Who here is waiting for Ryzen II on AM4+?

I'm squeezing out performance from my i7 950 until then yeah

>>59144058
Not me. I need new pc asap. I might buy it in the future though with new laptop.

>>59144058
I'm waiting for $150 OLED 8K monitors and photonic CPU Intel i7 22000K or AMD k45 series.

>>59144058
Not specifically waiting buttIt'll probably replace my current skylake.
>tfw bought an computer 1.5years to early too use ryzen

>>59144058
Waitfags must truly enjoy suffering

>>59144058
Here
I still use x5470 and it is more than usuable

>>59144299
X5650 mustard race

>>59144058
I'm currently on a temp X79 build i got in late 2016.
I'm waiting on a 40 PCIe lane Quad channel version of Ryzen/X370.
If it's a cut down version of their Naples server chips even better, those tend to get better binning.
Intel's X series has spoiled me for platforms and i don't particularly need the higher IPC.

The two favourite things of the AMD shitter:
1. Waiting
2. Being disappointed


Zen is a perfect storm

>>59144058
The next gen Ryzen will still be on AM4.

>>59144058

im waiting for high ipc 8 core processer under 500 dollars

>>59145219
X79 is quite a win - those cheap octocore Xeon E5s on ebay.

>>59144058
Zen+/Ryzen II won't need a new socket unless PCIe 4.0 rolls out sooner than people expect.

DDR5 isn't coming before 2020 at the earliest, and Zeppelin successors won't add more IO lanes, since they'd be wasted in Naples MCMs, which already support 128 PCIe lanes and 8 DDR4 channels per socket.

I'll let them work out the kinks and get 2nd gen. my $70 locked ivy will hold until then I hope

>>59144278
impatient babby

>>59146570
1800X

>>59144058
I have a feeling that AMD just put their all into Ryzen and now both companies have hit the 14nm limit. There won't be anything special coming out of either company now for a while I think.

>>59147211

why is 14nm a limit? samsung 10nm is already available and intel 10nm is coming next year

>>59147211
AMD clearly has room for improvement though.

For starters:
bigger FPUs, which are currently much smaller than Intel's.
change denser caches for wider, higher performance caches like Intel uses for their mainstream i7 line.
memory controller, AMD seems to be lacking a bit here compared to Intel, this is gonna be a notable bump next generation.
Clockspeed, this is mostly due to GloFo but this can be increase with improved yields, so natural progression.
Improvement on their first gen branch predictor, either in power or performance.


Plenty of high level stuff there, and even with these drawbacks its competitive with Intel.

>>59147275
samshit 10nm is partially 10nm, just like their 14nm is bigger than intel. It's one of the reasons why intel still has problems while trying to achieve true 10nm.

File: cinebench.png (38KB, 351x803px) Image search: [Google]

38KB, 351x803px

>>59144058
If i had anything newer I'd wait, but this cpu is ready for the big sleep.

>>59147276
>bigger FPUs
-But why? It's Floating point performance per clock is like 25% higher than the 7700k.
-I agree its cache could be denser and more vertical. There is a pretty large distance from one side of the shared cache to the die.
-Its memory controller is probably "good enough". I think it's worth the cost savings here. Going with dual channel is part of why it's so competitive with intel. Not supporting quad channel and both DDR3 and DDR4 were significant cost savings.
I think it's clearly a good choice as most of the benchmarks AMD ran against Intel gave Intel quad channel memory so people couldn't claim the 6800k or 6900k was gimped by not having full memory bandwidth.
-Performance/watt is probably rightfully going to be a bigger focus with Zen+ than higher clocks.

>>59147596
>-But why? It's Floating point performance per clock is like 25% higher than the 7700k.
That's under 256bit ops, anything over that Intel has a significant advantage due to how AMD's current FMACs are setup, this might not be too important for consumers but this is pretty much the entire HPC market, and newer AVX versions need this.

It's memory controller according to leaks seems to be higher latency, that's the issue.

>>59147750
Eh okay.

But Intel can't really do heavy 512bit ops without overheating, either.

Anything larger than 64bit or 128bit floats are also really unused and will remain unused for most applications as well, as the GPU does 256bit and 512bit FPO so much better and more efficiently.

Half, single, and double precision on the CPU is far far more relevant.

>>59148059
It overheats, but it doesn't need 2 registers for one instruction, AMD loses badly here.
It's not a consumer upgrade, but it's important in HPC, which makes 20% of the server market and by far the highest margin server market.

>>59144058
Whats supposed to be diff about zen+, better efficency or like 10% max improvement in IPC? Like steamroller to piledriver?

>>59148258
Some things that didn't make it in Zen will be added in Zen+

>>59148258
Supposed to be 30% performance increase (probably not 30% IPC but a combination of IPC and higher clocks) and better performance/watt. It'll be 7nm(really 8.2nm or something) over the current 14nm(really 15.8nm or something).

I imagine it'll have some more features, too.

It'll still be AM4 but might need a new chipset for some new features that may get added. So may not have those features on current chipset boards, but shouldn't actually require a new board.

Coming 2019.