[情報] MCST 開發出可搭載4TB ram的16核CPU
來源:https://t.ly/rBc9
原文:
This Russian CPU surpasses AMD EPYC, Intel Xeon on one critical feature
By Anton Shilov 13 hours ago
MCST Elbrus-16C shown off in public, and is set to be volume-produced in 2022
(Image credit: MCST)
MCST, a developer of microprocessors and servers from Russia, has shown off
its first 16-core CPU for the first time.
The Elbrus-16C system-on-chip makes another leap forward in terms of catchingup with modern chips by AMD and Intel, but surprisingly the new server
platform can beat AMD’s EPYC server platforms in one critical metrics: DRAM
support.
The MCST Elbrus-16C CPU integrates 16 cores that use a proprietary
microarchitecture that run at 2GHz, an eight-channel DDR4-3200 memory
controller that supports ECC, 32 PCIe 3.0 lanes, four SATA 3.0 ports, and
built-in 10GbE and 2.5GbE interfaces. The chip is powered by a new
microarchitecture that supports hardware virtualization.
Here's our list of the best web hosting services around
Check out our list of the best small business servers on the market
We've built a list of the best dedicated server hosting providers of 2020
The developer says that its 16-core processor will offer performance of 1.5
FP32 TFLOPS as well as 0.75 FP64 TFLOPS. In general, such performance levels
can barely impress. For example, Nvidia’s A100 GPU can hit 19.5 FP32 TFLOPS
and 9.7 FP64 TFLOPS, which is about 13 times higher when compared to the
Elbrus-16C.
心得:
當年蘇聯可是使用自行開發的CPU繞月飛行,
後來有人跑出來開公司做CPU,一度還有挑戰i皇的機會
但當年在俄國募資不順,那個人就跑去intel當高階主管
也帶去MCST的部分授權,有傳言當年intel就是靠這個黑科技打敗AMD的k8
最近幾年回去MCST開發俄國國產CPU,
期待5年後會是3強鼎立XD~
--
電子垃圾
用Optane的話4TB很輕鬆吧
台積16nm 靠台GG就能跟美俄結盟
但EPYC一樣可以4TB的RAM阿...
特定的Xeon也是可以
16條256的模組搭配能存取的記憶體控制器
8 channel 3200 icx的規格
這其實很弱 俄國黨政軍扶持的自主IC
Elbrus16是台積電16nm代工
前一代是Elbrus8, 台積電28nm代工
VLIW架構用binary translation相容X86
大概兆芯Via馬甲等級實力 毛子自主版
核心不行,那就堆料吧...
反正俄黨政軍工業自用 不用美國貨就好
美國現在主要軍事對抗敵人是中國
所以不太在意GG幫俄國代工
反正性能不行 不值得打壓
2 Way SMP的EPYC其實也可以做得到
或者一些加裝memory repeater的Xeon也可以
2 Way SMP算是最容易而且不衝核心數量也不太貴
會像羅斯的航空發動機一樣,數據嚇死人,壽限笑死人
嗎?
樓上 應該是宣傳嚇死人 實際笑死人
69
[情報] Xbox Series X 詳細規格公布CPU 8x Cores @ 3.8 GHz (3.66 GHz w/ SMT) Custom Zen 2 CPU GPU 12 TFLOPS, 52 CUs @ 1.825 GHz Custom RDNA 2 GPU Die Size 360.45 mm237
[情報] AMD宣布以第二代EPYC 攜 NVIDIA A100 打造 DGX A100 超級電業界沒有永遠的敵人, AMD 宣布以第二代 EPYC 攜 NVIDIA A100 打造 DGX A100 超級電 腦系統 by Chevelle.fu 2020.05.18 05:37PM NVIDIA 在上週宣布旗下新世代 GPU 架構 Ampere 安培以及採用 Ampere 架構之超算 AI24
[情報] i皇新10nm Xeon 32核打敗AMD 64核心i皇確定要在2021 Q1量產這顆10nm server CPU (不過每次都說確定量產 阿延期也是你們喊得) Ice Lake-SP 3rd Gen Xeon 主板最高支援兩顆CPU,每個CPU最高2.2GHz,最多32核心27
[情報] AMD很高興幫助Intel和Nvidia啟用SAMAMD is happy to help Intel and Nvidia enable Smart Access Memory 連結: In context: Smart Access Memory, or SAM, is a software/hardware trick that takes advantage of a PCI Express feature called Base Address Register. So far, only AMD is using with their new Radeon RX 6000 GPUs to eek out a little26
[情報] VMware改變授權機制對64核EPYC的傷害?原文來自tom's Hardware : 簡單來說原本的授權費用是取決於你的系統CPU數量,1顆CPU就授權一顆,2顆授權兩顆依此類推,不管核心數量為多少 新的則變成1顆CPU授權上限32核,超過的則要另外購買授權。也就是說假設你的CPU是64核,就算只有一顆也要買兩顆的授權 文中也提到資料中心等使用48核以上EPYC的用戶可能會首當其衝,畢竟整套授權下來可能還比單買CPU還貴15
[情報] 基於Intel 7的新一代HEDT平台!56顆P核Okay so I have a short and sweet update on Intel's upcoming Sapphire Rapids-SP, Sapphire Rapids-112L and Sapphire Rapids-64L platforms. Officially dubbed the 4t h Generation Intel Xeon Scalable Xeon W-3400 and Xeon W-2400 CPUs and powered by the company's W790 platform, Intel is gearing up for launch very very soon. Intel's high core count work station processors are coming10
[情報] 2-DIE Xeon? 下一代CPU SPR-SP照片流出寫在前面: Intel為了對飆AMD CPU的核心數 並不是第一次使用膠水(Cascade Lake-AP) 多了一個Die自然核心數,記憶體和PCIE通道數倍增 良率可能就...2
Re: [情報] 蘇媽Radeon Pro W6800曝光:Navi21 32GAMD公布的數據 Radeon PRO W6800 GPU架構:AMD RDNA2 SP數量:3840