謎底終於解開了. NV新GPU用HBM2

本帖最後由 MirageKnight 於 2015-9-22 14:07 編輯

http://www.guru3d.com/news-story ... rtedly-spotted.html

Yesterday some shipping info popped up on the web, indicating that nvidia is moving around what seems to be Pascal., Pascal is the next-generation Nvidia flagship graphics processor tagged under codename GP100.

It seems that a batch of completed chips have been sent from TSMC towards NVIDIA for testing and evaluation.

It is rumored that the flagship GPU will get 17 billion transistors made on the 16 nm FinFET+ node at TSMC. The GPU will get HBM2 memory with allowance up-to 32 GB of memory (very unlikely, more like 8 or 16 GB).

Anyway, aside from the tracking info below, take this news from 3dcenter with a huge grain of salt please.

DDR5真係要byebye了. 完成歷史任務.

咁咪幾好~

TOP

17000M transistors....
可以預期28nm -> 16nm
顯卡性能應該會有一定增長

但係16nm Finfet會唔會又俾apple打爆嫁

TOP

17000M transistors....
可以預期28nm -> 16nm
顯卡性能應該會有一定增長

但係16nm Finfet會唔會又俾 ...
danny9428 發表於 2015-9-22 14:11
應該要十二月Apple需求下降先開始大規模量產Pascal…
但除左NV同Apple人地其實仲有其他大客,所以…

TOP

no more Dram problems.

TOP

提示: 作者被禁止或刪除 內容自動屏蔽

TOP

提示: 作者被禁止或刪除 內容自動屏蔽

TOP

NV 用HBM=技術進步
AMD 用HBM=高成本,賣唔出倒米
BennyG 發表於 2015-9-22 16:44



    N粉又話bit多冇x用 今次上到 4096 bit仲唔爽到飛機都打多幾隻
    仲有N卡可以超貧 A卡冇得超

TOP

NV 用HBM=技術進步
AMD 用HBM=高成本,賣唔出倒米
BennyG 發表於 2015-9-22 16:44

HBM1 VS HBM2
(AMD) HBM1=EXPENSIVE, WASTE

i hope there will be a serious reduction in costs of using hbm2 in the coming year
otherwise we are both f_ _ _ ed

TOP

N粉又話bit多冇x用 今次上到 4096 bit仲唔爽到飛機都打多幾隻
    仲有N卡可以超貧 A卡冇得 ...
EVIS 發表於 2015-9-22 17:07


yes, reviews need to be fair, overclocking is not fair, and high bits of hbm is fair

TOP