Why is 'Skylake is a meme' a meme? Explain this

File: 68158308.jpg (36KB, 239x239px) Image search: [Google]

36KB, 239x239px

>>57769708
DELET THIS

>>57769708
Because poor faggot fucks who are on their shitty 2600Ks with shitty old motherboards and PCs are trying to justify not upgrading their system even though a stock 6600 (non-K) beats an overclocked 2600K.

>>57769708
Ignorant fags that demand 50% performance increases every generation.

>>57769802
I went from a 2600k to a 6700k and the only thing that ran better (20 fps better) was GTA V.
No other game or application showed any improvement.
It runs much cooler though.

>>57769834
I'm sure you benchmarked every game you played and compared the results before you made this statement.

My 4690k @ 4.7 shits all over memelake

>>57769859
Post cpumark scores fag

>>57769858
I did actually, with several games and applications and wrote a full review (it's in italian but I can link it if you want). GTA V was the only one that ran better. Everything else was around a 5% improvement. That's pretty shitty considering my old 2600k was already 5 years old at the time.

>>>/v/

>>57769826
>Ignorant fags that demand 50% performance increases every generation.

The same thing we had throughout the 80s, 90s and 00s?

>>57769875
>artificial benchmarks

Please end yourself

>>57769901
Dennard scaling is dead, gramps.

File: whenthehotlineblingjustright.png (243KB, 600x652px) Image search: [Google]

243KB, 600x652px

>when Zen is gonna btfo all jewtel chips soon anyway
I seriously hope you didn't do a new build recently anons.

>>57769904
I'll trust them before I trust your personal, subjective opinion. Please enlighten us about your issues, specifically, with pcmark in relation to skylake vs. haswell architecture.

>>57769826
Still happening with GPUs and mobile chips. 40-50% performance increases every 18 month.
The only reason it's not happening with x86 CPUs is Intel being lazy.

>>57769892
Yeah sure link it. Because everything I've read shows improvement. Pretty bold claim that there is no appreciable difference between the two architectures.

>>57769904
>only muh real life tests are valid
And tell me, are they completely repeatable and consistent?

>>57769926
I really hope so, otherwise, Intel will have another decade of 1-2% IPC gain per node

>>57769802
>a new CPU beats its 5-year-old predecessor even when the latter is overclocked

HOLY PENIS WHAT A SCOOP

https://www.youtube.com/watch?v=frNjT5R5XI4


Here.

>>57769963
Craming more cores works great in gpu's and not so well in cpu's. Also, CPUs can have massive gains from new instruction sets, but they tend to be underutilized, partially out of a need for legacy support. And unlike GPUs, there's not much you can cut out to save on die space. The clock speeds on gpus are also rising each generation, making a large difference

>>57770013
Intel only cares about the xeon lines and "mobile" (y, h, and u) lines. They are coming out with a 32 core xeon next. Maybe you can so the math and tell me how much of an improved that is from their 22 core last gen Xeon. Or from a SB xeon. Scaling isn't dead, its just not ipc anymore.

>>57769859
POOR
O
O
R

>>57770160
Intel bought the new company that does the hardware level multithreading, so maybe one of the newer generation intel cpu will get a boost from that. Maybe canonlake but I doubt it, probably the one after canonlake. Its supposed to make the multicore scaling much better (than current implementation)

>>57769802
It's this. Sandy Vag faggots and their "never obsolete" memes trying to convince themselves that there have been no performance improvements since then.

>>57770079
I love how they compare a non K 6500 vs an 2500k.

File: wow.gif (1MB, 300x174px) Image search: [Google]

1MB, 300x174px

>>57769708

>>57770194
The real problem is most desktop stuff doesn't scale all that well with moar corez. Sure, if you have sandy vag era servers that love parallel workloads, upgrading to v5 Xeons will show massive improvement.

>>57770194
Yeah Intel is developing chips with lasers and shit. I wonder how long it will take for us plebs that game and encode our anime or whatever will get software that can use FPGA's.

>>57770093
Xeons now have 50% more cores for the same price as 5 years ago. Abysmal improvements, but still improvements.

None of this has made it to the consumer space. Quadcore processors cost more than they did 5 years ago.