i7-7700K first benchmarks: 40% up in single threaded, 20% up in multithreaded (UPDATED)

[RejZoR]

I have it set to sync all cores, but mine doesn't boost higher than 3337MHz for some reason, it should be just over 3.6 (since it's a bit OC'd due to XMP memory profile).

If you have Sync All Cores and all set to Auto it'll behave the same as in Auto mode. You need to manually input the desired multiplier. 36 for example. Then, all cores will go up to 3600 MHz. If it'll hit thermals or other limits it'll drop to lower clocks.

 

[silentbogo]

There's probably something I'm missing in these results:
https://browser.primatelabs.com/v4/cpu/compare/194754?baseline=583064

 

[EarthDog]

Its quite obvious that in this benchmark when its tested at the same clockspeed, you won't find the IPC anywhere close to 40% over Skylake. I'm sure members here will get out their calculators and say, well if Skylake turbo was 4.2Ghz and KL is 4.5GHz, that 's a 300 Mhz difference and 7.x% IPC.... yada yada. But we need to see it heads up.

 

[Ferrum Master]

There's probably something I'm missing in these results:
https://browser.primatelabs.com/v4/cpu/compare/194754?baseline=583064

Nothing. It seems I will still use my SB-E.

 

[biffzinker]

Skylake turbo was 4.2Ghz

Wait a minute, I thought Skylake Turbos to 4.4 GHz, and the 4 core load clock speed is 4.2 GHz the same as Haswell?

 

[EarthDog]

Wait a minute, I thought Skylake Turbos to 4.4 GHz, and the 4 core load clock speed is 4.2 GHz the same as Haswell?

Skylake (6700K) is not the same as Haswell (4790K) - http://ark.intel.com/products/88195/Intel-Core-i7-6700K-Processor-8M-Cache-up-to-4_20-GHz

 

[Filip Georgievski]

Reading this thread makes me giggle a lot.
Everyone states that Kaby and Sky are same arch, yet you guys here fight over performance on both.
How dumb can you get??
It is like watching a race between 2 Ferrari cars, same model, same power, but 1 has traction control off and can get faster to finish...
The only thing here mentioned about Kaby that is true is the iGPU.
Nothing else has changed.
This reminds me of the episode about 390X being a rebranded 290X.

 

[LAN_deRf_HA]

I'm not really expecting much at all in regards to IPC but I am hopeful high boost equals high overclock. I think we've waited long enough for a chip that can do 5ghz 24/7.

 

[BirdyNV]

Either fake or CPU was running at very high frequency. There's no way they could have achieved 40% performance improvement on the same tech process and same architecture. I could bet that actual performance increase will be well bellow 5%. And here I'm being very optimistic...

Exactly, I can only imagine the backlash if they come with a 40% increase..."Where was all this performance before?!?!?"

 

[qubit]

Exactly, I can only imagine the backlash if they come with a 40% increase..."Where was all this performance before?!?!?"

Yeah, I was kinda thinking that. It's not unheard of though, since I remember NVIDIA did this a few years back with drivers to better compete against AMD when they released new cards.

I guess their strategy was to give AMD a lower performance target to aim for so that NVIDIA could crush them. Dirty tricks where the customer loses out, but unfortunately true.

 

[Bill_Bright]

Just because a site mentions why their article may be/is off, doesn't show they have integrity...

I said it's "a sign of integrity" because it, in fact, calls into question those benchmark results - which is what any journalistic article should do - question the facts instead of blindly repeating them (sending them viral) and partake in rumormongering.

 

[biffzinker]

It's Geekbench so consider it another reason not to take the scores seriously. Does anyone remember the fanboy spats on the internet proclaiming the iPad Pro 12" caught up to Intel's Core i3/i5 going off Geekbench scores?

 

[BirdyNV]

Yeah, I was kinda thinking that. It's not unheard of though, since I remember NVIDIA did this a few years back with drivers to better compete against AMD when they released new cards.

I guess their strategy was to give AMD a lower performance target to aim for so that NVIDIA could crush them. Dirty tricks where the customer loses out, but unfortunately true.

Honestly, I'm content with my PC so..this stuff just kinda goes over my head most of the time, I only get mad when we are being cheated as consumers.

 

[Frag_Maniac]

What I recall of Kabylake is first being hopeful about it, since Skylake wasn't really a gaming performance gain over Haswell. When I'd read architecture wise it was only improved via iGPU though, I lost interest, and have yet to even bother looking for credible benches on it.

It doesn't even make sense to me that they'd waste time and money improving the graphics, when it's a chip that is clearly more in the high end gaming category, vs HTPC. It's about as crazy as Intel being obsessive about ray tracing on everything.

No serious gamer uses built-in graphics. So honestly, I'm actually more interested in what Zen can do. I'd rather not pay for improved, but still far too weak graphics architecture that I don't need and won't use.

As for GPU driver boosts, vs same architecture CPU generation boosts, seriously? That's clearly apples and oranges. It's not like CPUs live or die off driver writing like GPUs do.

With all chips, it's always a case of whether they are proven worthy by consumers, and Kabylake is far from that stage yet.

 

[sneekypeet]

Thread has been thoroughly cleansed and multiple infractions have been passed out to various members. It is only due to my sick nature that I open this thread back up to see if anyone else would like to join the crowd. Moderation stick swung, scored a home run. Anyone else posting off topic or wants to derail TPU threads can join the list. This Is The One and Only Public Warning!

 

[silentbogo]

Anyway, WCFTech has just posted an update.
As first assumed - they've used a reference bench of i7-6700K from Geekbench 3.2.2, while i7-7700K was tested on Geekbench 4.
More excuses follow, though foul play not admitted.

I apologise for the err on my part in not conducting the necessary due diligence.

 

[EarthDog]

I said it's "a sign of integrity" because it, in fact, calls into question those benchmark results - which is what any journalistic article should do - question the facts instead of blindly repeating them (sending them viral) and partake in rumormongering.

Regardless, I expect that of sites. Wcftech is still considered a rumor mill. Just like other rumor mills, they can be hit or miss. They are not high on my list either. But I do read it and met a couple of guys there/were there. I think their update above speaks volumes for both the rumor mill (that it is) and its integrity(in a positive light) all in one fell swoop.

Anyway, WCFTech has just posted an update.
As first assumed - they've used a reference bench of i7-6700K from Geekbench 3.2.2, while i7-7700K was tested on Geekbench 4.
More excuses follow, though foul play not admitted.

It makes more sense now, the different versions among other things in why you saw such an increase. I know from benching Geekbench from its infancy to now(ish), it can be an inconsistent benchmark. Run it enough times, it 'pops' a higher score once in a while.

Also, (my editing has to be driving people nuts who are 'active' in the thread LOL) it was said that KL was more of an update, not an upgrade. But hey, with the market/competition how it is, any improvement is better than none.

EDIT: And now this post was edited by a mod????

 

[sneekypeet]

EDIT: And now this post was edited by a mod????

Consider it a pleasantry, I could have just infracted you for being off topic. Better yet, since the solution to all the rage has been posted, I may as well just close up shop since it seems everyone is unable to follow the rules of this forum.