Tom's Hardware i7-7700K OC Preview

[nathanddrews]

http://www.tomshardware.com/reviews...00k-overclocking-performance-review,4836.html

Meh.

 

[Sven_eng]

At this rate of progress the i7-8700K should be as hot and slow as Bulldozer.

 

[Arachnotronic]

At this rate of progress the i7-8700K should be as hot and slow as Bulldozer.

7700K didn't get slower than the 6700K.

 

[Lepton87]

A new process node(no to confuse with a shrink, it's like LPP and LPE for 16nm on TSMC) and a new name along with a chipset change and all we get is an improvement that's lesser then what we used to get from a stepping change. The most recent examples being the core 2 quad and the first generation of core i7. Core 2 Quad had a C1 to G0 stepping change which was a very big improvement and i7 saw the change from C0 to D0 which easily improved overclocking by upwards of 300MHz. As far as I rember C0 typically reached 3.6GHz while for D0 4GHz was commmon.
Another way to look at this is that the first stepping of a shipping CPU is way more mature than it used to be and that's why a simple stepping change won't give any meangful improvements to power efficiency or frequency headroom

 

[PPB]

There is no new node. It was just PR talk.

Intel with 14nm already had 2 variations of 14nm: one density optimized, one fmax/performance optimized. 6700K was on the latter and little could Intel do to improve things for 7700K, which at this point is only a better binned CPU paired with a more robust uncore (peeps gotta get their 4K netflix, you know).

7 years ago this wouldn't pass as a new generation, just a new stepping a la D0 920's. The milking has to stop.

 

[Arachnotronic]

There is no new node. It was just PR talk.

Intel with 14nm already had 2 variations of 14nm: one density optimized, one fmax/performance optimized. 6700K was on the latter and little could Intel do to improve things for 7700K, which at this point is only a better binned CPU paired with a more robust uncore (peeps gotta get their 4K netflix, you know).

7 years ago this wouldn't pass as a new generation, just a new stepping a la D0 920's. The milking has to stop.

14nm+ uses a significantly changed transistor, among other things, relative to the processes that shipped in Broadwell/Skylake.

The LP vs HP refers to the types of transistors and metal stack options that Intel makes available to the SoC implementation teams. Completely orthogonal to the 14nm vs 14nm+ discussion.

 

[kaesden]

There is no new node. It was just PR talk.

Intel with 14nm already had 2 variations of 14nm: one density optimized, one fmax/performance optimized. 6700K was on the latter and little could Intel do to improve things for 7700K, which at this point is only a better binned CPU paired with a more robust uncore (peeps gotta get their 4K netflix, you know).

7 years ago this wouldn't pass as a new generation, just a new stepping a la D0 920's. The milking has to stop.

the milking won't stop until AMD provides something actually competitive. Intel has no motivation to innovate currently.

 

[Arachnotronic]

the milking won't stop until AMD provides something actually competitive. Intel has no motivation to innovate currently.

Intel wants to sell CPUs to people at a faster rate than they currently do. They also don't want to be blindsided by competitive threats (AMD announced Zen two years ago, complete with their IPC and overall performance goals). What more motivation could they have?

 

[Dave2150]

The 'new' 14nm+ process appears to be extremely disappointing. Power consumption is through the roof.

I suspect this 14nm+ process 'improvement' is a marketing gimmick more than anything else, simply due to them not being able to continue with the tick tock model Any changes from 14nm to 14nm+ would probably not have been mentioned in previous years (such as 22nm improvements, from 4770k to 4790k etc).

Hopefully Zen proves to be interesting, otherwise it's a long time before Cannon/Coffee lake.

 

[PPB]

The 'new' 14nm+ process appears to be extremely disappointing. Power consumption is through the roof.

I suspect this 14nm+ process 'improvement' is a marketing gimmick more than anything else, simply due to them not being able to continue with the tick tock model Any changes from 14nm to 14nm+ would probably not have been mentioned in previous years (such as 22nm improvements, from 4770k to 4790k etc).

Hopefully Zen proves to be interesting, otherwise it's a long time before Cannon/Coffee lake.

This guys gets it. And this only shows that there is no reliable PR. PR statements need to be taken with salt whichever the company it comes from.

Just as a mental exercise: Imagine what would have happened if AMD released Vishera-K as a new generation of FX CPUs? Looking back, that was the last example of a good stepping change giving meaningfull perf/watt and transistor performance uplifts.

 

[Abwx]

The 'new' 14nm+ process appears to be extremely disappointing. Power consumption is through the roof.

I suspect this 14nm+ process 'improvement' is a marketing gimmick more than anything else, simply due to them not being able to continue with the tick tock model Any changes from 14nm to 14nm+ would probably not have been mentioned in previous years (such as 22nm improvements, from 4770k to 4790k etc).

Hopefully Zen proves to be interesting, otherwise it's a long time before Cannon/Coffee lake.

They said 12% better perf, but for obvious reasons they wont disclose what are the transistors parameters that were improved.

At worst the transistor characteristics are unchanged in term of global perf/Watt but with a better frequency ceiling, mathematicaly this amount to extend the voltage range within wich the device keep following a square law, there s of course an improvement in perf/Watt at the extremity of the curve where the previous process is less efficient because it s already above a square law.

Other possibility is that transconductance has been improved accordingly, there will be an improvement within the full voltage range of the device in respect of previous process, and this would manifest by eventually an extended frequency range, but in all cases by a 25% perf/Watt improvement, hence the frequency/voltage curve is shifted in the graph by the improvement factor wich is 1.12.

Assuming the 6700K is working at 4GHz then the measured 8-9W between the two CPU amount to 4.5% higher frequency at square law rate, so that s apparently the first case, that is, frequency ceilling is improved but not perf/Watt since the frequency is not high enough, but at close to max frequency the perf/watt improvement is visible since the difference is the same despite the devices being 200MHz apart, at this point the 6700K efficency is more degraded at 4.6 than for the 7700K at 4.8.

 

[LTC8K6]

the milking won't stop until AMD provides something actually competitive. Intel has no motivation to innovate currently.

Yes, that explains the Pentium with HT and the unlocked i3. Also the i3 with turbo. Intel doesn't want to sell you a new chip...

 

[Arachnotronic]

The 'new' 14nm+ process appears to be extremely disappointing. Power consumption is through the roof.

I suspect this 14nm+ process 'improvement' is a marketing gimmick more than anything else, simply due to them not being able to continue with the tick tock model Any changes from 14nm to 14nm+ would probably not have been mentioned in previous years (such as 22nm improvements, from 4770k to 4790k etc).

Hopefully Zen proves to be interesting, otherwise it's a long time before Cannon/Coffee lake.

We should reserve judgement about 14nm+ for desktops until the retail chips are out and tested in proper boards (14nm+ in mobile gives measurable performance/watt improvements), but from these preliminary results the power consumption/temperatures of 7700K are disappointing.

The higher overclocking capability at a given voltage shown in the tests is encouraging, though. Silicon Lottery says that only the top 19% of tested 6700K chips can hit 4.8GHz or better...at 1.424V (or less).

https://siliconlottery.com/products/6700k48g

 

[dark zero]

Yes, that explains the Pentium with HT and the unlocked i3. Also the i3 with turbo. Intel doesn't want to sell you a new chip...

In fact, we have the OC SKL old chips still around there. And Intel is still not increasing he core count... AMD might surprise Intel if they are goofing like that. Similar thing with Core M. Apple is ready to counter them with their 16nm improved chip.

 

[frozentundra123456]

Yea, I dont know what people expected. Given the high levels of overclocking on skylake already, 200 mhz more is an accomplishment. The temperatures are quite disappointing though. But I would say we have to wait for more data before making a final evaluation. And of course, stock clocks are higher both on desktop and mobile.

 

[moonbogg]

Yep, Skylake 8 core then forget about CPU's for a decade. This "new" CPU is laughable. The only thing new that it gives you is Netflix 4k streaming. What a racket. Zen is going to kick this thing's ass.

 

[cbn]

Is the memory controller the same on Kabylake? (ie, DDR4 2133)

Or it DDR4 2400 or better?

(I believe it is DDR4 2400, but I can't find the info)

 

[coercitiv]

We should reserve judgement about 14nm+ for desktops until the retail chips are out and tested in proper boards (14nm+ in mobile gives measurable performance/watt improvements), but from these preliminary results the power consumption/temperatures of 7700K are disappointing.

The higher overclocking capability at a given voltage shown in the tests is encouraging, though. Silicon Lottery says that only the top 19% of tested 6700K chips can hit 4.8GHz or better...at 1.424V (or less).

Et tu, Brute?

Is there no one in this thread asking themselves how come the default 4.5Ghz stock turbo voltage for 7700K on that GA-Z170X-Ultra is the same 1.3V they used to overclock the chip to 4.8Ghz? This is the frequency domain where every increase in frequency should require a voltage sacrifice, yet KBL is cruising at 1.3V.

Our motherboard set the Core i7-7700K to 1.30V at its 4.50 GHz max turbo, whereupon it behaved exactly the way we’d expect our Core i7-6700K to act when overclocked to 4.50 GHz at 1.30V.

This Core i7-7700K sample can be overclocked on this motherboard to 4.80 GHz at 1.30V, compared to the Core i7-6700K’s 4.60 GHz.

Even for the people who don't have the time to read the article, the Power consumption graph should also raise eyebrows: power usage at stock higher than Skylake, but overclocked power usage in line with Skylake? The evidence is so clear here every Intel fan who did not react should get an infraction. I'm not even going to mention the idle power usage difference, that would be a complete joke (-13W for idle power usage just from going from 6700K to 7700K)

As for the people who act completely disappointment by this clearly minor improvement, what exactly were you expecting? What did that PR machine promise anyway, except for higher overclocks?

This is how I read this preview, maybe it will help others understand it better as well:
my i5 6600K needs 1.44V for 4.8Ghz, 1.35V for 4.7Ghz
the i7 7700K needs 1.3V for 4.8Ghz, and will likely clock a bit higher at 1.35V and 1.4V respectively

I can only hope this launch will finally signal some people to ask what is the true relation between Intel stock voltage and motherboard stock voltage on desktop chips. Back when some AMD fans talked about it for Dozer family they were greeted with shouts about AMD incompetence, maybe this time people will be able to focus better.

 

[Arachnotronic]

Et tu, Brute?

Is there no one in this thread asking themselves how come the default 4.5Ghz stock turbo voltage for 7700K on that GA-Z170X-Ultra is the same 1.3V they used to overclock the chip to 4.8Ghz? This is the frequency domain where every increase in frequency should require a voltage sacrifice, yet KBL is cruising at 1.3V.


Even for the people who don't have the time to read the article, the Power consumption graph should also raise eyebrows: power usage at stock higher than Skylake, but overclocked power usage in line with Skylake? The evidence is so clear here every Intel fan who did not react should get an infraction. I'm not even going to mention the idle power usage difference, that would be a complete joke (-13W for idle power usage just from going from 6700K to 7700K)

As for the people who act completely disappointment by this clearly minor improvement, what exactly were you expecting? What did that PR machine promise anyway, except for higher overclocks?

This is how I read this preview, maybe it will help others understand it better as well:
my i5 6600K needs 1.44V for 4.8Ghz, 1.35V for 4.7Ghz
the i7 7700K needs 1.3V for 4.8Ghz, and will likely clock a bit higher at 1.35V and 1.4V respectively

I can only hope this launch will finally signal some people to ask what is the true relation between Intel stock voltage and motherboard stock voltage on desktop chips. Back when some AMD fans talked about it for Dozer family they were greeted with shouts about AMD incompetence, maybe this time people will be able to focus better.

What a really good post. I wish I could "double like" it, but I only have one "like" to give

 

[Arachnotronic]

This is how I read this preview, maybe it will help others understand it better as well:
my i5 6600K needs 1.44V for 4.8Ghz, 1.35V for 4.7Ghz
the i7 7700K needs 1.3V for 4.8Ghz, and will likely clock a bit higher at 1.35V and 1.4V respectively

I picked up a 6700K for one of my systems recently, and I have been spending some time overclocking it.

I simply can't get beyond 4.7GHz ASUS RealBench stable, even when I force-feed the thing 1.45V. It's not a temperature/core problem, since the core temperatures don't go higher than, say, 80-85C.

I had this same issue with the other 6700Ks that I had in the past...4.7GHz with lots of volts was the limit, and I could forget about 4.8GHz or beyond.

If 7700K can do 4.8GHz+ regularly, especially at 1.3v, that'd be really nice.

 

[Grooveriding]

Wait for reviews on independently bought retail samples. I always see pre-release and review samples of Intel CPUs overclocking better than they tend to average out at for end-users buying retail samples.

 

[arandomguy]

Let's give the 7700k a 10% clock speed and a generous 10% performance advantage from that. Say they OC to 5.1ghz vs 4.6ghz for the 6700k. This is trivial in real terms, how many usage scenarios are there in which that 10% difference is actually meaningful?

And in the real market come release time you are going to be paying more for it in an equivalent build. What will be more silly is those buying a 7600k builds likely at the same price they could've gotten a 6700k one just last week.

 

[VirtualLarry]

I had this same issue with the other 6700Ks that I had in the past...4.7GHz with lots of volts was the limit, and I could forget about 4.8GHz or beyond.

My i5-6400 wasn't full stable @ 4.51Ghz, with 1.410V, and I didn't feel like going higher. (Already hitting temp limits, and I upgraded the cooling once, now using a Zalman CNPS5x 92mm tower heatpipe cooler.)

 

[NTMBK]

Anyone else remember the slow, continuous improvement of 45nm over the life of the Phenom II? http://www.anandtech.com/show/2754 Process improvements aren't new, it's just slapping a + on the end that is different.

 

[LTC8K6]

Yep, Skylake 8 core then forget about CPU's for a decade. This "new" CPU is laughable. The only thing new that it gives you is Netflix 4k streaming. What a racket. Zen is going to kick this thing's ass.

I'm not sure why anyone was expecting KL to be revolutionary. It was never going to be anything but an optimization.

This was Anandtech's August piece on KL:

The fundamental microarchitecture between Skylake and the new Kaby Lake parts is practically unchanged (DMI 3.0 now allows PCIe 3.0 x4 NVMe drives from the integrated PCH), but the updated fin profile and reduced ‘strain’ by the larger fin pitch is being quoted as giving a 12% performance increase due to process alone, typically through additional frequency for the same power. The main benefits to KBL will be in that frequency due to the 14nm+ process as well as the new media capabilities.