Intel Skylake / Kaby Lake

[Wyrm]

G.Skill has a new and faster memory DDR4-4600MHz CL19 but only in dual channel: https://www.gskill.com/en/press/vie...0mhz-extreme-performance-trident-z-memory-kit. Hopefully, the quad channel will follow soon.

 

[Zucker2k]

What I love about all this is how the resident AMD fans are downplaying the potential, if not obvious, performance figures we've seen in previous leaks. It'll only make the 8700k look even more impressive when the reviews start flying in and it trades blows with R7, in multithreaded workloads - which it will.

 

[mikk]

https://twitter.com/ryanshrout/status/907280680207503361

Computerbase says the same, the scores are too low hinting towards a Turbomode issue. They are also saying that there is some more work to do on the Bios front.

https://www.computerbase.de/2017-09/intel-core-i7-8700k-benchmarks-turbo/

 

[tamz_msc]

Computerbase says the same, the scores are too low hinting towards a Turbomode issue. They are also saying that there is some more work to do on the Bios front.

https://www.computerbase.de/2017-09/intel-core-i7-8700k-benchmarks-turbo/

Wow, you guys are ignoring your own earlier leaks and just focusing on this "low" score and what everybody in the universe has to say about it in order to drum up expectations even further:

Everyone paying attention to the previous leaks will notice this score is bogus. Earlier leak showed 1.410 pts @ CB R15 MT. Retail systems using fast RAM could do even better.

Ryan Shrout better keep his mouth shut until he gets to review the actual product because clearly he seems to have jumped the gun:
https://twitter.com/ryanshrout/status/907290230377664513

 

[beginner99]

It'll only make the 8700k look even more impressive when the reviews start flying in and it trades blows with R7, in multithreaded workloads - which it will.

They will then use Cinebench as reference point which ironically they called irrelevant when AMD had supbar performance in it.

But both sides are kind of ridiculous. If gaming especially high FPS gaming is your main concern, then the 8700k is the top dog. If you also do encoding, DC /multi-threaded efficiency matters, then the 1700 is hard to beat, if you game at 4k, a 1600(x) is probably the best bet as you will be 100% GPU limited anyway with the exception of a few games like SC2.

 

[VirtualLarry]

Intel i7-8700K Worse Than Ryzen R5 1600X in Cinebench
https://www.eteknix.com/intel-i7-8700k-worse-ryzen-r5-1600x-cinebench/

LOL...

 

[Arachnotronic]

Intel i7-8700K Worse Than Ryzen R5 1600X in Cinebench
https://www.eteknix.com/intel-i7-8700k-worse-ryzen-r5-1600x-cinebench/

LOL...

Dude, pay attention to the discussion. Those scores have been confirmed to be far too low by Ryan Shrout.

 

[tamz_msc]

Intel i7-8700K Worse Than Ryzen R5 1600X in Cinebench
https://www.eteknix.com/intel-i7-8700k-worse-ryzen-r5-1600x-cinebench/

LOL...

Larry this only adds fuel to the fire. See my response above. These numbers can be explained entirely on the basis of what we know of Kaby Lake. When it comes to extrapolating Cinebench scores the 8700K is nothing extraordinary.

 

[tamz_msc]

Dude, pay attention to the discussion. Those scores have been confirmed to be far too low by Ryan Shrout.

It seems that you haven't been paying attention, just like Ryan Shorut.

 

[mikk]

Larry this only adds fuel to the fire. See my response above. These numbers can be explained entirely on the basis of what we know of Kaby Lake. When it comes to extrapolating Cinebench scores the 8700K is nothing extraordinary.

You still don't get it. The recently leaked MT score is 1230 and that's too low. We are talking about this all the time here through the last pages.

https://imgur.com/PId8WeK

 

[tamz_msc]

You still don't get it. The recently leaked MT score is 1230 and that's too low. We are talking about this all the time here through the last pages.

https://imgur.com/PId8WeK

Why don't you discuss the 1410 score from Sweeper's leak then? Go down the thread of the tweet of Ryan Shrout. The reason for the low score is clear. Now Computerbase has picked up on this and the circlejerk continues. The 1230 score is a dead horse.

 

[beginner99]

Intel i7-8700K Worse Than Ryzen R5 1600X in Cinebench
https://www.eteknix.com/intel-i7-8700k-worse-ryzen-r5-1600x-cinebench/

LOL...

locked at 3.7 GHz as has been discussed the last 10 pages or such....

And even more funnier that literally 1 min ago I posted that Cinebench will be the new reference point for AMD fanboys and they did not wait to deliver.

Secondly it's actually only worth in MT, and already faster in ST (162 vs 196). At 4.7 Ghz single-core turbo that would be 249 points in ST and with 4.3 Ghz all-core turbo 1429 in MT which is faster than a stock 1700 with 2 more cores and not too far from a 1700x. To add to that even a 1800x at 4.1 Ghz only gets 171 in ST or said otherwise the 8700k is 46% faster in ST.

Still that doesn't invalidate my previous post. Ryzen are good CPUs but don't let yourself be fooled about single-threaded performance. You can never have too much of it on the other hand it can very quickly be a bottleneck requiring a full platform upgrade.

 

[tamz_msc]

locked at 3.7 GHz as has been discussed the last 10 pages or such....

And even more funnier that literally 1 min ago I posted that Cinebench will be the new reference point for AMD fanboys and they did not wait to deliver.

Secondly it's actually only worth in MT, and already faster in ST (162 vs 196). At 4.7 Ghz single-core turbo that would be 249 points in ST and with 4.3 Ghz all-core turbo 1429 in MT which is faster than a stock 1700 with 2 more cores and not too far from a 1700x. To add to that even a 1800x at 4.1 Ghz only gets 171 in ST or said otherwise the 8700k is 46% faster in ST.

Still that doesn't invalidate my previous post. Ryzen are good CPUs but don't let yourself be fooled about single-threaded performance. You can never have too much of it on the other hand it can very quickly be a bottleneck requiring a full platform upgrade.

The 196 score in ST is the same as a 4.5GHz 7700K. At 4.7 GHz the 8700K ought to score just north of 200 points, not 249.

How much faster is a 5GHz 7700K over stock in gaming in non-esports titles?

 

[mikk]

Why don't you discuss the 1410 score from Sweeper's leak then? Go down the thread of the tweet of Ryan Shrout. The reason for the low score is clear. Now Computerbase has picked up on this and the circlejerk continues. The 1230 score is a dead horse.

Because it's an older leak, this is nothing new. But even in this leak the ST score looks too low, it's even lower than the recent leak in fact. And Ryan Shrout wasn't commenting on this old leak, he referred to the recent leak.

 

[hnizdo]

At 4.7 GHz the 8700K ought to score just north of 200 points, not 249.

U sure? Because 8700K have more cache. Intel itself satted +11% ST performance.

 

[tamz_msc]

Because it's an older leak, this is nothing new. But even in this leak the ST score looks too low, it's even lower than the recent leak in fact. And Ryan Shrout wasn't commenting on this old leak, he referred to the recent leak.

Why waste time typing about leaks which can be explained with a simple query as to whether turbos are working or not?

1410*(3.7/4.3) = 1213.

These numbers aren't technically "wrong", just needs clarification as to what the actual clocks are. It's no big deal like it's being made out to be in the past 5 pages of this thread.

Jumping on this bandwagon is just plain silly.

 

[tamz_msc]

U sure? Because 8700K have more cache. Intel itself satted +11% ST performance.

Cinebench is a benchmark which has minimal thread-to-thread crosstalk and very little movement of data between cache.

We all know how the Skylake-X cache hype ultimately panned out.

 

[PhonakV30]

locked at 3.7 GHz as has been discussed the last 10 pages or such....

And even more funnier that literally 1 min ago I posted that Cinebench will be the new reference point for AMD fanboys and they did not wait to deliver.

Secondly it's actually only worth in MT, and already faster in ST (162 vs 196). At 4.7 Ghz single-core turbo that would be 249 points in ST and with 4.3 Ghz all-core turbo 1429 in MT which is faster than a stock 1700 with 2 more cores and not too far from a 1700x. To add to that even a 1800x at 4.1 Ghz only gets 171 in ST or said otherwise the 8700k is 46% faster in ST.

Still that doesn't invalidate my previous post. Ryzen are good CPUs but don't let yourself be fooled about single-threaded performance. You can never have too much of it on the other hand it can very quickly be a bottleneck requiring a full platform upgrade.

What ? 8700K@4.7 gets 249cb ? Think about it , an intel 7700K@4.7 gets 192cb ( or more ) and 249/192 = 29% , That's not possible when Intel claims 11% more performance and It's still vague what's that mean.

 

[SpoCk0nd0pe]

If you run your 8700k at 4.8 ghz vs an 1700 at 3.8 ghz, the 8700k has 26% higher frequency and due to higher IPC you are already close to 30% higher ST performance. In my book the higher power use is worth the higher ST performance. Games are still mostly limited by a main thread.

Games are limited by a main thread and L3 performance. That's the reason a core complex oc'd ryzen 5 and 7 gives a frame experience similar to the 7700K. But high core complex oc's require a ton of research and tweaking so it's not for everyone. The fact that SMT gives lag spikes in gaming on Ryzen is also really annoying.

Tbh, I think performance differences due to CPU in gaming are generally negligible. Performance gain in the CPU department (not synthetic benches) was disappointingly small over the past years. Saving some bucks for a bigger GPU is the way to go when concerned about gaming performance. But I guess some people just like to boast about 2 more average FPS in game X by spending a holiday's money on a high performance rig before even considering the GPU.

Sorry for my little rant but I think people sometimes loose perspective and get all worked up about things that ultimately just don't matter (this is not directed at beginner99)

 

[ezodagrom]

Tbh, I think performance differences due to CPU in gaming are generally negligible. Performance gain in the CPU department (not synthetic benches) was disappointingly small over the past years. Saving some bucks for a bigger GPU is the way to go when concerned about gaming performance. But I guess some people just like to boast about 2 more average FPS in game X by spending a holiday's money on a high performance rig before even considering the GPU.

Posted this in another forum, may as well post it here too:

People tend to upgrade GPUs faster than they upgrade CPUs, and GPUs become obsolete much faster.
I rather spend a bit more on a better gaming CPU which I won't upgrade again for many years than spend more on a GPU that I'll upgrade in just a few years.

 

[hnizdo]

Cinebench is a benchmark which has minimal thread-to-thread crosstalk and very little movement of data between cache.
We all know how the Skylake-X cache hype ultimately panned out.

In that case SK-X should excell in Cinebench , right? Because of its big L2 and non-inclusive L3. Which is not the case.

 

[tamz_msc]

In that case SK-X should excell in Cinebench , right? Because of its big L2 and non-inclusive L3. Which is not the case.

Cinebench isn't a benchmark that makes the big L2 come into effect. You're more likely to see it's effect when compiling code and running scientific applications.

 

[Zucker2k]

Posted this in another forum, may as well post it here too:

People tend to upgrade GPUs faster than they upgrade CPUs, and GPUs become obsolete much faster.
I rather spend a bit more on a better gaming CPU which I won't upgrade again for many years than spend more on a GPU that I'll upgrade in just a few years.

This was and should be the norm, but the people on the slower gaming cpu found a way to feel good about their setups with talks of "now" and "real world." And yet, they're the first to talk about platform longevity. It's always performance deferred with these guys; "too expensive," "only two fps," "no noticeable difference," etc.

 

[Zucker2k]

It seems that you haven't been paying attention, just like Ryan Shorut.

Shouldn't you rather be criticizing those, like VirtualLarry, who have misconstrued the 'actual result' of this benchmark and are openly celebrating it, instead of the guy trying to correct the mistake, even if he fails to notice that the benchmark was run at pegged base clock?

 

[PeterScott]

More than likely. It will have a slight frequency advantage over Ryzen, but I question whether it will have a power-efficiency advantage. I think Ryzen is better in that dept. We'll see who the real fanboys are, as they decry "oh, who cares about power consumption, as long as it's faster"... when not so long ago, the rhetoric was, "You should get Intel, it's more power-efficient than FX."

Sure and the Ryzen 1800X will have a slight core count advantage. Those might come close to balancing out on highly(>90% parallel) multi-threaded applications.

But on less multi-threaded ones, the Intel chip will be clearly ahead.

Power usage remains to be seen.