Haswell info

[ICIT2LOL]

Don't know if anyone else has seen this but it is starting to look real good for the next build.

Haswell: Cheat Sheet | TechRepublic

 

[paulpicks21]

As long as your not into heavy overclocking then they will be fine. I have been keeping a close eye on 2 forums where lots of members now have the K series and they are getting very high heat with anything over 4.4ghz, many chips won't OC to past 4.5ghz either.

Paul.

 

[ICIT2LOL]

I don't OC Paul but that is worth knowing as that chip I was always given to believe that these were going to run that much cooler as I was intending having a go at it when I build. Now I'm not so sure of what to do..

 

[Lady Fitzgerald]

I've been taking a hard look at Haswell systems, especially the ASUS MOBOs, to see how they compare to my present build (which I'm not replacing—I've too much money tied up in it—but I was curious to see if I was wise to go ahead and build when I did instead of waiting). The X87 Deluxe Dual, X87 WS, and X87 Deluxe, in that order, especially caught my attention. They could meet my needs, albeit barely, but I have concerns about the CPU itself.

Intel is using a TIM between the chip and the heat spreader, like they did with the Ivy Bridges, instead of soldering, like the Sandy Bridge-E and earlier. The Ivy Bridge CPUs have had problems with the chips overheating due to poor heat transfer from the chip to the heat spreader, resulting in many owners delidding them so the TIM can be replaced. The new Haswell CPUs are also not soldered and there already has been a handful of early user reports about them running hot.

I don't regret going with an X79 build instead of waiting for the Z87s. I needed a new desktop back when I built mine. I would have had to wait at least five months for the Haswells (my two top choices still aren't available yet) and it's always fiscally wise to not be a first adopter of anything since the first runs are usually buggy so I would have wound up waiting a year and wound up with a system that wouldn't have been much better than what I have now (albeit less expensive, though). Still, it's interesting to see what is, and is going to be, out there.

 

[JackALope52x]

Thanks for the link John, that's pretty interesting. I haven't really read too much about the new Haswell processors and this really shined some light on them for me. The overheating problem and lack of OCing potential is a little depressing though.

 

[Layback Bear]

The market for those who want to clock over 4.5 is a very small so I don't think that will be a concern to Intel.
Having a much faster cpu using a lot less power to me is the big step forward.
I will be watching as time goes on of any cooling problems.
Time will only tell.
If the facts are really facts about the better use of battery life it will be a great improvement. This improvement along should reach out and grab market share.
Us procrastinators will just wait and see.
Thanks for the information ICit2lol!

 

[FuturDreamz]

I've heard that the reason why intel uses TIM instead of solder is because the chips are too small for heat to disperse evenly, so the solder is cracking/lifting and so limiting surface contact. Don't quote me on that though.

If that is true, then all Intel has to do is to make the chips bigger, preferably by adding more cores (maybe we can see 16-core haswell-E processors?)

That being said, I'm looking forward to building a tiny gaming computer using the i7-4770R with iris pro. It won't knock your socks off, but it will handle all the games I like.

 

[Lady Fitzgerald]

I've heard that the reason why intel uses TIM instead of solder is because the chips are too small for heat to disperse evenly, so the solder is cracking/lifting and so limiting surface contact...

Sounds like marketing department spin. Funny how the delidded chips work better.

 

[ICIT2LOL]

Hum making me have second thoughts now I was sorts attracted by the RAM speed specs on the boards and I suppose when you think about it logically because the architecture is so small heat is going to be even more of an issue.

By that I mean on the actual die for example the minute connections and components that I assume are in the tiniest sizes one would care to think about would be very vulnerable to currents applied to them.


For example say a 10v x 10 amp current passed through a 10mm wire might cause it to get warm pass that same current through a 1mm may possibly melt it. After all it doesn't require O2 for melting to occur.

The heat spreader issue also seems to me one that the manufacturers have rushed into I would prefer to pay a tad more to know that the chip is fully in contact with the entire heat spreader.

Plus we also have to remember we are paying a reasonably good prices for these devices.

 

[paulpicks21]

Here is a small article from today on Haswell and heat - Haswell heat surprises system builders | bit-tech.net

 

[ICIT2LOL]

Here is a small article from today on Haswell and heat - Haswell heat surprises system builders | bit-tech.net

Neat article Paul now it really has got me thinking re OC because to me and I might have it completely wrong but OC is for pure experimenting with or using because you can.

My current i5 Ivy is really quite fast in a standard setting and I have no real idea what I would see if I say took the clock speed to say 4Ghz. So is OC purely for those who game or does it really put the pedal to the metal for everything?.

 

[paulpicks21]

Really with all the latest gen i7's and i5's OC'ing is for fun, they are usually powerful enough at stock to run anything you might need. I believe overclocking the CPU can be useful if say you have 2 or 3 GPU's as it can reduce a bottleneck.

I have my Ivybridge overclocked at 4.6ghz and I had my previous Sandybridge running at the same, they run well and never exceed 65C under normal use so I keep it as a 24/7 OC. I use speedstep so when the CPU is not being used much it will idle at 1.6ghz with low voltage and it steps up to 4.6ghz when needed.

The article backs up my point that if you are not into OC'ing then they will be fine but if you enjoy OC'ing like many of us do then it's not the chip for you.

I was surprised to read that even at stock Haswell in most cases is running 15C hotter than the pre-production engineering samples and also surprised to read that one manufacturer was unable to to reach 4.2ghz at safe temps with any of the 40-50 retail units they got.

Paul.

Edit: added i5

 

[ICIT2LOL]

Thanks Paul that was very interesting and easy to understand explanation - I for one thoroughly appreciated it

Again I suppose it goes back a bit to what I was referring to with large currents through small components.

 

[Layback Bear]

I'm some what curious why the pre sale chips (R&D) given for testing worked so well and cool and the for sale chips (retail) did not perform as good and were hotter.
What corners are Intel cutting to mass produce a chip with such gross difference between tested chips (R&D) and chips ready for retail. The fact that the R&D tested chips worked so good tells me Intel has the ability to make a chip that can do the things it says it can do.
Somebody dropped the ball between R & D chips and ready for Retail chips.
****

For me a simple home PC user I don't care if the cpu is the size of a pack of cigarettes. If it does more and runs cool is really all I care about.
The motherboard, and cpu cooler companies will make things that work with the big chip and all of it will fit in my Tower with no problems.
The few dollars of extra electricity it will use in a year is nothing to worry about.
Taking your kids to Micky D's one time for Happy meals will cost more.
Everything doesn't have to be made smaller just because you can.

 

[sygnus21]

I don't regret going with an X79 build instead of waiting for the Z87s. I needed a new desktop back when I built mine. I would have had to wait at least five months for the Haswells (my two top choices still aren't available yet) and it's always fiscally wise to not be a first adopter of anything since the first runs are usually buggy so I would have wound up waiting a year and wound up with a system that wouldn't have been much better than what I have now (albeit less expensive, though). Still, it's interesting to see what is, and is going to be, out there.

I find myself in this same boat as I'm looking to build a new system. I'd initially put together a x79 system but others advised me to wait and see what Haswell has to offer.

I've got a few months to figure it out as I get my finances in order. By then there should be a bit more info and benchmarks to make a more informed decision by then. In the end though I may wind up going X79 as I'm am somewhat leery about new tech, especially motherboards.

We'll see.

 

[ICIT2LOL]

Well I think you make some pretty darn valid points Bear because if for example the machine is using say 200watts at 240v why on earth are we concerned about say 25 watts at 12v?? the difference is just about insignificant and as said before it all seems to be playing around semantics really.

Again I will say that small components are bound to get hotter by the very nature that they are small. Plus I am guessing the test /mock up trial processors were far more stringently build controlled than what we are likely to buy. For example I could go buy a car of the same model as Joe Blow and the quality, efficiency, and longevity be very different indeed - dare I say a Friday made processor LOL!!

 

[Layback Bear]

It seem to me quality control during mass production is the problem not the design per say.

 

[ICIT2LOL]

It seem to me quality control during mass production is the problem not the design per say.

Totally agree goes across the board and when you see the sites particularly in the major manufacturing countries and the conditions in them it is little wonder the quality is not as I said so stringent as testing. Plus I suspect that maybe either the product brand dictates how well the products are made and some made by the actual production people cutting corners / not following production guidelines.

 

[Britton30]

Here's one explanation of the Haswell heat, the chip have the voltage regulator on the die, rather than the motherboard. Good for mobile computers bad for desktops and overclocking.
Going from a Sandy Bridge to Haswell may be a good idea but not from an Ivy Bridge.?

http://www.extremetech.com/computin...cpu-in-the-world-unless-youre-a-pc-enthusiast

 

[Dude]

Reminds me of the Prescott release, them bad boys ran hot!!