Con Lake Con firmed [Warning: AdoredTV]

[jigger]

What basis do you have to say that Intel sent reviewers binned coffee-lake chips?

They don't add up with retail chips. 8pack was getting less than a handful per 100 IIRC to perform at review levels.

Although I don't know 8 pack very well I'd say he'd wipe the floor with the average youtube reviewer and most of the so called big tech sites when it comes to configuring a system for overclocking. Come to think of it I think Intel was asking him for advice for bining silicon.

 

[gavinh87]

They don't add up with retail chips. 8pack was getting less than a handful per 100 IIRC to perform at review levels.

My retail must be binned then.

 

[AndreiD]

@Jono8 You're wasting your time, all he does is just repeat speculation and rumors.
In fact, the few outlets that actually tested the ES Intel provided chips vs retail found the opposite: https://www.io-tech.fi/artikkelit/core-i7-8700k-retail-vs-engineering-sample/
Finnish outlet found that the retail 8700K ran at a lower voltage and thus was cooler and used less power than their Engineering Sample 8700K that was provided by Intel. Retail chip was also a better overclocker.

 

[AmateurExpert]

If they are buying medion PC's or a HP 'gaming' PC they are getting exactly what they deserve

Its like any performance part if you put it in a mediocre system you can't expect the headline performance the reviews show

Absolutely. But the elephant in the room is that unless competitors use the same grey area to list conservative base clocks while simultaneously profiting from benchmarks on systems that gain favourable (and often invisible) boosts in the grey area (all-core performance) then the playing field is not level in terms of comparison between CPU models, and also between fully-built systems.

 

[jigger]

@Jono8 You're wasting your time, all he does is just repeat speculation and rumors.
In fact, the few outlets that actually tested the ES Intel provided chips vs retail found the opposite: https://www.io-tech.fi/artikkelit/core-i7-8700k-retail-vs-engineering-sample/
Finnish outlet found that the retail 8700K ran at a lower voltage and thus was cooler and used less power than their Engineering Sample 8700K that was provided by Intel. Retail chip was also a better overclocker.

The Finish site shouldn't be reviewing ES chips at all. Very unprofessional.

 

[AndreiD]

That's just a misunderstanding with how Intel's Turbo Boost Technology works, it's opportunistic in that a lot of factors can influence max clocks, workload, current, voltage, cooling, TDP limits, etc... Usually the limiting factor is either TDP or cooling.
Two chips should perform the same given a similar scenario and environment, the issue is that comparing a chip from a review which most likely used adequate cooling with what you find in pre-builts isn't going to give you the same results at times.

 

[jigger]

That's just a misunderstanding with how Intel's Turbo Boost Technology works, it's opportunistic in that a lot of factors can influence max clocks, workload, current, voltage, cooling, TDP limits, etc... Usually the limiting factor is either TDP or cooling.

Or the quality of silicon. Yah know, just like Intel state...

 

[AndreiD]

Or the quality of silicon. Yah know, just like Intel state...

Nowhere does Intel state that and I highly doubt you can provide an official source to back up your false statement.

The Finish site shouldn't be reviewing ES chips at all. Very unprofessional.

Those are the chips Intel provided to reviewers, most of them got Engineering Samples. Track back to various outlet 8700K reviews and look at the IHS.

 

[jigger]

Nowhere does Intel state that and I highly doubt you can provide an official source to back up your false statement.

I also doubt it would make a difference. Maybe all this is just a misunderstanding. Go back through the other 10 or so threads we had on this topic.

 

[AmateurExpert]

Again though, this is a review/buyer research issue, not anything Intel is lying about.

What problem am i supposed to accept? There is no problem as far as i can see. The cpu's work as described.

Let's start again. What is that you take issue with, in terms of how the coffee lake range is described by Intel?

Intel set out only the base and single-core boost speeds, leaving a grey area in multi-core operation. Base speed and TDP are relatively low leaving a large margin of variability in performance above.

Reviews and benchmarks are done which give absolute and relative results for all-core operation, with little/no clarity on the speed and TDP configuration the CPUs are operated at, but many are in fact in systems representative of enthusiast builds, but not retail.

With no clarity of the all-core speeds achieved in comparative benchs or achievable in retail PCs, there's likely to be a misconception among non-enthusiast consumers about the true performance that Intel CPUs will offer, as many reviews and videos do not make this clear. The only way they will discover this is if they find a product-specific review of the retail PC - of which there is in fact one for the PC in the OP - but sadly this is silent on the ~20% performance that they're missing out on because of the TDP limitations.

There has of course always been compromises when buying mass market vs premium builds, but never before has the margin been so wide and the specifications been so obscure. This is what some take issue with.

 

[Jono8]

Intel set out only the base and single-core boost speeds, leaving a grey area in multi-core operation. Base speed and TDP are relatively low leaving a large margin of variability in performance above.

Reviews and benchmarks are done which give absolute and relative results for all-core operation, with little/no clarity on the speed and TDP configuration the CPUs are operated at, but many are in fact in systems representative of enthusiast builds, but not retail.

With no clarity of the all-core speeds achieved in comparative benchs or achievable in retail PCs, there's likely to be a misconception among non-enthusiast consumers about the true performance that Intel CPUs will offer, as many reviews and videos do not make this clear. The only way they will discover this is if they find a product-specific review of the retail PC - of which there is in fact one for the PC in the OP - but sadly this is silent on the ~20% performance that they're missing out on because of the TDP limitations.

There has of course always been compromises when buying mass market vs premium builds, but never before has the margin been so wide and the specifications been so obscure. This is what some take issue with.

So the real issue is the availability of reviews then, or even perhaps the transparency with specs/settings used by the oem partners (such as Medion)?

As i said though, there is no con from Intel's part and this is nothing exclusive to coffee-lake (ie how boost clocks work). Intel make it very clear how their cpu's operate on their website if anyone cared to learn or read about it.

 

[AmateurExpert]

So that is good. No one can explain what the actual "con" is.

It is indeed debatable whether the term "con" is applicable. But that's missing the point - whether the grey area created by the move to keep actual operating speeds secret is going to harm the typical consumer.

I don't see any evidence that Intel are seeding specially binned chips to reviewers though, even though this type of shenanigans has happened before. Intel's scale and quality of production renders this unnecessary.

 

[AndreiD]

@AmateurExpert There's actually no grey area there, the issue is just that most people have no idea how Intel's Turbo works. If given the same workloads and environment, 2 or more of the same chip will perform the same (reaching the same Turbo clocks). The reason why they don't specify the maximum all core turbo, which mind you wasn't specified in ark or their marketing previously, is because Turbo is opportunistic and there are a lot of limits and factors involved into what turbo clocks you'll in any given scenario. If you have adequate cooling and there are no issues with your motherboard/PSU, when you play a game you'll most likely get an all core turbo close to the maximum turbo since that's a fairly light workload, but if you run something proper multithreaded you will run into the current or TDP limits and your clocks will be lower.

 

[Jono8]

@AmateurExpert There's actually no grey area there, the issue is just that most people have no idea how Intel's Turbo works. If given the same workloads and environment, 2 or more of the same chip will perform the same (reaching the same Turbo clocks). The reason why they don't specify the maximum all core turbo, which mind you wasn't specified in ark or their marketing previously, is because Turbo is opportunistic and there are a lot of limits and factors involved into what turbo clocks you'll in any given scenario. If you have adequate cooling and there are no issues with your motherboard/PSU, when you play a game you'll most likely get an all core turbo close to the maximum turbo since that's a fairly light workload, but if you run something proper multithreaded you will run into the current or TDP limits and your clocks will be lower.

Exactly.

For example, with my 8700, if i put Windows in high performance mode with nothing open/running(ie idling) all cores go to 4.6hz. When you start using it, this fluctuates between 4.3-4.6 (depening on how many threads are being used obviously), and when starting something like prime or realbench, the cores then all drop down to and stay at 4.3ghz. If in balanced mode, the clocks will vary from 800 mhz all the way up to 4.6mhz dependent on the load but again, start something that stresses all cores at 100% and they will all sit at 4.3ghz.

 

[jigger]

It is indeed debatable whether the term "con" is applicable. But that's missing the point - whether the grey area created by the move to keep actual operating speeds secret is going to harm the typical consumer.

I don't see any evidence that Intel are seeding specially binned chips to reviewers though, even though this type of shenanigans has happened before. Intel's scale and quality of production renders this unnecessary.

Coffelake production is still poor. The best chip go to make Xeons and mobile products. Very few sites got chips to test on release and believe some had to hand them back after testing.

 

[AmateurExpert]

So the real issue is the availability of reviews then, or even perhaps the transparency with specs/settings used by the oem partners (such as Medion)?

We can certainly agree that transparency is the real issue.

There's actually no grey area there.

The lack of multi-core boost specs is the grey area I'm referring to. It's grey because Intel now claims this is proprietary information that they choose to no longer disclose. The suspicion is that this'll harm consumers in some way, and there seems to be some validity in it, as the variation in MT CB scores imply. At the moment this type of workload is rare for consumers, so it's fortunate that they see little harm in practice. As a development though it's not necessary, it's not beyond AMD to do similar (or worse) and I don't welcome it.

 

[humbug]

Even Jayz2cents has said retail chips are not clocking as high as the sample he got from Intel, his view is all, nVidia, Intel and AMD give reviewers the very best binned products for reviewing.

Of course they do, but this isn't about that, this is about advertising the CPU as 3.7Ghz 95 Watts TDP when infact they could boost on all cores as high as 4.7Ghz if you run them on very high end Motherboards and coolers, that is a massive difference, its the difference you get from reviewers scoring 1200 points in Cinebench and 1500 points, what those scoring 1500 points are saying is anyone can expect that out of the box and its only running at 3.7Ghz (so lots and lots of performance headroom from overclocking)
That is incredibly misleading, the fact is a retail CPU will not score much more than 1500 point overclocked as far as it will go, even when binned if they run at 5Ghz that is still only a 5% gain.

 

[Jono8]

if you run them on very high end Motherboards and coolers,% gain.

The 8400 build i did was running on the cheapest coffeelake mobo you could buy (the asrock a pro) and the stock cooler and boosted to 3.8ghz on all cores at 100% load and 3.9-4.0ghz on a mixed core load. Hyperbole doesnt help anyone.

 

[humbug]

The 8400 build i did was running on the cheapest coffeelake mobo you could buy (the asrock a pro) and the stock cooler and boosted to 3.8ghz on all cores at 100% load and 3.9-4.0ghz on a mixed core load.

You still haven't watched the bloody video, if you had you would know Adored even said exactly that himself about the 8400, he isn't talking about the 8400 in the sense that its misleading, he is talking about the 8700/K and its misleading advertised clock rates and TDP, whether or not that's the fault of reviewers or Intel Adored is shining a light on the fact that depending on which reviewer you go to and what hardware they are using dictates how much over 'or in some cases not' the chip is performing above Intel's ratings.

The fact is different reviewers are getting vastly different performance refor the same chip, for example between 1200 points in Cinebench to 1500 points for the same chips, Adored is simply shedding light on that and explaining why that's happening.

 

[CAT-THE-FIFTH]

The 8400 build i did was running on the cheapest coffeelake mobo you could buy (the asrock a pro) and the stock cooler and boosted to 3.8ghz on all cores at 100% load and 3.9-4.0ghz on a mixed core load. Hyperbole doesnt help anyone.

As a result, the clock of the processor in the Medion Erazer X67015 drops from 4.3 GHz to 3.6 to 3.7 GHz after about 15 seconds, while classic PC mainboards allowed the CPU to permanently consume more power and the maximum specified by Intel maximum turbo for 6-core loads of 4.3 GHz is also held permanently (in Prime95, there are even only 3.1 instead of 4.3 GHz). But what is correct?

Remember the Core i7 has HT and more L3 cache which should consume more power,and lead to a bit more heat being produced than the Core i5 CPUs. This says to me,that at a 65W TDP,the 6C Intel CPUs are probably going have all cores Turbo under sustained load of upto 3.8GHZ,and any higher is for very short term all cores usage,especially if you are using the stock cooler.

This sounds a bit like the Xeon E3 CPUs I have had which are TDP capped,so under sustained all cores load will clock a bit lower than for a shorter all cores workload.

With better motherboards(like most of the consumer Z series ones),it should be easier to breach TDP especially with better cooling,but OEM PCs are not known to have the best motherboards or cooling.