I would say that is not true. Intel basically skipped Westmere on the desktop (other than a few dual cores), so I would say 32nm was as mature for SB as 14nm is for the Skylake launch.
SB launched on a mature 32nm process. It will be interesting to see how SKL does as 14nm matures. It would have been interesting to see how SKL would have done on a more mature 14nm process.
So Skylake overclocks what... the same or 100 Mhz better than Haswell, and still worse than Sandy Bridge. Is that something to cheer about, when Skylake is on 14 nm, Haswell on 22 nm and Sandy Bridge on 32 nm?
If you have a CPU at say 4.5Ghz and get a 6700k at 4.5Ghz the 6700k will be faster. The pure clockspeed doesn't mean that much when IPC goes up. Plus you get the extra PCIe 3.0 lanes from Z170 which as I've said before is the main benefit right now. You can run multiple GPUs, USB 3.1, and PCIe SSDs without worrying about occupying all your PCIe lanes and dropping speed. There are 20 PCIe 3.0 lanes on the Z170.
Fjodor2001 is just fishing for reasons to crap all over Intel. And it is getting tiresome.
So either Intel can release a slower clocked cpu with more over clocking headroom (like the 2600k), or they can release a higher clocked cpu with lower over clocking headroom (4790k, 6700k), but at the end of the day, the final OC speeds are very similar.
What is the difference?
Would it make you feel better is the Skylake was released at 3.4Ghz with the ability to OC to 4.8Ghz? As opposed to 4.0-4.8?
You argument is invalid in my opinion.
You're missing the point. For OC what's important is of course the max frequency (and IPC), regardless of what frequency it is rated at by Intel.
I should report you for posting deliberate lies. I never told Broadwell wasn't a generation (made it clear at least 3 times), I told it didn't replace any of the existing desktop Haswell parts which is true. Broadwell-K belong to a new niche segment (LGA Iris Pro on desktop) while regular 2013 Haswell parts where replaced by Haswell Refresh & Devil's Canyon last year and now Skylake-S.
Skylake is not an earth shattering leap in performance. In our IPC testing, with CPUs at 3 GHz, we saw a 5.7% increase in performance over a Haswell processor at the same clockspeed and ~ 25% gains over Sandy Bridge. That 5.7% value masks the fact that between Haswell and Skylake, we have Broadwell, marking a 5.7% increase for a two generation gap.
The max frequency the average user gets isn't going up significantly anymore but CPU performance still does. You're not going to see 4Ghz parts getting 5.5Ghz overclocks.
..you told Broadwell shouldn't count as a disagreed with you however in their review of desktop Skylake:
I think no one mentioned it. Since Base Clock and other stuff like the PCIe clock domain are now independent, what is the likehood of going back to Base Clock overclocking like in the old days? The important part of it is that if you can do a reliable Base Clock overclock (Instead of now that you're limited by other Buses being more suceptible), you should be able to overclock ALL Processors, not just the K Series. Assumming that Intel sells Xeons E3 V5 with similar specs than Core i5/i7 at better price points (That is what they did with the V2 and V3) and you can overclock them, they will be THE superior option. And we will be finally be able to grasp if they use higher quality bins for them.
Isn't the best part of the BCLK overclocking the fact that it enables more options for pushing up DDR4 memory? Correct me if I'm wrong. I am still on the idea that better and better DDR4 is going to keep pushing this platform's performance up.
The clock difference is obviously offsetting part of the IPC gain. Even then Core i7 6700K still managed to outperform Core i7 4790K by ~7-10% @ Hardware.fr application and gaming suite. Not bad at all.
Didn't AT already measure the IPC and came up with average 5.7% (ok ,let's round it to 6%)?Interesting, after reading the other thread I was surprised about the Turbo clock difference between Core i7 6700K and Core i7 4790K.
- 1 core Turbo
Core i7 6700K: 4.2GHz
Core i7 4790K: 4.4GHz
- 2 core Turbo
Core i7 6700K: 4.0GHz
Core i7 4790K: 4.4GHz
- 3 core Turbo
Core i7 6700K: 4.0GHz
Core i7 4790K: 4.3GHz
- 4 core Turbo
Core i7 6700K: 4.0GHz
Core i7 4790K: 4.2GHz
Core i7 4790K is running at 5-10% higher Turbo clocks depending on the load.
www.intel.com/support/processors/corei7/sb/cs-032279.htm
The clock difference is obviously offsetting part of the IPC gain. Even then Core i7 6700K still managed to outperform Core i7 4790K by ~7-10% @ Hardware.fr application and gaming suite. Not bad at all.
If anyone expected a 4.0Ghz CPU to hit 5.5Ghz or something crazy, they are really out of their mind. a 3.4Ghz part can hit 4.5-4.7Ghz fine, but don't expect a 1Ghz overclock on every CPU.
I'm actually thinking seriously about an upgrade from my Z77 platform to Z170 for the extra PCIe lanes since I could run a PCIe SSD and USB 3.1 devices without dropping the speed of my x16 slots for the GPUs. The HT CPU with better IPC would be a nice bonus.
Didn't AT already measure the IPC and came up with average 5.7% (ok ,let's round it to 6%)?
Fry's and Microcenter need to hurry up and get the 6700k in stock.
This hardly matters. There is a 0-1% drop moving from PCIe 3.0 x16 to PCIe 3.0 x8 on a 980.
Even moving to PCIe 2.0 x8 / 3.0 x4 is just a 4% drop at 1440P.
Upgrading from 970 SLI to 980TI SLI would net a massive performance increase.
As an example in TW3 a single 980Ti ~ 980 SLI
Or when SLI doesn't work at all, 970 SLI, the performance delta between 980Ti and 970 is massive.
Essentially a single 980Ti OC ~ 970 SLI OC.
As far as your example of using a PCIe SSD in the rig, a 512GB one costs almost $400 US. Chances are someone who can afford a $400 512GB PCIe SSD, $200 board, $350 CPU can afford Skylake-E. Since IPC increases are slowing down for Intel, while DX12 will bring more emphasis on multi-core scaling in games, a Skylake-E 5820K successor system should give a gamer a peace better performance over the next 5 years. Even with 5820K, more members of AT voted for it over i7-6700K, nevermind Skylake-E. Some have tried to argue how overclocking X99 CPUs isn't guaranteed but given the $350 US price of 6700K, it's not a big stretch to get a $419 binned 4.6Ghz 5820K.
I think patience will reward Skylake-E buyers since those CPUs should overclock similarly but run cooler since they will have proper solder. Over the next 4-5 years, having 2 extra cores for barely more $ is going to pay off as well. I mean at that point in 2019-2020 we could even get PS5/XB2 consoles. Since it's already August 2015, it's not out of the question that a Skylake-E will survive to next gen consoles. Who is going to be that there will be less than 3 AAA games that benefit from 6-core CPUs over the next 5 years? We already have some games today that benefit. I remember how the entire forum kept recommending i5 over i7 when I got my i5 2500K and I hate to say it but most of those people were wrong. Learning from past mistakes, I think it's going to be history repeating itself.
I am also frankly amazed how many PC gamers on AT are hyping up Skylake upgrade over 4.5Ghz SB/IVB when almost none of these individuals even have a single 980Ti, nevermind 980Ti SLI (or Fury X CF, etc.). Sounds like a lot of people are just looking for excuses to upgrade, which is fine since upgrading is fun but it's doesn't change the fact that it's still not logical when it comes to gaming performance upgrade priorities unless one's monitor and GPUs have been upgraded first. CPU upgrades no longer mean what they did 5 or especially 7-10 years ago. Today, it's all about the GPU(s) and monitor ugprades imho when it comes to gaming experience.
Not to mention, most of the gaming benchmarks online comparing Skylake to previous generation of CPUs aren't using highest in-game settings, skimping on AA as well, or using low resolutions. That's not how most real world gamers play their games. Most gamers do not buy $600+ of GPUs to play games at 1080P non-maxed out. In fact, most gamers will use DSR/super-sampling methods when there is enough GPU horsepower which will make the system even more GPU limited.
When comparing real world gaming situations of proper AA/highest settings, the performance upgrade from even SB i7 2600K OC is trivial.
http://www.bit-tech.net/hardware/2015/08/05/intel-skylake-review/7
Raise the gaming resolution to 1440P and upgrade to a quad-core Skylake is going to be even less.
On x99 for example with certain Asus boards running wifi or usb 3.1 shares bandwidth with pcie slot 2. What they do is drop pcie slot 2 to X1 when those slots are filled. Means your sli performance is crippled. That is unless you have the $500+ cpu with 40 pcie lanes. The 5820k with 28 is insufficient. With the z170 board I have 20 pcie 3.0 lanes available for extra peripherals.
But the thing about IPC is that you want all other things to be equal. You don't know if for instance Haswell would also become 10% faster with the faster memory, then the IPC difference would still be almost zero.You mean the same review where overclockable Skylake-S was crippled by slow DDR4 (for the sake of highest officially supported speed even though it's a 'K' chip) to the point where Haswell was faster in games (and even some applications)? Sorry but I can't take it seriously after checking out other reviews.
Even in the same applications the results differ:
7-Zip 9.2
- AnandTech: Skylake 1% slower than Haswell per clock
- Hardware.fr: Skylake 7% faster than Haswell per clock
- Hardware Canucks: Core i7 6700K 11.8% faster than Core i7 4790K with lower turbo clocks
POV-Ray
- AnandTech: Skylake 9.7% faster than Haswell per clock
- HardOCP: ''POV-Ray is a ray tracing for creating high quality graphics. We are using the benchmark included in the software and using its multicore ability. Again we see Skylake stretch its legs. Skylake rewards us with a 16% decrease in render time compared to Haswell, a 21% decrease compared to Ivy Bridge, and a 27% decrease compared to Sandy Bridge.'' (per clock, fixed 4.5GHz)
WinRAR
- AnandTech: Skylake 0.1% slower than Haswell per clock
- Hardware.fr: Skylake 10.85% faster than Haswell per clock
- Lab501: Core i7 6700K 9.4% faster than Core i7 4790K with lower turbo clocks