Intel haswell i7-4770K preview article-TMSHW

[nossy23]

ivy was around 10% faster than sb, so i'm guessing haswell will be roughly around 15% to 20% faster than sb.

I really was under the impression that Ivy was only about 5% faster than SB in most benchmarks. See for example here:

http://www.tomshardware.com/reviews/...k,3181-24.html

(this shows it's even only around 3.7% faster on average, and I know the base clock for the 2700K is 100 Mhz higher, but I would think under load both CPU's would be at about the same Ghz)

At the same clocks you can look here:

http://vr-zone.com/articles/ivy-brid...own/15637.html


The speed improvements are very low, far removed from 10%, for the most part going from Sandy to Ivy.


And I expect the same kind of clock for clock improvement going from Ivy to Haswell. So that would be around 10% from Sandy to Haswell if all goes well. Of course for a limited number of things the new CPU can be a lot faster.

[BababooeyHTJ]

And I expect the same kind of clock for clock improvement going from Ivy to Haswell. So that would be around 10% from Sandy to Haswell if all goes well. Of course for a limited number of things the new CPU can be a lot faster.

Why? We aren't looking at a tick here. Look at the improvement going from core2 to Nehalem or Nehalem to sandy bridge. Thats what I was expecting with Haswell. Haswell looks to be a pretty big disappointment so far.

Also, still no hex core on the "mainstream" socket? Really, after all these years?

[hersounds]

Haswell using the new power supply standard, the need to upgrade existing power


Haswell has done a lot of improvements for energy conservation


The Haswell processor for energy-saving made many improvements, including C6/C7 saving state required the reduction of the current. There have been too much processor power, the problem of shortage of power supply, and now the current is too large will be a problem. The existing ATX power supply with Haswell what problems will arise? What kind of power supply in order to meet the needs of the Haswell?

The Hasell supply improved

Earlier today, we reported Haswell processor for energy-saving done a lot of improvements, including C6/C7 saving state required the reduction of the current, minimum current requirements reduced from the current third-generation Intel Core processor 0.5A to ten one of the stars, only 0.05A. We finally came to a special period of time, used to be too much processor power, shortage of power supply, and now is the minimum current is too large. Using existing ATX power problems that will arise? What kind of power supply in order to meet the needs of the Haswell?


The Haswell minimum current requirements from 0.5A to 0.05A

The current Intel ATX12V v2.31 design guide book from the Lynnfield period for the processor power supply +12 V2, all the way to the minimum requirements of the output current is 0.5A. Intel has not released an updated design guide book, but this is related to the electrical characteristics of the power supply itself is not a design specification can be solved, our analysis together.


Intel ATX12V V2.31 specification of 300W power output requirements

Low current to reduce problems that may occur

Haswell processor to enter C6/C7 power-saving state, the +12 V2 the required current will be greatly reduced, and a host of other accessories will enter a power saving state to reduce power consumption. Mainstream power supply +12 V2 with +12 V1 is in fact the same route, this time need to focus on not just the load on the +12 V2 +12 V, but the total load.

Single magnetic amplifier power, for example, if the +12 V total load is too low, the power supply may be from the continuous conduction mode (CCM Continuous Current Mode) into the discontinuous conduction mode (DCM, Discontinuous Current Mode), in the continuous conduction mode into discontinuous conduction mode, the same duty cycle will result in a high voltage, in order to stabilize the output voltage in this case is necessary to significantly reduce the duty cycle of the +12 V drop back to normal, but this time +5 V the voltage will follow the decline may occur when stabilized +12 V High +5 V low.

To conclude, a single magnetic amplifier, dual magnetic amplification and DC-DC power supply will appear the following conditions:

The magnetic amplification: +12 V and +5 V output voltage significant changes possible +5 V is pulled low;
Dual magnetic amplification: In addition to the +12 V, +5 V and +3.3 V output voltage may be pulled low;
DC-DC: +5 V and +3.3 V low-voltage two-way will not be affected, of course, the power advantage here;

The output voltage is too low may occur is to crash or the host can not wake up from a power-saving state. The introduction of the the magnetic amplification technology, and the DC-DC technology can refer to our article: " PC power supply Introduction and advanced, from the magnetic amplifier voltage stability of the technical talk , Super annual Juxian: seven high-end power Hengping The test papers , DC-DC output technology introduction. "

Actual use readers can go back and read sections of the power evaluation cross-load point A light load, a power +12 V total load point A fixed 5.8W (12V2 output 0.242A) look the brightest voltage. The following example is some time before our silver the Yan ST45SF-G watch point A abscissa load 5.8W.


450W power supply cross-load pull out the chart, pay attention to point A


Cross-load performance DC-DC power supply low-pressure two-way will not be affected at light loads

Conclusion: ATX power Haswell is not a big problem

As host of other accessories, including motherboards, graphics cards and hard disk more or less there will be a certain load, the total load of +12 V at light loads generally are higher than our test point A. So small that the current ATX power Haswell processor will not be much of a problem (cottage power, and our evaluation of cross-load failed except). We can safely open the processor C6/C7 energy saving, voltage output deviates too big or system instability, show, you can try to pick a few +12 V chassis fan to increase the +12 V load power supply out of discontinuous conduction pass mode in order to solve this problem.

Source: http://www.expreview.com Using Google Traductor

[nossy23]

Why? We aren't looking at a tick here. Look at the improvement going from core2 to Nehalem or Nehalem to sandy bridge. Thats what I was expecting with Haswell. Haswell looks to be a pretty big disappointment so far.

Also, still no hex core on the "mainstream" socket? Really, after all these years?

I completely agree with you here. I would have expected a lot more performance, but it doesn't seem to be the case. Really disappointing. Even no extra cores. Looks like Intel does not want to take my money, so I bought me a Geforce Titan instead.

[FlanK3r]

or disable lowest C-state (if someone has poor PSU)

[Olivon]

4770K left, Xeon E3-1280 V3 right

Some time ago once there are friends broke the news to us that Haswell core i7-4770K can support FSB frequency setting operation, and send an FSB to 184.82Mhz CPU-Z screenshot up.

http://www.chinadiy.com.cn/html/45/n-8845.html



http://www.chinadiy.com.cn/html/75/n-8775.html

[rk7p5]

Maybe it's too early to tell but in those cpuz shots it looks like intel is limiting the multipliers if you want to use the fsb multi's. This is really looking sad. I don't need to upgrade for more performance, I want to upgrade for a better platform and for some overclocking fun.

[behrouz]

Haswell using the new power supply standard, the need to upgrade existing power
.....................

Report: Not all PSUs support Haswell's low-power states


WTF? Too much little power From PSU?
Hopefully one day , we will see PSU with firmware that can update via Web.

[PatRaceTin]

ahhhh FSB multi may be some fun DDR3 overclocking

[Olivon]

http://www.comptoir-hardware.com/act...rclocking.html

Haswell doing 3'333MHz on IMC piece of cake - OCaholic



Thanks stasio !

[stasio]

More interesting info:

http://www.ocaholic.ch/modules/news/index.php

[sin0822]

wow today must have been lets leak day hahaha.

[Dumo]

No leaks with Gigabyte board?

Intel must be pissed.

[paulbagz]

Omfg dat RAM speed lol.

Be nice to see leaked pics of the Maxi VI.

-PB

[paulbagz]

lol spoke too soon!;



-PB

[Zoran]

Do you have pics of the Maxi VI Gene too? Thanks in advance

[Alex-Ro]

haswell launch is going to be very exciting it seems ...

[sholvaco]

lol spoke too soon!;

http://cdn.overclock.net/9/90/900x90...7d_i84ipn.jpeg

-PB

Like OMFG is that a good looking board!

Anyway, here's ASUS offerings:

http://wccftech.com/asus-z87-motherb...hoon-pictured/

[STaRGaZeR]

One of these Maximus VI Gene will be mine!

[informal]

At least boards look semi exciting

[Olivon]

http://www.redquasar.com/forum.php?m...read&tid=12622

via WCCFTech

[zalbard]

Preview by ChinaDIY -> HERE

[Olivon]

Core i7-4770K Review - ChinaDIY



1.4v in BIOS / 1750MHz IGP overclocking / IB-HW IGP comparison

However, we measured, the air-cooled overclocking 4.5G long-term stability is very difficult to, only 1.13V 4.5GHz frequency pager, easily exceed 90 ? C in the pursuit of the highest frequency of the air-cooled, 5.1GHz CPU-Z certification is still possible, but carried Super PI 1m test, for the sake of overheating causes blue screen.

After much debugging, I found the hands of Fengyun i7-4770K processor to a minimum 1.13V voltage, CPU to 4.5GHz frequency stability by baking machine test by IntelBurnTest software; Also, with the 2400rpm fan of violence blessing, CPU each core under full load average temperature of 81.8 ? C, not suitable for daily long-term use.

Our with 1600rpm speed fan to suppress 1.13V i7-4770K processor clocked at 4.5GHz, core temperature is maintained at about 92 ? C, long-term use? Anyway, I would not dare.

Cooler is Prolimatech Megahalems Basic-81 for this review.
Haswell seems to be hot while overclocking on air

edit : burned by Zalbard

[informal]

I wonder what is causing these high temps? Is it bad ihs compound or the way intel attaches die to ihs? In any case it's bad news since everyone hoped for higher OCs on air Vs IB. Looks like SB, ironically, will OC higher on air vs both Haswell and IB.

[Andi64]

So, less than 5% average improvement over 3770K, and less than 1% on games. I'm glad I went LGA2011