AMD永恆~~GTX 1080非公可超至2.5GHz

dmm.co.jp

轉貼機神譯:
NVIDIA的GeForce GTX1080在液體冷卻抵達版為2.5GHz
哇：在的GeForce GTX據報導，1080會看到模型，其GPU主頻的* INSANE*2.5GHz的（納斯達克代碼：NVDA，TPE：2357，TPE：2376，TPE：2377）

據報導，會有的GeForce GTX1080:.方正版（參考版），正常AIB卡（風冷），自定義AIB卡（風冷），然後自定義AIB卡（水冷）索泰，七彩虹的四個版本，華碩，微星，技嘉和無數其他都會有一些超級動力的GTX1080顯卡炫耀，與技嘉即將推出的复刻遊戲GTX1080應在GPU達到2.4GHz的傳聞，都在空氣中 - 如果這是真的，準備好了你的頭腦吹。

液體的冷卻挑逗的GeForce GTX1080與GPU主頻為2.5GHz應該是足夠的人留下深刻的印象，因為它會允許4K60FPS的遊戲沒有問題。我們應該期望幾AIB合作夥伴有一些瘋狂的設計，在兩週的時間2016年Computex上展示過。


NVIDIA GeForce GTX 1080 to arrive in liquid cooled version at 2.5GHz
WHOA: The GeForce GTX 1080 will reportedly see models with its GPU clocked at an *INSANE* 2.5GHz (NASDAQ:NVDA, TPE:2357, TPE:2376, TPE:2377)



There will reportedly be four versions of the GeForce GTX 1080: the Founder's Edition (reference edition), the normal AIB card (air cooled), the custom AIB card (air cooled) and then the custom AIB cards (watercooled). ZOTAC, Colorful, ASUS, MSI, GIGABYTE and countless others will have some super-powered GTX 1080 video cards to show off, with the rumor that GIGABYTE's upcoming Xtreme Gaming GTX 1080 should reach 2.4GHz on the GPU, all in air - and if that's true, get ready to have your mind blown.

The tease of a liquid cooled GeForce GTX 1080 with its GPU clocked at 2.5GHz should be enough for anyone to be impressed, as it'll allow for 4K 60FPS gaming without a problem. We should expect a few AIB partners to have some insane designs, shown off at Computex 2016 in two weeks time.

http://www.tweaktown.com/news/52 ... n-2-5ghz/index.html

sunnykoomos

咩料......

ricky1992

係呢
點解cpu 上到3-4ghz
gpu 上得1ghz多小小

燕飛

回復 3 #ricky1992

設計問題

via HKEPC Reader for Android

danny9428

Fermi -> Kepler其實都有相似轉變

Kundera

係呢
點解cpu 上到3-4ghz
gpu 上得1ghz多小小
ricky1992 發表於 2016-5-15 02:04

   
Meanwhile quad core skylake's less than 150mm²

aeolian2009

真係好期待啊

rayad

非公應該都會出水凍，宜家興

danny9428

回覆 6# Kundera
i7 3820好似都有返咁上下die size

當然CUDA/GCN core count同x86 core比又好似係另一回事

Kundera

回覆  Kundera
i7 3820好似都有返咁上下die size

當然CUDA/GCN core count同x86 core比又好似係另一回事 ...
danny9428 發表於 2016-5-15 12:46

This applies to x86 as well. There's a sweet point when Δclock speed = Δpower consumption in any manufacturing node (higher leakage = steeper curve). A desktop cpu may pass that point (turboboost) to maximise IPC. But throughput oriented designs (xeon phi and GPUs), in any given thermal limit, core count is more favourable than clock speed (until you get poor yield)