Intel Skylake / Kaby Lake

[Asterox]

Weird, wonder why the single thread is slower than the 7700K.

Well you have several options, IPC is higher or lower, or CR15 score is fake, or i7 8700K TurboCore is deliberately disabled. But for what reason, fake CR15 score that does not make much sense.These scores are not top military Aliens secrets, way fake CR15 Singlethread score if official tests are coming soon.

https://forums.anandtech.com/thread...s-out-page-554.2428363/page-607#post-39052497

 

[Arachnotronic]

Here's my 7820X @ 4.6GHz w/ mesh @ 3.2GHz:

It would seem that L2 cache bandwidth is much higher on the 7900X than it is on the 7820X. I wonder why that is...

 

[eddman]

RAM hasn't been connected to Intel chipsets since 2008.

DMI 3.0, or PCI Express 3.0 x4, has bandwidth ~4 GB/s per direction. Fine for most users, but not enough for multiple NVME drives.

You didn't seem to read my comment properly. I've known about that since day one.

Even if RAM is connected to CPU, loading data from storage still needs to go through the chipset. How else can it reach RAM.

You didn't answer my question. IINM, copying data between storage devices is direct, so DMI would play no role here.

Besides loading data from storage to RAM (which I suppose even 4 GB/s is sort of enough, even for NVMe), what other scenarios are there where DMI's bandwidth might come into play?

 

[Timmah!]

Here's my 7820X @ 4.6GHz w/ mesh @ 3.2GHz:

It would seem that L2 cache bandwidth is much higher on the 7900X than it is on the 7820X. I wonder why that is...

U have BOTH 7900x and 7820x? :-O

 

[Ajay]

They are very close.

Yes, the differences are within the well within the margin of error for a test of two samples.

 

[Ajay]

What evidence? I haven't seen a single evidence. Or do you mean the KBL misread entries with CFL due to the KBL PCH? This is no evidence because we know that Z370 uses KBL PCH.

Is that was happened? It did look like same CFL samples were running on Z270.

 

[mikk]

Is that was happened? It did look like same CFL samples were running on Z270.

There is zero evidence that it run on a Z270 board.

 

[Ajay]

There is zero evidence that it run on a Z270 board.

Thanks.

 

[Dayman1225]

8700k Geekbench Score
http://browser.geekbench.com/v4/cpu/3797188

8700 Geekbench Score
http://browser.geekbench.com/v4/cpu/3806297

8400 Geekbench Score
http://browser.geekbench.com/v4/cpu/3800240

 

[LTC8K6]

These leaks seem to suggest that the 4.7 single core boost number might be incorrect.

 

[Arachnotronic]

These leaks seem to suggest that the 4.7 single core boost number might be incorrect.

4.7GHz is the correct number. But don't assume that these benchmarks are actually running at that boost number. Very tricky to get single core boost speeds in an operating system where you have a lot of lightweight threads doing work all the time.

 

[majord]

4.7GHz is the correct number. But don't assume that these benchmarks are actually running at that boost number. Very tricky to get single core boost speeds in an operating system where you have a lot of lightweight threads doing work all the time.

Never been a problem for any other processor, why suddenly now? Sounds more like the 1 core turbo is more akin to XFR on ryzen than a traditional turbo state

 

[Dave3000]

Well you have several options, IPC is higher or lower, or CR15 score is fake, or i7 8700K TurboCore is deliberately disabled. But for what reason, fake CR15 score that does not make much sense.These scores are not top military Aliens secrets, way fake CR15 Singlethread score if official tests are coming soon.

https://forums.anandtech.com/thread...s-out-page-554.2428363/page-607#post-39052497

Shouldn't the 8700k score a little higher than the 7700k on the single-thread CBR15 test since it's been mentioned that it does 4.7GHz on single-core turbo and 4.6GHz on dual-core turbo if the specs are going to be true? I know Windows sometimes has light background tasks going on that are transparent to the user but then at worst shouldn't it drop to 4.6GHz on the 8700k and 4.4GHz on the 7700k if the same background tasks were happening during the test (if there were background tasks during testing)?

 

[Arachnotronic]

Never been a problem for any other processor, why suddenly now? Sounds more like the 1 core turbo is more akin to XFR on ryzen than a traditional turbo state

Have you ever actually used an Intel processor? Try to "force" 1-core turbo to work, and you'll find that it's not as easy as you make it sound.

 

[maddie]

Have you ever actually used an Intel processor? Try to "force" 1-core turbo to work, and you'll find that it's not as easy as you make it sound.

So what's happening on the 7700K system then? I thought that 3 of these tested are Intel? I could write more but won't.

 

[Arachnotronic]

So what's happening on the 7700K system then? I thought that 3 of these tested are Intel? I could write more but won't.

7700K runs 4-core turbo @ 4.4GHz. If 8700K is being pushed into 4-core turbo mode then scores should be similar, and if 5-6 cores are being hit with work, then you're going to hit 4.3GHz across the cores.

 

[Drazick]

@Arachnotronic,
Are you the lawyer / PR of Intel or something?
I really don't understand the logic you present.

If one CPU Single Core Turbo is higher than the other I'd assume that for 1 core test it would have better performance.
The above holds for any other generation of Intel processor (If the architecture is the same and only boost turbo is higher).

It doesn't happen here, so either:

1. This is not the Turbo.
2. Intel changed its Turbo definitions.
3. The test isn't valid.

It seems you keep trying explaining Intel in very curvy ways.
Starting to sounds similar to press officer on a press conference after a scandal.

 

[mikk]

Have you ever actually used an Intel processor? Try to "force" 1-core turbo to work, and you'll find that it's not as easy as you make it sound.

This is quite easy in benchmarks where you can force it to run with 1 Thread only which is the case for Cinebench or Geekbench ST. I think this is a Bios problem, ST Turbo not working there.

 

[Pandora's Box]

Overclocked my ram tonight. from DDR4-3200 @ 14-14-14-34 1.35V to DDR4-3800 @ 16-16-16-34 1T 1.40V

 

[LTC8K6]

@Arachnotronic,
Are you the lawyer of Intel or something?
I really don't understand the logic you present.

If one core Turbo is higher than the other I'd assume that for 1 core test it would have better performance.
The above holds for any other generation of Intel processor (If the architecture is the same and only boost turbo is higher).

It doesn't happen here, so either:

1. This is not the Turbo.
2. Intel changed its Turbo definitions.
3. The test isn't valid.

It seems you keep trying explaining Intel in very curvy ways.
Starting to sounds similar to press officer on a press conference after a scandal.

I think the single core turbo might be the same as the 7700K.

 

[frozentundra123456]

So much ado about a hundred mhz or two and a few points in Benchmarks. People, the base clock on the 8700K is a lot lower than the 7700k. Seems a gross oversimplification to expect even a "single thread" task to run exclusively on one core of a six core processor. Plus there is variation from chip to chip. Personally, I would *expect* a six core chip with a 500 mhz lower base clock to be slightly slower in even a "single threaded" task than the 7700K. TBH, I really dont attach much weight to the turbo 3.0 or whatever intel calls that one thread turbo (same for the AMD equivalent). These IMO are very ideal situations, made mainly for marketing purposes, and probably wont ever be seen in real life.

And no, I am *not* a lawyer for intel, and have no financial interest in either Intel or AMD.

 

[LTC8K6]

So much ado about a hundred mhz or two and a few points in Benchmarks. People, the base clock on the 8700K is a lot lower than the 7700k. Seems a gross oversimplification to expect even a "single thread" task to run exclusively on one core of a six core processor. Plus there is variation from chip to chip. Personally, I would *expect* a six core chip with a 500 mhz lower base clock to be slightly slower in even a "single threaded" task than the 7700K. TBH, I really dont attach much weight to the turbo 3.0 or whatever intel calls that one thread turbo (same for the AMD equivalent). These IMO are very ideal situations, made mainly for marketing purposes, and probably wont ever be seen in real life.

And no, I am *not* a lawyer for intel, and have no financial interest in either Intel or AMD.

Ahhh...but the 8700K supposedly has a dual core turbo of 4.6, which should still have it ahead of the 7700K with two cores boosting...

None of the regular Intel desktop chips have the 3.0 turbo. Just the regular 2.0 turbo.

 

[majord]

So much ado about a hundred mhz or two and a few points in Benchmarks. People, the base clock on the 8700K is a lot lower than the 7700k. Seems a gross oversimplification to expect even a "single thread" task to run exclusively on one core of a six core processor. Plus there is variation from chip to chip. Personally, I would *expect* a six core chip with a 500 mhz lower base clock to be slightly slower in even a "single threaded" task than the 7700K. TBH, I really dont attach much weight to the turbo 3.0 or whatever intel calls that one thread turbo (same for the AMD equivalent). These IMO are very ideal situations, made mainly for marketing purposes, and probably wont ever be seen in real life.

And no, I am *not* a lawyer for intel, and have no financial interest in either Intel or AMD.

Why is it an oversimplification? the concept of a single core turbo is simple, and the number of cores has very little to do with it. The PM should be quite capable of dynamically clocking up which ever core the task is running on - this has been the case for several generations of turbo mode, from both Intel and AMD.

But I agree, much ado about nothing.. I would have been happy to just give benefit of the doubt that maybe turbo mode isn't operating properly on very early BIOS/uCode.. which, you know is perfectly reasonable given how far off release is.. but, no people instead went off on this illogical tangent, which instead suggests that Intel's turbo mode is flawed, and incapable of actually reaching its advertised max turbo clock under even the most ideal circumstances (a single threaded benchmark) GG!

 

[nvgpu]

http://www.dell.com/learn/us/en/usc...l-unveils-new-laptops-vr-headset-and-services

XPS and Inspiron notebooks getting Kaby Lake-R refreshes.

 

[Timmah!]

Well, 28. august is here, no reviews of 7920x? Or we are still waiting for exact hour of the day?