A larger instance size (m6i.32xlarge) with 128 vCPUs and 512 GiB of memory that makes it easier and more cost-efficient to consolidate workloads and scale up applications.
However, when compared to G4dn the new G4ad instances enable up to 45% better price performance for graphics-intensive workloads, including the aforementioned game streaming, remote graphics workstations, and rendering scenarios. Compared to an equally-sized G4dn instance, G4ad instances offer up to 40% improvement in performance.
Ideally we want to have the ARM performing at or above 50% of the Xeon performance per core. This would make sure we have no performance regressions, and net performance gain, since the ARM CPUs have double the core count as our current 2 socket setup.
In this case, however, I was disappointed to discover an almost 4X slowdown.
Not one to despair, I figured out that applying the same optimizations I did for Intel would be trivial. Surely the NEON instructions map neatly to the SSE instructions I used before?
With the new implementation Centriq outperforms the Xeon at batch reduction for every number of workers. We usually run Polish with four workers, for which Centriq is now 1.3 times faster while also 6.5 times more power efficient.
這篇在提醒之後在 ARM 上寫最佳化時,不要只從 SSE porting 到 NEON,要多看一下有沒有其他指令集是有幫助的...
兩者差了 85W,如果以五年來算就差了 3723 度的電,另外再考慮 PUE 與機櫃空間租用的成本,長期應該是頗有機會換掉原來的 x86 系統。反過來看,短期有轉換測試成本以及 (可能會有的) 較高的故障率 (畢竟是白老鼠 XD),再來是機器本身價錢差距,這些都是會想要知道的...
在 tweet 後 Matthew Prince 有回答一些問題,另外可以看到後續會有更多細節會整理出來,但感覺應該是調整的差不多決定會換過去了?這邊算是延續去年十一月「Cloudflare 測試 ARM 新的伺服器」這篇所做的事情,當時他們拿到 ARM 的工程板在測試,就已經跟 Xeon 打的差不多 (有輸有贏),現在應該又改善更多...
看 retweet 數可以看出來大家還滿期待的,畢竟 ARM 上面的 Linux 本來就因為行動裝置很熱,現在主要還是差在有沒有穩定的伺服器可以用。
The X1 instances will be powered by up to four Intel® Xeon® E7 processors. The processors have high memory bandwidth and large L3 caches, both designed to support high-performance, memory-bound applications. With over 100 vCPUs, these instances will be able to handle highly concurrent workloads with ease.
X1 系列預定 2016 年上半年會開放使用:
We expect to have the X1 available in the first half of 2016. I’ll share pricing and other details at launch time.
另外是 T2.Nano,只有 512MB RAM,預定是今年會開放使用:
Later this year we will introduce the t2.nano instance. You’ll get 1 vCPU and 512 MB of memory, and the ability run at full core performance for over an hour on a full credit balance. Each newly launched t2.nano starts out with sufficient CPU Credits to allow you to get started as quickly as possible.