Whats after Haswell?

WingZero30 · 15 Apr 2013 at 22:54

Yeah clock-for-clock i7 4770k seems to be around 7-10% faster than i7 3770k.

However if the speculation of haswell achieving 8Ghz 'easily' can be verified as fact then that would be huge performance!! (highly unlikely

)

Martini1991 · 15 Apr 2013 at 22:55

There isn't any speculation of Haswell hitting 8GHZ easily.
Only that multiplier goes upto 80?

WingZero30 · 15 Apr 2013 at 23:03

Hence I put 'easily' to emphasise that it would be a near impossible task to achieve 8Ghz as the guy who wrote the article on IDF beijing gave a speculative mentioning of haswell achieving 8Ghz

lettuce · 18 Apr 2013 at 19:55

So my aging i7 920 Bloomfield @ 4Ghz, would Haswell be 40% faster at the same OC?....also Haswell would probably achieve 4.5Ghz without too much hassle?, so then i could be looking at 60% performance increase????

silvagti · 18 Apr 2013 at 22:26

lettuce said:
So my aging i7 920 Bloomfield @ 4Ghz, would Haswell be 40% faster at the same OC?....also Haswell would probably achieve 4.5Ghz without too much hassle?, so then i could be looking at 60% performance increase????

60% faster than an i7-920 sounds about right based on the benchmarks I've seen.

Whatever happened to Moore's law where CPU performance used to double every 18 months! Now it's taking 4 years

melvis · 18 Apr 2013 at 22:42

correct me if im wrong but moores law is about transistor density not performance and still on track afaik

jonneymendoza · 19 Apr 2013 at 00:11

bru said:
I'm not quite sure where you get that idea from.
.

the idea came from people here who said haswell wont be more then 20% faster then my 920 clock for clock

Orcvader · 19 Apr 2013 at 00:33

melvis said:
correct me if im wrong but moores law is about transistor density not performance and still on track afaik

It is. Nothing about performance at all.

http://en.wikipedia.org/wiki/Moore's_law

Moore's law is the observation that over the history of computing hardware, the number of transistors on integrated circuits doubles approximately every two years. The period often quoted as "18 months" is due to Intel executive David House, who predicted that period for a doubling in chip performance (being a combination of the effect of more transistors and their being faster).

PhysicsMan94 · 19 Apr 2013 at 01:22

Drangueos said:
not only the nm process but the speed that some materials can work.
"soon" they will move away from silicon.

That is where things will start getting very interesting - moving away from the conventional silicon-based circuitry in computer components that we've seen pretty much ever since the invention of computers in the first place.

Wonder what's next? Perhaps a move towards more "organic" oriented computing systems given that someone earlier mentioned DNA being used in research.

Martini1991 · 19 Apr 2013 at 08:34

jonneymendoza said:
the idea came from people here who said haswell wont be more then 20% faster then my 920 clock for clock

But it clocks further.
No one's said there's going to be 20% maximum difference in performance.
Haswell might only raise the IPC bar another 5%, but it could raise the OC'ing bar another 5% too.

illuz · 19 Apr 2013 at 10:16

Haswell is meant to clock like a dream, if you read through the reddit AMA with the Intel Blue Badge he also said that if you're an overclocker, you'll love Haswell.

Production quality of Ivy Bridge chips has also stepped up recently, which may give an indication of what to expect with Haswell. There's been a batch made recently by Intel which have been reaching *ridiculous* clocks on stupidly low voltages, golden chips. v1.22 for 5GHz 3570ks, reaching up to 5.4GHz or so v1.4. ON AIR.

Heres some clocks: http://www.overclock.net/t/1368961/newest-batch-of-3570ks-ocing-like-crazy

PhysicsMan94 said:
That is where things will start getting very interesting - moving away from the conventional silicon-based circuitry in computer components that we've seen pretty much ever since the invention of computers in the first place.

Wonder what's next? Perhaps a move towards more "organic" oriented computing systems given that someone earlier mentioned DNA being used in research.

When we move to Graphene, that will show us some extreme improvements and allow us to ramp the voltage right up as it is a hell of a lot tougher than silicon.

jonneymendoza · 19 Apr 2013 at 11:21

Martini1991 said:
But it clocks further.
No one's said there's going to be 20% maximum difference in performance.
Haswell might only raise the IPC bar another 5%, but it could raise the OC'ing bar another 5% too.

clock for clock its only like 10% faster though

Martini1991 · 19 Apr 2013 at 11:23

jonneymendoza said:
clock for clock its only like 10% faster though

It'll probably be closer to 20% than 10%.

gronf · 19 Apr 2013 at 11:23

jonneymendoza said:
clock for clock its only like 10% faster though

U sure?

jonneymendoza · 19 Apr 2013 at 11:24

jomiyo said:
U sure?

According to people in this thread. yes

Martini1991 · 19 Apr 2013 at 11:44

jonneymendoza said:
According to people in this thread. yes

No one's said that though.

Wucked · 19 Apr 2013 at 12:49

wolvers said:
U sure?

Yes.

wolvers said:
U sure?

?

wolvers said:
U sure?

Maybe.

wolvers said:
U sure?

Not any more

wolvers said:
U sure?

/Facepalm

I think Haswell will be a worthy upgrade for i7 9XX. It'll be a faster and more efficient CPU that supports a new instruction set, on a new socket with an upgrade path to Broadwell. And from what I've read it'll be good for overclocking too, hopefully even better than SB. So perhaps nothing to wet your pants about in terms of IPC, but still looking pretty good to me.

Sorely tempted to build a mini-ITX system for Haswell, monolithic gaming towers of doom are unnecessary now. You can fit some serious performance into SFF chassis atm, and by the looks of it Haswell is set to make it even easier.

snips86x · 19 Apr 2013 at 12:59

Martini1991 said:
There isn't any speculation of Haswell hitting 8GHZ easily.
Only that multiplier goes upto 80?

WingZero30 said:
Hence I put 'easily' to emphasise that it would be a near impossible task to achieve 8Ghz as the guy who wrote the article on IDF beijing gave a speculative mentioning of haswell achieving 8Ghz

If anyone can get to 8ghz, it'll be 8Pack!!

V3teran · 24 Apr 2013 at 12:39

Kaynar said:
I got an i7 930 and im upgrading to haswell at the start of next university year. While clock-clock there is not much difference, its hard to go over 4Ghz with a 920 or 930 while its dead easy to go at 4.5Ghz+ on current CPUs and from what I read today Haswell makes high-end OC even more achievable with expensive motherboards due to new flexible VRM systems. That means we may reach near 5Ghz without much effort under a simple AIO wc system. Also don't forget X58 has problem with Sata-3. I see that on my SSD every day. Write speeds and random seek is not as good as with newer CPUs. The list of updates could keep going but for sure its not a dramatical upgrade like we were used to once upon a time.

I expect a highly overclocked 4770K to be more than 50% faster than a 4Ghz i7 920/930.

I have a 930 running at 4.4 with HT on fully stable in linx,prime 95,occt and Ibt.

It is on custom water, if i had higher memory and turned Ht off i reckon i could hit 4.5 easy.

Haswell is not required.

jonneymendoza · 24 Apr 2013 at 13:02

Martini1991 said:
No one's said that though.

read the thread