:49

Martini1991 said:
Theory and practical are radically different, hence netburst.

Planning in advance only gets so far, they physically wouldn't have had anything near 22nm, let alone 32nm back in 2006!.

And yet were able to get the solder to degrade and crack on much larger process nodes...

Funny that all the big dies are soldered but the smaller, denser ones aren't :cool:

Martini1991 · 3 Mar 2014 at 19:54

almighty15 said:
Funny that all the big dies are soldered but the smaller, denser ones aren't

That in itself is not proof though.

Solder cracking is nothing new as well.

almighty15 · 3 Mar 2014 at 19:57

Martini1991 said:
That in itself is not proof though.

You can wait for all the proof you want but it's obvious to anyone that reads a few white papers and has basic understanding of metallurgy that small dies + high density = problems.

Kaapstad · 3 Mar 2014 at 20:06

almighty15 said:
You can wait for all the proof you want but it's obvious to anyone that reads a few white papers and has basic understanding of metallurgy that small dies + high density = problems.

You remember what they say about Assume

What is your background in metallurgy as a matter of interest ?

almighty15 · 3 Mar 2014 at 20:09

Kaapstad said:
You remember what they say about Assume

What is your background in metallurgy as a matter of interest ?

Studied it for a while.... personally crap happened... never went back into it..

I wouldn't say I'm an expert by any means but I know enough to understand papers, research and stuff.

I should get back into it....

Avenged7Fold · 3 Mar 2014 at 21:08

almighty15 said:
The interesting thing is how will AMD handle the issue? People have already blamed this TIM issue on them avoiding a lower process node.

In the past they have been and are currently using TIM on their APU's.

If people think that TIM over solder is what is holding back AMD from smaller nodes, they are idiots. I am sure the main thing holding them back is budget for development.

I don't doubt the solder can crack (still not convinced that soldering is impossible though) but there are a number of possible solutions than crap TIM. Why not something similar to CLU if it is an issue?

8 Pack · 3 Mar 2014 at 21:25

Tim saves money.

Who cares so long as they clock well.

setter · 3 Mar 2014 at 21:43

Avenged7Fold said:
Why not something similar to CLU if it is an issue?

This, ill probably end up doing it on mine. Temps hitting 74c in games at 4.5. No warranty on mine anyway so dont have to worry about that.

smilertoo · 3 Mar 2014 at 22:04

100mhz bump is useless if you already own a decent intel cpu, but if you're buying new then its a welcome little bonus.

chaparral · 3 Mar 2014 at 23:32

Terminal_Boy said:
Is there anyone actually impressed at all by the Haswell family of CPUs?

Am not the for sure...
All my i7-4770K benches I done can't even beat my old i7-2600k bench scores...

And my i7-4770 is also a crap overclocker..

setter · 3 Mar 2014 at 23:34

chaparral said:
Am not the for sure...

All my 4770K benches I done can't even beat my old 2600k bench scores...

What speed was the 2600k at, and whats the most stable for benching you eeked out of the 4770k? Curious as i found my hw at 4.2 to be roughly equal to my previous 3770k @4.5.

Dicehunter · 3 Mar 2014 at 23:58

Less than a 3% performance improvement over a 4770K, Guess this means I'll be sticking with my old 4770K for a while longer then.

humbug · 4 Mar 2014 at 13:38

almighty15 said:
Intel have already stated you can't use solder on things that small.... That's why they're looking into a high performance replacement or a more durable solder.

But the solder they use today, here and now does not work.

The interesting thing is how will AMD handle the issue? People have already blamed this TIM issue on them avoiding a lower process node.

There was an article on this somewhere, AMD said why they are avoiding the lower process node (GloFlo 20nm)

Its so small its causing more electrical resistance resulting in higher temperatures and less stability at higher clocks, efficiency improvement are ~20% at best and it costs; i think it was 48% more (vs 28nm)

AMD think they can get better APU's by sticking with 28nm, GloFlo 28nm has yet to mature, Excavator is still on 28nm.

About a year ago both AMD and Nvidia made the same complaints about TSMC's 20nm, which is why they are in no rush to move to it, you may have noticed both are still developing new architectures on 28nm while 20nm is already up and running.

I think if Intel were honest they would tell you they are finding the same problems as they go smaller and smaller.

We are starting to hit a dead end on silicon.