Early Core i7-7700K Testing Reveals Higher Clocks And Power Usage

[HardOCP News]

According to the crew over at the Tech Report, a website owned by an AMD employee, supposed leaked Intel Core i7-7700K benchmarks shows higher clock speeds, performance and power usage from Intel's upcoming Kaby Lake processor.

The basic specs of the Core i7-7700K that Tom's reports are about in line with the rumors and leaks we've seen so far. Tom's says the chip has a 4.2 GHz base clock and a 4.5 GHz Turbo speed, and its four cores and eight threads slip into a 95W TDP—up slightly from the Core i7-6700K's 91W figure. That TDP increase is borne out by a small increase in power consumption in the site's testing—141W under a stock-clocked Prime95 Small FFTs load for the Kaby Lake chip, up from 133W for the Core i7-6700K running the same torture test.

 

[Shintai]

The idle numbers shows something is not entirely right. There is also the question if MCE was on.

Even OC seems to be a problem for Toms to get right.

 

[AceGoober]

I'm still rocking a 4790k @ stock 4Ghz and turbo of 4.4Ghz. Looking mighty tempting to jump up to a 7700k/Z170/DDR4 setup.

 

[cyclone3d]

I'm still rocking a 4790k @ stock 4Ghz and turbo of 4.4Ghz. Looking mighty tempting to jump up to a 7700k/Z170/DDR4 setup.

Why not just overclock y our current chip to start with?

 

[Shintai]

I'm still rocking a 4790k @ stock 4Ghz and turbo of 4.4Ghz. Looking mighty tempting to jump up to a 7700k/Z170/DDR4 setup.

There will be Z270 boards too.

 

[necrosis]

There will be Z270 boards too.

This is what I am waiting for. No massive need to upgrade now so I might as well wait for the Kaby's and Z270 boards to hit. Slap a Pro 960 into it with my Titan X to make a tasty small powerful rig.

 

[Meeho]

The idle numbers shows something is not entirely right. There is also the question if MCE was on.

Even OC seems to be a problem for Toms to get right.

What am I missing here?

 

[Deleted member 93354]

Hard to say what's going on. It might be an engineering sample. It might be a bum sample. The Z170 board might need a BIOS update. So it's still too early to tell.

 

[Loose Nut]

I stay away from that site, but thanks for the reminder why.
website owned by an AMD employee? that's funny actually

 

[iQuasarLV]

Steve:

Why the 3rd degree re-post. Why not follow the link to the source and post that?

(1)You-(2)Them-(3)Them's reference.

 

[Dan_D]

Hard to say what's going on. It might be an engineering sample. It might be a bum sample. The Z170 board might need a BIOS update. So it's still too early to tell.

The CPU is almost certainly not a retail sample. That said, its probably pretty damn close to final silicon. Combining the CPU with a Z170X based motherboard might have a major impact on the numbers. One thing you guys need to keep in mind is that the BIOS support has to be there for the old and new CPUs or things may not behave correctly. Ordinarily, you wouldn't think the board would have that much of an impact on things but we saw big differences with some more recent motherboards which centered around the F-clock values and something else in the BIOS / UEFI which allows ASUS to trounce everyone else in the benchmarks for quite some time. Eventually that gap closed but there was something going on there which impacted performance in CPU limited tests to a significant degree.

Also, the 890FX and 990FX chipsets for AM3 / AM3+ were virtually identical. The latter boards supported a new C-state the other motherboards couldn't. The VRM/VRD design was somewhat different on the 990FX motherboards. Those motherboards were also a disaster with the older chips as they were developed and QC'ed with Bulldozer. It was so bad that we quit reviewing those for quite awhile until things got sorted out. Of course then Bulldozer came out and was lackluster so no one cared.

The point is that these numbers should be taken with a grain of salt.

 

[Shintai]

What am I missing here?

For the OC part? 20% faster with 4.3% clock advantage?
Or you thinking about power consumption with the wide spread idle usage?

 

[Meeho]

For the OC part? 20% faster with 4.3% clock advantage?
Or you thinking about power consumption with the wide spread idle usage?

For idle usage I'm guessing it is due to increased voltage. Performance varies from 2.5 to 11% faster if I'm reading the chart right. There doesn't seem to be anything too suspicious there so I wasn't sure what you meant by "even OC seems to be a problem for Toms to get right".

 

[martinmsj]

I stay away from that site, but thanks for the reminder why.
website owned by an AMD employee? that's funny actually

It's Toms Hardware who has the sample and benchmarked it. The article from TechReport just talks about what Tom's Hardware found.

 

[nutzo]

It might finally be time to upgrade my old i7 860

Even without overclocking the new chip, I would see a 70-80% increase over my current system.

However, I think I'll put more importance on making it as quiet as possible over slightly higher performance.
A couple cheap 1TB SSD's would let me dump some of my old noisy spinners.

 

[Track Drew]

The idle numbers shows something is not entirely right.

What doesn't seem right?

Could improved power gating or faster clock speed changes on the new part account for the lower idle?

 

[THUMPer]

Power is off the charts! Intel is going bankrupt!

 

[Shintai]

What doesn't seem right?

Could improved power gating or faster clock speed changes on the new part account for the lower idle?

Anyone with an Intel CPU, specially a later one should know it draws extremely little idle. Kaby Lake isn't going to draw 13W less AC side than Skylake. Because then Kaby Lake would be producing power.

 

[IdiotInCharge]

I stay away from that site, but thanks for the reminder why.
website owned by an AMD employee? that's funny actually

You stay away from the site that pioneered frame-time analysis?

Okay.

In other news, Kaby Lake looks like a mediocre bump; at best, you'd jump for the new platform support, you know, if you can actually make use of it. Still hard to justify if your current system is running good though, the speed gains of yore just aren't there and neither is the need for them.

 

[nilepez]

According to the crew over at the Tech Report, a website owned by an AMD employee, supposed leaked Intel Core i7-7700K benchmarks shows higher clock speeds, performance and power usage from Intel's upcoming Kaby Lake processor.

The basic specs of the Core i7-7700K that Tom's reports are about in line with the rumors and leaks we've seen so far. Tom's says the chip has a 4.2 GHz base clock and a 4.5 GHz Turbo speed, and its four cores and eight threads slip into a 95W TDP—up slightly from the Core i7-6700K's 91W figure. That TDP increase is borne out by a small increase in power consumption in the site's testing—141W under a stock-clocked Prime95 Small FFTs load for the Kaby Lake chip, up from 133W for the Core i7-6700K running the same torture test.

Why does it matter who owns Tech Report? They're just re-reporting what Tom's Hardware found (never really understood why [H] doesn't just link to the original website...sometimes your links are 3 or 4 links removed for the original publisher)

 

[Track Drew]

Anyone with an Intel CPU, specially a later one should know it draws extremely little idle. Kaby Lake isn't going to draw 13W less AC side than Skylake. Because then Kaby Lake would be producing power.

As I have a i7-4790 sitting in the system on the desk next to me, I fall in the bucket of "anyone with an Intel CPU". I'm going to go from memory - but my kill-a-watt would show idle numbers that make the i7-6700K idle draw believable to me.

You could give some ballpark numbers of what you believe they should be and create much less confusion.

Article also says:

Idle energy also dropped to 24W after several hours, but an idle period of that length exceeds normal test procedure.

So that number wasn't typical - it was higher up to that point. The previous graph shows that.

 

[STrAYeR]

The (7700) non k would be a good deal if you still holding onto older cpu's. The k version is 40 dollars more but that will shoot up $20-30 more once the stores get it.

 

[Nebulous]

I'm still rocking a 4790k @ stock 4Ghz and turbo of 4.4Ghz. Looking mighty tempting to jump up to a 7700k/Z170/DDR4 setup.

Me too except I'm @ 4.7Ghz. I'd like to upgrade when the need arises over the wants

 

[Loose Nut]

It's Toms Hardware who has the sample and benchmarked it. The article from TechReport just talks about what Tom's Hardware found.

You stay away from the site that pioneered frame-time analysis?

Okay.

In other news, Kaby Lake looks like a mediocre bump; at best, you'd jump for the new platform support, you know, if you can actually make use of it. Still hard to justify if your current system is running good though, the speed gains of yore just aren't there and neither is the need for them.

"shakes head" have to explain everything the these kids nowadays, Its Tom's Hardware website that I stay away from. I mean that's who and what the article is about.

 

[Shintai]

http://www.tomshardware.com/news/intel-core-i7-7700k-kaby-lake-overclocking-update,33119.html

Intel Core i7-7700K Retest: More Oveclocking, Less Heat

MSI board this time.

 

[Decko87]

Probably still going to rock my 2500k until Zen releases, hoping it's good enough to drive down prices a bit.

 

[STrAYeR]

I'm hoping AMD can do something with Zen. The memories of Bulldozer and the hype/lies are still strong. I will take any numbers AMD releases with a grain of salt till benches prove otherwise. I really hope the SR7+ is not really 500 dollars. That will hurt them if it is a flop.

 

[Decko87]

I'm hoping AMD can do something with Zen. The memories of Bulldozer and the hype/lies are still strong. I will take any numbers AMD releases with a grain of salt till benches prove otherwise. I really hope the SR7+ is not really 500 dollars. That will hurt them if it is a flop.

I'll take any competition, anything that knocks their prices down a bit. I don't care who it comes from, or how it happens, but it needs to happen.

 

[Shintai]

I'll take any competition, anything that knocks their prices down a bit. I don't care who it comes from, or how it happens, but it needs to happen.

CPU prices as such has never been lower since 2006. And people should be careful when starting to match over in the server lines. AMD isn't going to sell anything for cheap from there. The prices will be no different from today, its just a matter of who is charging it.

 

[DedEmbryonicCe11]

I'm curious if when they refer to 1.3v is that the auto setting that still spikes quite a bit higher when AVX is used or a fixed voltage? I didn't see any mention of that skimming through both articles.

 

[pxc]

Why does it matter who owns Tech Report? They're just re-reporting what Tom's Hardware found (never really understood why [H] doesn't just link to the original website...sometimes your links are 3 or 4 links removed for the original publisher)

I think it's actually interesting HOW it's being reported. THG had several reservations about what its testing revealed, and stressed it might have been an issue with the motherboard. Yesterday THG released an update with different results using a different motherboard: http://www.tomshardware.com/news/intel-core-i7-7700k-kaby-lake-overclocking-update,33119.html

tl;dr Analysis froma site run by an AMD marketing employee probably isn't going to go out of its way to make sure the information is complete or up to date on a competitors product. IMHO, that makes such analysis worse than worthless since misleading information is worse than mere ignorance.

 

[cageymaru]

Translator needed. Seems to have big gains!
http://www.expreview.com/51056-4.html

 

[cageymaru]

 

[smaudioz]

I guess I might as welll just go ahead with a 6700K build then. I'm just waiting for the [H] review next week now really and any others that come in between now and then to see if it's worth waiting for.

 

[cageymaru]

I guess I might as welll just go ahead with a 6700K build then. I'm just waiting for the [H] review next week now really and any others that come in between now and then to see if it's worth waiting for.

If you read the translation of the website that I linked, they were able to get it to 5.0GHz on air. That's really good!

http://img.expreview.com/review/2016/12/Corei7_7700k/5g_1hour.png

 

[smaudioz]

And the cores hit 91C. I don't think that really counts, I could probably clock my 2500K to 5GHz on air too if I didn't care about running it at ~90C.

 

[Dayaks]

And the cores hit 91C. I don't think that really counts, I could probably clock my 2500K to 5GHz on air too if I didn't care about running it at ~90C.

It being stable at 5Ghz and 91C is impressive since high temps makes holding frequency even harder. Makes you wonder what it can do with a decent AIO.

 

[DedEmbryonicCe11]

It being stable at 5Ghz and 91C is impressive since high temps makes holding frequency even harder. Makes you wonder what it can do with a decent AIO.

At those temperatures won't it degrade though? So yeah, you would need at least a 2x120mm AIO to get it back down to something that could be used long term.

 

[Araxie]

And the cores hit 91C. I don't think that really counts, I could probably clock my 2500K to 5GHz on air too if I didn't care about running it at ~90C.

you can't expect same Cooling potential on planar architecture versus FinFet (aka 3D Transistor[Tri-gate]) which by itself increase the transistor density together with the reduced process of 14nm.. Kaby Lake by itself is no more than a improved skylake it's just a refinement of the same process so nothing great to be expected.

It being stable at 5Ghz and 91C is impressive since high temps makes holding frequency even harder. Makes you wonder what it can do with a decent AIO.

make me wonder what can reach with a direct die cooling, or at least delidded with Liquid metals.. the main limiting factor since ivy bridge is always the damn TIM used.

 

[Araxie]

At those temperatures won't it degrade though? So yeah, you would need at least a 2x120mm AIO to get it back down to something that could be used long term.

Not really, is not the temperature what cause noticeable degradation overtime, is the voltage. does the chip run at 90C? yeah that's under unreasonable stress FPU test which tend to be at least 15C higher than any other real world application being gaming or professional usage or even folding... hitting 90C with Aida64 FPU only test?. man, that's marvelous.. that's a thing that couldn't happen with any other i7 since Sandy Bridge, 3770K/4770k/4790K/6700K (not count broadwell as not a common factor). all would reach instantly 100C+ under a FPU only stress test overclocked.. that just as Dayaks said make one wonder what kind of things can reach on better cooling (preferably delided)..