The launch of Intel’s new Kaby Lake processors may not be an easy one. Not only has the NDA been broken in every way imaginable but after the relatively uninspiring performance uplifts of Skylake and (to a lesser extent) Haswell, expectations are understandably diminished this time around. There is some hope though since I’m here to say that if you’ve been waiting to upgrade your Sandy Bridge or Ivy Bridge system, now may be a great time.
The performance, overclocking capabilities and other elements of Kaby Lake may have been leaked to no end by unscrupulous elements of the media but beyond that, there is a lot to like about this refreshed architecture. It runs at higher speeds, pricing remains competitive and it happens to be backed up by an extremely capable new chipset.
At its heart Kaby Lake actually represents a significant tectonic shift for Intel. This is the first processor generation that officially moves away from their ubiquitous “tick-tock” cycle of a manufacturing process technology shrink followed by a core architecture change. While the rollout from Haswell to Devil’s Canyon could also be considered a minor example of this, the 14nm Skylake architecture along with its subsequent refreshes like Kaby Lake and its upcoming successors breaks a longstanding tradition.
In short, the 14nm node will be around for the foreseeable future as tick-tock gives way to a three-step of a new manufacturing process followed by an architecture change and then core optimizations / refreshes. Kaby Lake can be considered an optimization of Skylake but there’s enough new here –at least on some levels- to pique your interest.
Before I get too far into this article and its nitty-gritty architectural details, let’s discuss Intel’s refreshed Kaby Lake-S lineup. It may not look all that interesting at first glance but there are some noteworthy additions and more than a few surprises. For the most part it is a clone of what we saw with Skylake, just with higher core speeds at every SKU level without any associated TDP increase. Remember, this is a refresh and not a whole new architecture.
By and large what we are seeing here is a 200MHz to 300MHz increase in raw base / boost clocks over the previous generation. There are a few other changes lurking behind these numbers that also have a positive effect upon inter-generational IPC performance metrics but for the most part much of Kaby Lake’s benefits will like within its achievable frequency range. As a result the uplift over Skylake chips should be somewhere in the neighborhood of 5-12% depending on the scenario.
Kaby Lake’s pricing also remains aligned with Skylake but there are a few moves to very slightly lower brackets. I’d expect this to change as AMD’s intents for Ryzen become clearer and Intel adapts their strategies accordingly.
Moving a bit further down-market and this is where I think Intel’s new lineup really comes into its own. As we’ve seen in the past, the more affordable i5 and i3 options actually provide some great bang-for-buck performance while –if you are a gamer- providing in-game framerates that are extremely close to those achieved by higher end CPUs. Even here there are some incremental 200MHz clock speed improvements when moving to Kaby Lake and the dual core, quad thread i3-series processors still lack a Turbo mode.
With that being said, arguably the most interesting Kaby Lake processor lies right at the center of this chart: the i3-7350K. While the Skylake generation received the unlocked Pentium Anniversary Edition as a more affordable overclocker-friendly processor, this time around Intel is moving their entry-level K-series part into the mid-range. The i3-7350K commands a $20 premium over the locked and 100MHz slower i3-7320 and it should prove to be one of the most popular CPU’s in the Kaby Lake lineup when it is released into retail channels in late January or early February. I can see this thing pairing perfectly with a compact ITX system.
From a core architectural standpoint there hasn’t really been any changes between Skylake and Kaby Lake. It is still built upon a 14nm process node and incorporates between two and four logical cores. However, due to manufacturing process efficiencies, Intel has been able to expand this refresh’s effective frequency range upwards without negatively impacting TDP values.
Behind the scenes there are some minor upgrades built into Kaby Lake. First and foremost Intel’s Speed Shift technology has undergone a refresh. Like its predecessor in Skylake, the second generation Speed Shift allows the operating system to hand off control of the CPU’s Turbo mode to the processor itself. This means quicker response times to performance requests (about 30 milliseconds) and very fine granularity for P-States. Meanwhile, the second generation of this technology further cuts down the response time to about 15ms.
Another rather significant addition is the incorporation of support for Intel’s next generation Optane memory at the platform level. Billed as a low-cost option to drastically increase performance through the use of 3D Xpoint technology and non-volatile memory, Optane is supposed to combine memory and storage subsystems into one blazing-fast device. In its most basic implementation you will be able to pair up Intel Optane Memory with a traditional hard disk to boot system and application load times. However, this riff on Intel’s older Smart Response Technology is just the beginning and its capabilities will surely expand at a breakneck pace provided costing remains competitive.
The only real limiting factor with Optane compatibility is that it will need to be enabled by motherboard vendors on their products and it will require a Kaby Lake processor. This will be done via the addition of an Optane-certified M.2 slot which has access to a minimum of four PCIe 3.0 lanes from the Z207 chipset.
All in all it looks like with the addition of fast speed bins, a new mid-tier overclockable SKU and Optane memory support, there could finally be a good reason for some hold-outs with older systems to upgrade. And yet even with these moves towards a more competitive product stack, will Kaby Lake really offer enough to capture the hearts and minds of enthusiasts? Let’s find out.
The performance, overclocking capabilities and other elements of Kaby Lake may have been leaked to no end by unscrupulous elements of the media but beyond that, there is a lot to like about this refreshed architecture. It runs at higher speeds, pricing remains competitive and it happens to be backed up by an extremely capable new chipset.
At its heart Kaby Lake actually represents a significant tectonic shift for Intel. This is the first processor generation that officially moves away from their ubiquitous “tick-tock” cycle of a manufacturing process technology shrink followed by a core architecture change. While the rollout from Haswell to Devil’s Canyon could also be considered a minor example of this, the 14nm Skylake architecture along with its subsequent refreshes like Kaby Lake and its upcoming successors breaks a longstanding tradition.
In short, the 14nm node will be around for the foreseeable future as tick-tock gives way to a three-step of a new manufacturing process followed by an architecture change and then core optimizations / refreshes. Kaby Lake can be considered an optimization of Skylake but there’s enough new here –at least on some levels- to pique your interest.
Before I get too far into this article and its nitty-gritty architectural details, let’s discuss Intel’s refreshed Kaby Lake-S lineup. It may not look all that interesting at first glance but there are some noteworthy additions and more than a few surprises. For the most part it is a clone of what we saw with Skylake, just with higher core speeds at every SKU level without any associated TDP increase. Remember, this is a refresh and not a whole new architecture.
By and large what we are seeing here is a 200MHz to 300MHz increase in raw base / boost clocks over the previous generation. There are a few other changes lurking behind these numbers that also have a positive effect upon inter-generational IPC performance metrics but for the most part much of Kaby Lake’s benefits will like within its achievable frequency range. As a result the uplift over Skylake chips should be somewhere in the neighborhood of 5-12% depending on the scenario.
Kaby Lake’s pricing also remains aligned with Skylake but there are a few moves to very slightly lower brackets. I’d expect this to change as AMD’s intents for Ryzen become clearer and Intel adapts their strategies accordingly.
Moving a bit further down-market and this is where I think Intel’s new lineup really comes into its own. As we’ve seen in the past, the more affordable i5 and i3 options actually provide some great bang-for-buck performance while –if you are a gamer- providing in-game framerates that are extremely close to those achieved by higher end CPUs. Even here there are some incremental 200MHz clock speed improvements when moving to Kaby Lake and the dual core, quad thread i3-series processors still lack a Turbo mode.
With that being said, arguably the most interesting Kaby Lake processor lies right at the center of this chart: the i3-7350K. While the Skylake generation received the unlocked Pentium Anniversary Edition as a more affordable overclocker-friendly processor, this time around Intel is moving their entry-level K-series part into the mid-range. The i3-7350K commands a $20 premium over the locked and 100MHz slower i3-7320 and it should prove to be one of the most popular CPU’s in the Kaby Lake lineup when it is released into retail channels in late January or early February. I can see this thing pairing perfectly with a compact ITX system.
From a core architectural standpoint there hasn’t really been any changes between Skylake and Kaby Lake. It is still built upon a 14nm process node and incorporates between two and four logical cores. However, due to manufacturing process efficiencies, Intel has been able to expand this refresh’s effective frequency range upwards without negatively impacting TDP values.
Behind the scenes there are some minor upgrades built into Kaby Lake. First and foremost Intel’s Speed Shift technology has undergone a refresh. Like its predecessor in Skylake, the second generation Speed Shift allows the operating system to hand off control of the CPU’s Turbo mode to the processor itself. This means quicker response times to performance requests (about 30 milliseconds) and very fine granularity for P-States. Meanwhile, the second generation of this technology further cuts down the response time to about 15ms.
Another rather significant addition is the incorporation of support for Intel’s next generation Optane memory at the platform level. Billed as a low-cost option to drastically increase performance through the use of 3D Xpoint technology and non-volatile memory, Optane is supposed to combine memory and storage subsystems into one blazing-fast device. In its most basic implementation you will be able to pair up Intel Optane Memory with a traditional hard disk to boot system and application load times. However, this riff on Intel’s older Smart Response Technology is just the beginning and its capabilities will surely expand at a breakneck pace provided costing remains competitive.
The only real limiting factor with Optane compatibility is that it will need to be enabled by motherboard vendors on their products and it will require a Kaby Lake processor. This will be done via the addition of an Optane-certified M.2 slot which has access to a minimum of four PCIe 3.0 lanes from the Z207 chipset.
All in all it looks like with the addition of fast speed bins, a new mid-tier overclockable SKU and Optane memory support, there could finally be a good reason for some hold-outs with older systems to upgrade. And yet even with these moves towards a more competitive product stack, will Kaby Lake really offer enough to capture the hearts and minds of enthusiasts? Let’s find out.
