AMD IS FINSIHED

>>59202597
Yeah, but muh video games.

yeah but mu CSGO at 768p

>>59202597

Wprime isn't indicative but winrar single could improve.

>>59202656
Winrar seems like a real outlier, considering Zen has more than enough bandwidth and integer performance.

>>59202597
yeah but my 4k netflix

>>59202597
You're forgetting about the really important things like CS 1.6 at 480x320

Amdrone...

>>59202597
>6950X
>Implying anyone buys that thing except millionaires and trust fund babbys

>>59203045
Isn't it nice that you can buy that performance for $500 now ;)
Or $320 if you bother with overclocking and worse perf/watt

>>59202597
It's been confirmed that it supports ECC, so yeah, damn good performing workstation CPU for dirt cheap(compared to competition at least)

>>59202597
yeah but muh ecc

File: Ryzen power.jpg (75KB, 711x533px) Image search: [Google]

75KB, 711x533px

>>59202597
>over all the 1800X has 99% of the performance of the 10c/20t $1700 i7 6950X.
>consumes less power
>has ECC support
>virtualization

Pretty good.

>>59203767
In some cases, the 6950X is 10 cores, which is 20% more than the 6900K, the 1800X loses some stuff even to the 8 core 6900k, but in some cases it can even beat the 6950X, which the 6900K will never, ever beat in productivity. The 6900k will never beat a 6950X in supported software.

>>59202682
>1800X boards support 2 channels of memory vs the 6950x supporting 4 channels of memory
>both support DDR4
Maybe this is why and on top of that the 6950x is just more powerful in general.
Obviously the 1800X is still amazing for the price.

>>59202597
why winrar performance is so poor?

>>59205067
dual channel vs quad channel from what we can muster.

It's one of those 'why is this happening' situations like situations where it beats a 6950X with these new chips.

>>59205083
Its actually from the thread incurring huge latency penalties by moving around different areas of the L3.
7zip doesn't have that problem.

File: Capture2.png (52KB, 1349x385px) Image search: [Google]

52KB, 1349x385px

>>59202682
>>59202656
>>59205067
>>59205083

I think it's the memory latency being a problem, more so than quad vs dual channel.

Benches on guru3d show the 1800X hitting 50GB/s, which is in and around quad channel territory anyway, but it's latency is much worse. It doesn't seem to go lower than 80ns, where intel chips can hit 40 ns.

That's my theory anyway

File: RYZEN7-1800X-50.jpg (89KB, 614x401px) Image search: [Google]

89KB, 614x401px

Here's hardwarecanucks Winrar, for some reason it doesn't lose like a tard here.

>>59205129
they fucked up memory again?

>>59205109
>>59205129

probably some zip NSA backdoor optimized for ME, the more you know but yeah looking forward for the next revision to see if it can be fixed

File: DataCenter.png (1MB, 1326x693px) Image search: [Google]

1MB, 1326x693px

>>59205154
No, they have massive bandwidth for a dual channel but high latency.
Probably a design decision.

Guess that 'disruptive bandwidth' promise was kept.

>>59205193
Having marginally higher IMC efficiency is not disruptive.
What is meant on that slide is that Naples supports 8 memory channels. That is disruptive bandwidth in the enterprise market.

>>59205129
>>59205193

Could be RAM related, still waiting on passmark AMD ddr4 benches to be populated