You could get into HEDT relatively cheaply, I suppose. Then upgrade later to a lot more cores.I still don't understand how KBL-X makes any sense whatsoever. There are basically three reasons why an enthusiast might want to go to Intel's HEDT platform: more cores, more PCIe lanes, and quad-channel RAM. KBL-X offers none of them. You're still paying for the more expensive motherboard, but you get the same 4 cores, 16 PCIe lanes, and dual-channel RAM that you would if you went with the cheaper LGA 1151 platform. What's the point? Are there really people who need the absolute fastest single-threaded performance and their task is critical enough that they won't overclock but not critical enough that they need ECC?
Unless it has something extra to offer in the form of significantly increased oc potential over mainstream, buying KBL-X for the upgrade path is the equivalent of paying in advance for something you may or may not need.You could get into HEDT relatively cheaply, I suppose. Then upgrade later to a lot more cores.
Whatever the chipset/board has to offer, it's also available with the cheapest SKL-X SKU.Why throw all the talk about KL-X around in isolation? What about the rest of the package? Aren't the chipsets and boards just as important?
Whatever the chipset/board has to offer, it's also available with the cheapest SKL-X SKU.
Imagine the following 2 choices on the same board:
1. $389 4c/8t overclockable to 5Ghz+
2. $439 6c/12t overclockable to 4.5Ghz+ (improved cache, more lanes etc)
Now take that $50 price delta and divide it by system cost (MB+CPU+RAM+whatever else is new).
Yup, you're damaged goods now. Excellent purchase thoughMaybe that ruined me
I'm hoping that skylake-x will match kabylake at single threaded performance even with a frequency disadvantage. It has a reworked cache system with a much bigger l2 cache and l2 cache is more important for single thread performance than l3 cache so it might get interesting.Unless it has something extra to offer in the form of significantly increased oc potential over mainstream, buying KBL-X for the upgrade path is the equivalent of paying in advance for something you may or may not need.
Looking at 4c/8t vs. 6c/12t prices, it seems to me like the difference between KBL-X i7 and the cheapest SKL-X will be less than $100. Why buy KBL-X now when the upgrade is already on the table? SKL-X is bound to oc higher than BDW, so ST difference won't be as high as it is today either.
.
Yeah, Microcenter ruined me, too. I picked up a pair of Q9300 Core2Quad CPUs on "clearance", back in the day, for $100 + tax ea. Thereafter, that became my benchmark for CPUs, and why I've never, well, until I finally broke down and bought a i5-6400 to BLCK OC, bought a CPU significantly over $100. (Ok, I bought an i3-6100 for $125 too.)Maybe that ruined me, but I refuse to pay more than $250 for a high end unlocked processor with less than 6 real cores.
Summit Ridge launches.... 9 months later... COFFEE LAKE! w/ AVX512 for all!
Great, so in 10 years when it actually starts getting used in games, it might improve framerates from 2fps to 3fps.
Why do people think that cpu architectural enhancements revolve around gaming?
Why do people think that cpu architectural enhancements revolve around gaming?
Probably because gaming is the only reason to own a PC that isnt for work.
I am going to assume that was sarcasm. I don't know anyone (outside of no life gamers) who can say that with a straight face.
If the "PC" was meant to convey a Windows machine I can say that gaming is the only reason that I keep a Windows PC. It's why when you read a CPU review these days more than half the pages are dedicated to gaming benchmarks.I am going to assume that was sarcasm. I don't know anyone (outside of no life gamers) who can say that with a straight face.
Definitely entertaining. Thanks for theHere's another entertaining look back when Keller joined Samsung.
http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=threads/jim-keller-joins-samsung.2452411/
Arachnotronic Intel's EMIB approach takes effect at 10nm++ . When do you think that happens ? If we take that we are only now at 14nm+ and 2018 for 14nm++ and seeing that there is no tock in 2018 Intel's plans and roadmap are just not clear. Skylake was manufactured at 14nm and Kabylake at 14nm+ . So If Icelake arrives in 2019 on 10nm then we are talking about 2022 to see Intel EMIB approach. Intel's roadmap as it pertains to microarchitecture and node revisions is basically vague at this point in time. 10nm is virtually a H1 2018 ramp. I doubt if Intel can even do token shipments of 10nm Cannonlake in Q4 2017. Right now Intel seems to be in disarray. We will see in 2018 how bad are their woes.
FYI, Intel is manufacturing 10nm CPUs now.
Will the real 10nm please stand up?And TSMC/Samsung's are 7nm now.
Also entertaining to go back and read comments about Jim Keller's departure and the future of AMD.
http://www.anandtech.com/show/9643/jim-keller-leaves-amd'
It may be a bit premature (no official benchmarks yet) but I've said many times in the past to never underestimate this guy .
Here's another entertaining look back when Keller joined Samsung.
http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=threads/jim-keller-joins-samsung.2452411/