Polaris and Pascal are on Samsung's 14nm and TSMC's 16nm, respectively, FINFET technologies and they come out this year. Both have working silicon already. It's not the case that they'll only be seen in another year like you just said. However, everything else you said is correct (I'm pointing this out so no-one assumes I've just disproved everything else you wrote, even if they're DJBarry).
Kaby Lake ? Common , it's "SkyLake Refresh" or basically SkyLake
In the past , Intel did so many "refresh" like this but they didn't dare to call them "new architecture"
For example : i7 4700MQ and i7 4710QM , i5 4670k and i7 4690k ...
10% IPC improvement is not likely to happen
It's not a refresh. This is painfully obvious from the IGP specs (the only part of the architecture that we actually know anything about at this point). Intel has so far done only 1 refresh on their consumer line and it was marketed as such. Stop making stuff up when you have no evidence to support it.
AnyOldName3 Wrote:Polaris and Pascal are on Samsung's 14nm and TSMC's 16nm, respectively, FINFET technologies and they come out this year. Both have working silicon already. It's not the case that they'll only be seen in another year like you just said. However, everything else you said is correct (I'm pointing this out so no-one assumes I've just disproved everything else you wrote, even if they're DJBarry).
Yeah I should change next year to late this year. I'm willing to bet they'll run into yield issues this year and will likely just stick to making enthusiast chips until next year or very late this year. FINFET is really tricky. The fact that Intel was able to get the yields up so fast during that transition is a testament to their pioneering work in material physics.
Yes it is a refresh
Quote:According to the results, the Core i7 7700K is a quad-core chip running at 3.60GHz (up to 4.2GHz Turbo) and features eight threads, 8MB of L3 cache, and includes integrated graphics with 24 execution units and the same 1,150MHz clockspeed as the Core i7 6700K (Intel HD Graphics 530, GT2, 24 units).
Running at the 4.20GHz Turbo clock frequency in Windows 10 x64 (and perhaps using an engineering sample), the benchmarks show 118.71 GOPS (giga-operations per second) across eight threads, 313.84 megapixels per-second in the multimedia test, 35.30 GOPS in the Microsoft .NET arithmetic benchmark, 5.59GB/s cryptographic performance, 23.2 nanoseconds DDR4 latency, and 37.41 megapixels per-second GPU performance.
http://fudzilla.com/news/40596-intel-core-i7-7700k-kaby-lake-cpu-benchmarks-leaked
This is
i7 6700k benchmark for comparison
i7 6700k is slightly faster than i7 7700k . The Lower base clock , the slower CPU like I suppected .
Side by Side comparison
http://www.mediafire.com/view/qdojmu822hv24mg/i7_7700k.JPG
Even if that's true (I have my doubts considering even the engineering samples of kaby lake aren't supposed to be available for months) it doesn't change the facts. Clock rates do not make something a refresh. Identical architectures make something a refresh. We already know from what limited info. we have from confirmed sources that usb 3.1, optane, HDCP 2.2, and 10 bit H.265 hardware decode/encode are all supported by the new architecture. So at the very least they've revamped the I/O and IGP logic. My next HTPC cpu is likely going to be a kaby lake celeron/pentium since I need 10 bit H.265 hardware decode for my 4K rips.
At this point I'm getting really annoyed at display technology - there are still no good 4k 120/144Hz panels at affordable consumer price! I'm probably gonna get a 2k one, which is better than 1080p, but... bleh, I want my 4k.
EDIT: Also, I have a question about expensive motherboards: are they ever worth it? As a concrete example, take
this one. I can't figure out how to link to the different versions on amazon, but check the "GODlike gaming carbon": $630. And its stats are essentially identical to the SLI Plus, which is $230. Now obviously some of it is hype for "hardcore gamers" to buy overpriced crap, but there has to be something to justify a 174% price increase, no?
Some manufacturers reserve certain features, like their best sound chips or USB 3.1 Type C and Thunderbolt for their top-end motherboards (although a PCIe card with the same features will likely be cheaper). Also, things like the quality and quantity of VRMs can differ, but that's only really important for major overclocks. I've also seen motherboards advertised as waterproof, which isn't even for watercooling (as that shouldn't leak) but instead for liquid nitrogen cooling, as that can cause condensation which can kill things.
tl;dr: There may be some people who'll benefit from the top end boards, but that group almost certainly doesn't include you.
Alright yeah that definitely doesn't include me. I use a sound card, and almost never use external drives so I wouldn't need usb3.1 and such, and I can never be bothered to OC too much.
Also why would you even use liquid nitrogen cooling? At that point it seems like it would be cheaper and easier to just get a server-like build with extra CPUs or something. Unless you need crazy single-core performance for some reason, but I can't think of any task right now which would require that - gaming is notoriously hard to parallelise, but if you're bottlenecked by your CPU when gaming and want to use liquid cooling you're doing something very wrong.
I guess you could use it to cool your GPUs but again, you can always add more. If 4-way SLI with whatever's the current highest-end card isn't enough for you, you should consider getting a server farm or something.
There's pretty much no way to use liquid nitrogen to cool a CPU that is reasonable to actually do to a computer you're going to use. It's basically a niche hobby to get ridiculous overclocks just to say you've done it.
Apparently the niche is wide enough for manufacturers to make specialised equipment for it, if you can actually buy a waterproof mobo. I can kinda see the appeal though.