[PCPro] Haswell vs. Haswell ES Overclocking

[BallaTheFeared]

4 OEM system builders have no reason to talk down a major product that their businesses depend on unless there is a problem.

Who was it?

Oh right, "anonymous"... Looks like their feelings weren't enough to stake their reputation on.

 

[Fx1]

Who was it?

Oh right, "anonymous"... Looks like their feelings weren't enough to stake their reputation on.

Not very smart.

You dont put your business name on record slamming intels latest CPU or you might find you can no longer buy them at wholesale prices. Also Intel pulls its rebates and other support and suddenly your business is dead.

Another Balla classic comment.

 

[Lonyo]

Who was it?

Oh right, "anonymous"... Looks like their feelings weren't enough to stake their reputation on.

http://www.extremetech.com/computin...-core-i7-4770k-takes-over-the-pole-position/3

Just because PCPro decided not to reveal who they spoke to, and neither did Extremetech, doesn't mean all of them said "don't mention our name".
For all we know, it could be editorial policy not to mention sources where not required.

You are assuming that because PCPro decided not to name names, they are "anonymous".
Well, why not provide some evidence that it's all made up and that Haswell is better or equal to IB at overclocking?

I would take the word of PCPro saying 4 boutiques told them something over you saying "they are anonymous sources and therefore they are talking crap".

 

[BallaTheFeared]

Not very smart.

You dont put your business name on record slamming intels latest CPU or you might find you can no longer buy them at wholesale prices. Also Intel pulls its rebates and other support and suddenly your business is dead.

Another Balla classic comment.

Without transparency it's worthless page hit material.

 

[rgallant]

http://hardforum.com/showthread.php?t=1765130

Grain of salt and all that but it's people sharing their retail results.

-lol not many posting their ram speed
-asus vid said there might be a trade off of high ram speed and high oc's eg. 2400 + 4.7 won't boot but 2133 + 4.7 will.
-so they might be turning up the vcore instead of turning down the ram speed.

-same thing happened with ib people turning up vcore on beta bios's looking to be first to post their golden chip clock and temp. only to declare ib is a dud after a hour.
-if I recall correctly ib es samples were from 09-2011 ,most retails on release were 11-2011. later ib's seem to get better + more data points.

 

[Abwx]

Without transparency it's worthless page hit material.

Because intel was transparent on the matter , isnt it..?.

That s why reviews and press exist , mind you...

That said , only the ocking crowd will be left
somewhat disapointed , the rest of the users
will see nothing.

 

[VirtualLarry]

If you are basing your purchase off of a review and an expected overclock, that's your biggest mistake.

But what about the boutique builders, that sell (or in this case, planned to sell) pre-overclocked systems?

ES chips are meant for these mfgs to qualify the new CPU chips for their systems.

In this case, those chips were NOT representative of retail production chips, which greatly affected (hurt) these boutique builders. Intel really screwed them over, and their own overclocking reputation beside.

No matter how you look at it, retail Haswell is a dog for overclocking. Might as well stick to Ivy.

 

[Fx1]

Without transparency it's worthless page hit material.

Yeh because without proof nothing ever happens.

Proof will be in the next few months when 4.2ghz becomes the default OC of the haswell CPU.

 

[Lepton87]

If you really expect an overclock, just buy 20 chips, test them and keep the best, return the rest. That's what I would do if I really wanted an over-clockable product. Next time I'm gonna do just that, I'm sick and tired of getting garbage-bin products, for example my Titan doesn't OC past +50/300MHz, it's fine at 130/300 in some games but total stability is only at +50MHz.
ps. I wouldn't expect a new stepping that improves OC. When was the last time they did that? Almost 4 years ago? (920 G0) The new stepping of SB-E only repaired what was broken, I don't expect anything more from a company that puts a gap between a CPUs die and its IHS. WHY? Higher temps should lead to more silicon degradation thus increase RMAs. Maybe that effect is too miniscule for them to worry.

This is very close to fraud. Don't do this
-ViRGE

 

[BallaTheFeared]

Seems people have Sandy Bridge Glory Days syndrome.


Sandy Bridge ES overclocked better and didn't need Internal PPL overvoltage.


Sandy Bridge 5GHz clubs aren't huge lists, and most chips need 1.45v or more without even AVX Prime, and needed water just to run AVX at those voltages/speeds.


Why don't you give it more than a few days before we start drawing conclusions? The average for SB on air was probably 4.4GHz, let's try to keep this in perspective.

 

[SOFTengCOMPelec]

In the past, as I remember it, Intel often released processors, which were not the best of overclockers, until about 3 .. 12 months (or longer) after initial release. As it took them time to iron out slight problems, and get all the chinks out of any digital high speed timing issues. Hence, my memories of the old days were that we (overclockers) had to (ideally) seek out the best (certain) chip batch numbers, which would be most likely to overclock the best.
Haswell has only just been released, maybe we have to give Intel time to improve (optimize) their chip.
They may also have to reconsider the heat sink (T.I.M.) issue. I don't know how viable going back to soldered (T.I.M.) is on these 22nm chips.

 

[tential]

Who was it?

Oh right, "anonymous"... Looks like their feelings weren't enough to stake their reputation on.

A LOT of people release things anonymously to stay protected. This comment sounds like you're just looking for something to complain about rather than learning the realities of the world.
I expect more of you.
You can anonymously whistle blow in MANY situations, and the press quotes anonymous sources all the time to protect their source from losing their job, and their livelihood as sometimes if you lose your job you can't find another one because you blew the whistle, even in situations where you were correct to do so.

Not saying whether this is true or not with the ES overclocks, just talking about anonymous sources in general.

 

[BallaTheFeared]

I have no doubt that ES Haswell overclocked better than retail, they're probably soldered too.

You're comparing binned validation chips from a small sample pool against retail, there are going to be retail chips when properly loved and cared for that blow the ES chips away too.

This is overclocking it's always a lottery, I had three SB chips, one did 5GHz at 1.5v but no higher no matter the voltage, another did 5,278MHz at 1.528v and another did 5.5GHz at 1.528v. I've seen plenty of people with chips that crapped out around 4.3GHz as well.

I don't see anything new here, what's the big conspiracy?

 

[Vesku]

-lol not many posting their ram speed
-asus vid said there might be a trade off of high ram speed and high oc's eg. 2400 + 4.7 won't boot but 2133 + 4.7 will.
-so they might be turning up the vcore instead of turning down the ram speed.

-same thing happened with ib people turning up vcore on beta bios's looking to be first to post their golden chip clock and temp. only to declare ib is a dud after a hour.
-if I recall correctly ib es samples were from 09-2011 ,most retails on release were 11-2011. later ib's seem to get better + more data points.

Yeah, I actually take those retail results to mean ASUS was pretty spot on with their estimates. Going to be the rare Haswell K that runs 4.6-4.7 stable on air at least without making some of the choices you have mentioned.

 

[Lonyo]

Without transparency it's worthless page hit material.

If transparency was required, no one would say anything for fear of getting shafted in the long run after they annoy whoever is giving them the thing they are making money from.

It's not the best plan to go and say "Intel lied to us and are messing us about with these new processors" and then expect Intel to keep treating them as nicely as they did before.

Transparency is nice in an ideal world, but because transparency ISN'T required, we are able to get this sort of story. Don't be so naive.

 

[tential]

I have no doubt that ES Haswell overclocked better than retail, they're probably soldered too.

You're comparing binned validation chips from a small sample pool against retail, there are going to be retail chips when properly loved and cared for that blow the ES chips away too.

This is overclocking it's always a lottery, I had three SB chips, one did 5GHz at 1.5v but no higher no matter the voltage, another did 5,278MHz at 1.528v and another did 5.5GHz at 1.528v. I've seen plenty of people with chips that crapped out around 4.3GHz as well.

I don't see anything new here, what's the big conspiracy?

I agree with this completely. If you seriously base your OCing results on Engineering Samples you really didn't do your homework. I don't know why OEMs even feel so hurt. All they had to do was purchase 10 retail chips, see how they OCed, and base what they sold on that. Why you would assume that you'd get overclocks that you saw in Engineering Samples is beyond me.

I expected to hit a little over 4 Ghz. It looks like OEMs over promised, and now are mad that they may have to under deliver. It's hard marketing to convince the average person to pay more for a 4.2 Ghz OCd Haswell compared to a 4.4Ghz OC'd ivybridge considering people care more about the number and less about which is actually faster but I'm sure they'll think of something.

Besides, intel wants to sell them faster chips anyway rather than have OEMs purchase a bottom of the line chip and OC them now they'll have to buy faster chips.

Intel is just doing what's best for them just like OEMs were doing what was best for them by buying cheaper chips and OCing them to get performance.

 

[WhoBeDaPlaya]

http://hardforum.com/showthread.php?t=1765130

Grain of salt and all that but it's people sharing their retail results.

I tested four chips from the local MC, on both an MSI Z87-GD65 and Gigabyte Z87-UD5.
Consistent results between the two boards :
http://hardforum.com/showpost.php?p=1039938086&postcount=50

 

[Vesku]

I tested four chips from the local MC, on both an MSI Z87-GD65 and Gigabyte Z87-UD5.
Consistent results between the two boards :
http://hardforum.com/showpost.php?p=1039938086&postcount=50

Cooling used, Antec Kuhler 920 in your sig? Temperatures?

Thanks for contributing.

 

[rgallant]

chip 3 is better[lower vcore] than my ib at 4.6-4.7

 

[notty22]

It's really not OEM's either. It's supposed 4 anonymous Boutique sellers from UK. I can't even name 4 boutique sellers in the USA that sell guaranteed o/c's. Maybe Falcon? It is a grab at web hits with a semi-sensational story.
We have K cpu's coming in at the same price as old K chips they replace. With a new socket and bios's for support for new features. It's not surprising to me that some tests show less efficiency and lower clocks in these early days.
In the launch of Lynnfield, we had sockets melting from, cpu/bios's that were new and not optimized. And plenty of fear mongering.
By the way, what is 'misleading again' and who is 'they say'.
I also heard 'they say' that a K Haswell chip replacing K IVY cpu, it's a upgrade.

 

[tential]

I tested four chips from the local MC, on both an MSI Z87-GD65 and Gigabyte Z87-UD5.
Consistent results between the two boards :
http://hardforum.com/showpost.php?p=1039938086&postcount=50

Could I hit my planned 4.2 Ghz OC with a stock cooler? Or even 4 Ghz?

 

[Ayah]

Pretty amusing article considering AFAIK Intel never guarantees overclocking performance/thermal behaviour.

 

[LagunaX]

Intel has been sending out ES samples which are far superior than the retail chips.

Back to its old tricks.

4.2ghz and 15c hotter at stock! than ES sample chips sent for review.

http://www.pcpro.co.uk/news/382267/intel-haswell-hotter-and-slower-than-expected

I was looking for a haswell build and now im going to wait to see what happens.

Francois was F.O.S...
https://twitter.com/FPiednoel/statuses/324389068186210304

 

[Gillbot]

Why?

I base all my purchases off reviews and expected overclocks. Plus other information.

Nothing is guaranteed... I never base my purchase off of a review or what someone else claims to have done. Too many sandbaggers out there exaggerating their overclocks for e-fame.

But what about the boutique builders, that sell (or in this case, planned to sell) pre-overclocked systems?

ES chips are meant for these mfgs to qualify the new CPU chips for their systems.

In this case, those chips were NOT representative of retail production chips, which greatly affected (hurt) these boutique builders. Intel really screwed them over, and their own overclocking reputation beside.

No matter how you look at it, retail Haswell is a dog for overclocking. Might as well stick to Ivy.

Boutique builders will have to test and verify the system before they sell it overclocked. I can assure you that none I know of, and when I built systems, I would NEVER base anything off of an ES evaluation. Most system builders know better, and they are also keyed in to more information than what leaks into the general populous. They have a batter handle on how much headroom there is, if any.

 

[tweakboy]

Haswell is 5 to 10 percent faster then Ivy Bridge QuadCore Performance of CPU and a really fast built in video card is all this is. You will disable the onboard card. gl