Core i7 7700k benchmark

[NumLock21]

Not much info on it. Just a leaked benchmark on the upcoming Intel Core i7 7700K based on the Kaby Lake architecture

 

https://browser.primatelabs.com/v4/cpu/583064

[sazrocks]

4.2 GHz clock speed, cool

 

[TheKDub]

Interesting, considerably higher score than the i7 6700k

Single Core:

i5 6600k: 5309
i7 6700k: 5822
i7 7700k: 6131

 

Multi-Core:

i5 6600k: 14568
i7 6700k: 19437

i7 7700k: 20243

 

i5 6600k: https://browser.primatelabs.com/v4/cpu/620296

i7 6700k: https://browser.primatelabs.com/v4/cpu/621525

i7 7700k: https://browser.primatelabs.com/v4/cpu/583064 (Same as in the main post)

 

Hopefully that'll mean some real world performance increases lol

[Spork829]

10-15% improvements! Oh boy oh boy!

[Kosios]

Not much better

[sazrocks]

1 minute ago, TheKDub said:

Interesting, considerably higher score than the i7 6700k

Single Core:

i5 6600k: 5309
i7 6700k: 5822
i7 7700k: 6131

 

Multi-Core:

i5 6600k: 14568
i7 6700k: 19437

i7 7700k: 20243

 

i5 6600k: https://browser.primatelabs.com/v4/cpu/620296

i7 6700k: https://browser.primatelabs.com/v4/cpu/621525

i7 7700k: https://browser.primatelabs.com/v4/cpu/583064 (Same as in the main post)

That's what I thought. But now I wonder if there will be any performance increase GHz for GHz. Anyone have a 6700k 4.4(assuming the boost of the 7700k is 4.4) GHz geekbench?

[Enderman]

Seems about the same as the difference between the 4790k and 6700k.

So that is a lot more than "just a refresh" level of improvement...

https://browser.primatelabs.com/v4/cpu/40469

https://browser.primatelabs.com/v4/cpu/621525

https://browser.primatelabs.com/v4/cpu/583064

 

4790k, 6700k, 7700k

[Geldar]

9 minutes ago, TheKDub said:

Interesting, considerably higher score than the i7 6700k

Single Core:

i5 6600k: 5309
i7 6700k: 5822
i7 7700k: 6131

 

Multi-Core:

i5 6600k: 14568
i7 6700k: 19437

i7 7700k: 20243

 

i5 6600k: https://browser.primatelabs.com/v4/cpu/620296

i7 6700k: https://browser.primatelabs.com/v4/cpu/621525

i7 7700k: https://browser.primatelabs.com/v4/cpu/583064 (Same as in the main post)

 

Hopefully that'll mean some real world performance increases lol

4.14% isnt "considerably higher"... not even the usual 6-8%

[Imakuni]

6 minutes ago, Enderman said:

Seems about the same as the difference between the 4790k and 6700k.

So that is a lot more than "just a refresh" level of improvement...

https://browser.primatelabs.com/v4/cpu/40469

https://browser.primatelabs.com/v4/cpu/621525

https://browser.primatelabs.com/v4/cpu/583064

 

4790k, 6700k, 7700k

Most (if not all) of it probably comes from the clock bump, though. +200mhz on 8 threads is sure to push stuff up. And if we assume a single core can boost to 4.5 rather than 4.4, it would explain the single core difference pretty well.

[patrickjp93]

10 minutes ago, sazrocks said:

That's what I thought. But now I wonder if there will be any performance increase GHz for GHz. Anyone have a 6700k 4.4(assuming the boost of the 7700k is 4.4) GHz geekbench?

7700K is 4.5 boost, not 4.4.

[Electronics Wizardy]

19 minutes ago, sazrocks said:

That's what I thought. But now I wonder if there will be any performance increase GHz for GHz. Anyone have a 6700k 4.4(assuming the boost of the 7700k is 4.4) GHz geekbench?

Same IPC. Its the exact same cpu as the 6700k only the gpu and other parts have changed.

[nobelharvards]

http://www.anandtech.com/show/10610/intel-announces-7th-gen-kaby-lake-14nm-plus-six-notebook-skus-desktop-coming-in-january/4

 

Quote

For users that speak in pure IPC, this may/may not be a shock. Without further detail, Intel is implying that Kaby Lake will have the same IPC as Skylake, however it will operate with a better power efficiency (same frequency at lower power, or higher frequency at same power) and for media consumption there will be more idle CPU cycles with lower power drain. The latter makes sense for mobile devices such as tablets, 2-in-1s and notebooks, or for power conscious users, but paints a static picture for the future of the desktop platform in January if the user only gets another 200-400 MHz in base frequencies.

 

[MageTank]

56 minutes ago, Imakuni said:

Most (if not all) of it probably comes from the clock bump, though. +200mhz on 8 threads is sure to push stuff up. And if we assume a single core can boost to 4.5 rather than 4.4, it would explain the single core difference pretty well.

This is exactly it. There isn't a "10-15% improvement". I don't get where people keep pulling that number from. 

 

1 hour ago, TheKDub said:

Interesting, considerably higher score than the i7 6700k

Single Core:

i5 6600k: 5309
i7 6700k: 5822
i7 7700k: 6131

 

Multi-Core:

i5 6600k: 14568
i7 6700k: 19437

i7 7700k: 20243

 

i5 6600k: https://browser.primatelabs.com/v4/cpu/620296

i7 6700k: https://browser.primatelabs.com/v4/cpu/621525

i7 7700k: https://browser.primatelabs.com/v4/cpu/583064 (Same as in the main post)

 

Hopefully that'll mean some real world performance increases lol

From these scores, we see nearly a 5.5% increase in single core performance. The difference between the 6700k and 7700k's single core boost clock speed is 300mhz, a 7% difference. While perfect scaling seems impossible, it's safe to assume that the 7% increase in clock speed is completely responsible for this 5.5% increase in single core performance, not an IPC advantage.

 

Let's look at the multi-core to be certain. Difference in clock speed? 200mhz (4.0 6700k vs 4.2 7700k) which is a 5% difference in clock speed. The difference in the scores however, was a 4% difference in multi-core. Again, safe to assume this comes entirely from the 5% difference in clock speed. 

 

I am not quoting you to call you out @TheKDub, your post just has the relevant information for me to do this math. You may or may not agree with me, I was just quoting it so others could see where my math is coming from.

 

This is yet another confirmation that people should not worry about Kaby Lake if they already own a high end Skylake machine. Now, if you are upgrading from something older, obviously Kaby Lake is a better option depending on how it's priced vs Skylake, but it's just not a viable upgrade option if you already have Skylake. Until we see how this new fin design works, and the impact it has on overclocking, I wouldn't get my hopes up for the previously reported gains. 

[patrickjp93]

11 minutes ago, MageTank said:

This is exactly it. There isn't a "10-15% improvement". I don't get where people keep pulling that number from. 

 

From these scores, we see nearly a 5.5% increase in single core performance. The difference between the 6700k and 7700k's single core boost clock speed is 300mhz, a 7% difference. While perfect scaling seems impossible, it's safe to assume that the 7% increase in clock speed is completely responsible for this 5.5% increase in single core performance, not an IPC advantage.

 

Let's look at the multi-core to be certain. Difference in clock speed? 200mhz (4.0 6700k vs 4.2 7700k) which is a 5% difference in clock speed. The difference in the scores however, was a 4% difference in multi-core. Again, safe to assume this comes entirely from the 5% difference in clock speed. 

 

I am not quoting you to call you out @TheKDub, your post just has the relevant information for me to do this math. You may or may not agree with me, I was just quoting it so others could see where my math is coming from.

 

This is yet another confirmation that people should not worry about Kaby Lake if they already own a high end Skylake machine. Now, if you are upgrading from something older, obviously Kaby Lake is a better option depending on how it's priced vs Skylake, but it's just not a viable upgrade option if you already have Skylake. Until we see how this new fin design works, and the impact it has on overclocking, I wouldn't get my hopes up for the previously reported gains. 

The 6700K at 4 cores full tilt stays at 4.2GHz. For the 7700K, we don't know what multicore boost clocks are.

[MageTank]

Just now, patrickjp93 said:

The 6700K at 4 cores full tilt stays at 4.2GHz.

On Z board with "Multi Core Enhancement" (No idea what non-Asrock boards call it) enabled, they do. However, on my H170 board, my clocks fall back down to 4ghz under multi-core load. http://www.intel.com/content/www/us/en/support/processors/000005523.html

[Lays]

The benchmark doesn't seem to make any sense though.

 

I ran it at 5 Ghz and my single core is much higher, but multicore is lower.

 

I ran both Geekbench 3 and Geekbench 4, multicore is higher on geekbench 3, and geekbench 4 gets better single thread.

I think what they did was run it on both, and take the highest score from each and make an image with both, since they both display the scores in the same way.

 

Geekbench 3:

 

 

Geekbench 4:

[Lays]

Ran it again a few times with more things closed, I think the multi-core score on their run is still goofy somehow, my single thread is much higher so the multi-core should be higher too, I'd think.

@MageTank what do you think?

 

[stconquest]

So we can safely deduce (from these numbers) that IPC improvement was not focused in this architecture.

 

They probably added specific support/improvement for hardware/software integration and IDK what else.  Raw speed is not what they will be showcasing.   Maybe a little energy efficiency improvement, where they could.

[patrickjp93]

38 minutes ago, MageTank said:

On Z board with "Multi Core Enhancement" (No idea what non-Asrock boards call it) enabled, they do. However, on my H170 board, my clocks fall back down to 4ghz under multi-core load. http://www.intel.com/content/www/us/en/support/processors/000005523.html

Hmm, weird. It stays up by default on Z170 boards.

[MageTank]

Just now, patrickjp93 said:

Hmm, weird. It stays up by default on Z170 boards.

Oh. I thought it was due to multi-core enhancement being enabled. Even checked ASrock's advanced turbo options, which at the time, seemed to be the reason. You are right though. I turned off the advanced turbo features and Prime95 still keeps my 6700k at 4.2ghz. Though, on my H170 Pro4, it goes down to 4ghz. Temps are still perfectly fine, as this is on a delidded 6700k with CLU, and i am not even running 48k (temps are only at roughly 66C in both test beds). 

 

Either way, It doesn't matter. You are right regardless because after checking the links for the tests above, they all used Zx70 boards, meaning all-core turbo was at the peak of their boost tables. I'll redo the math when I am less tired tomorrow. 

[patrickjp93]

5 minutes ago, MageTank said:

Oh. I thought it was due to multi-core enhancement being enabled. Even checked ASrock's advanced turbo options, which at the time, seemed to be the reason. You are right though. I turned off the advanced turbo features and Prime95 still keeps my 6700k at 4.2ghz. Though, on my H170 Pro4, it goes down to 4ghz. Temps are still perfectly fine, as this is on a delidded 6700k with CLU, and i am not even running 48k (temps are only at roughly 66C in both test beds). 

 

Either way, It doesn't matter. You are right regardless because after checking the links for the tests above, they all used Zx70 boards, meaning all-core turbo was at the peak of their boost tables. I'll redo the math when I am less tired tomorrow. 

Fair enough. Rest well.

[RagnarokDel]

2 hours ago, Spork829 said:

10-15% improvements! Oh boy oh boy!

Looks like a 4% improvement on multi-thread.

[patrickjp93]

12 minutes ago, RagnarokDel said:

Looks like a 4% improvement on multi-thread.

Amdahl's Law. Scaling gets diminishing returns even if there's 0 overhead to launching and merging threads. You can't just give a flat % comparison between the multithreaded scores to say multithreading is x% better.

[RagnarokDel]

1 minute ago, patrickjp93 said:

Amdahl's Law. Scaling gets diminishing returns even if there's 0 overhead to launching and merging threads. You can't just give a flat % comparison between the multithreaded scores to say multithreading is x% better.

he claimed up to 15% better which is clearly not the case and who cares about that stupid law? It's cool in theory but in world practice, if it takes 96 seconds to do one thing and 100 with the previous generation, it's 4% faster, that's it and that's all you truly need to know.

 

I'm sorry to say that but whenever I read stuff like that it reminds me of a WoW nerd saying you cant do one thing cause you're missing 20 dps or some shit.

[patrickjp93]

2 minutes ago, RagnarokDel said:

he claimed up to 15% better which is clearly not the case and who cares about that stupid law? It's cool in theory but in world practice, if it takes 96 seconds to do one thing and 100 with the previous generation, it's 4% faster, that's it and that's all you truly need to know.

 

I'm sorry to say that but whenever I read stuff like that it reminds me of a WoW nerd saying you cant do one thing cause you're missing 20 dps or some shit.

tbf 20dps in a 30-minute fight is 36,000 damage you're not doing.

 

Also, Amdahl's Law matters. Gustafson's Law does not change that certain portions of tasks will never get faster and will always be the bottleneck.