How to start: Account Application
ASGC Resources & Spec
CPU Clusters
Cluster CPU Nodes Cores-Per-Node Cores RAM(GB) Inter-connection Notes
EDR1 AMD Genoa 9654 @2.4GHz 10 192 1920 1536 100GbE Available since 1 Mar, 2024
CPU Cores Software Compiler Benchmark(TFLOPs) Notes
EDR1-192 Cores MPICH3 AOCC 4.1.0 2.9 TFLOPs --
MPICH3 AOCC 4.1.0 6.85 TFLOPs --
OpenMPI4 AOCC 4.1.0+AMD Library 7.1 TFLOPs Current SLURM doesn't support. Will upgrade in the future.
EDR1-384 Cores(2 nodes) MPICH3 AOCC 4.1.0 3.9 TFLOPs --
HDR1-128 Cores(1 nodes) MPICH3 AOCC 4.1.0+AMD Library 2.75 TFLOPs
FDR5 Intel® Xeon® CPU E5-2650 v4@2.20GHz 92 24 2208 128 IB:FDR, Eth:10Gbps
HDR1 AMD Rome 7662 @2.0GHz 6 128 768 1536 IB:HDR, Eth:100Gbps Available since 16 May, 2022
GPU Clusters
GPU Cluster GPU Model GPU-Boards GPU-RAM(GB) CPU-Model Nodes Cores Per Node Cores Sys RAM/Core(GB) Inter-connection
GPU-1080Ti NVIDIA GTX-1080Ti 8 11 Intel® Xeon® E5-2650L v4 8C@2.2G1GHz 8 16 128 24 IB:EDR, Eth:10Gbps
GPU-P100 NVIDIA P100 4 16 Intel® Xeon® E5-2650L v4 8C@2.2G1GHz 4 48 192 10.5 IB:EDR, Eth:10Gbps
GPU-RTX3090 NVIDIA RTX-3090 8 24 AMD EPYC 7302 16C @3GHz 4 32 128 16 IB:EDR, Eth:100Gbps
GPU-V100 NVIDIA V100 8 24 Intel® Xeon® Gold 6126 @2.6GHz 6 24 144 16 IB:EDR, Eth:10Gbps
GPU-A100 NVIDIA A100 8 80 AMD EPYC 7302 16C @3GHz 3 64 192 16 IB:EDR, Eth:100Gbps
GPU-RTX3090 (Dedicated for ASCEM user) NVIDIA RTX-3090 4 24 AMD EPYC 7402 24C @2.8GHz 6 48 288 10.5 IB:EDR, Eth:100Gbps
ASGC研究型服務收費(ASGC Pricing)
本表各系統每⼆⼗四⼩時使⽤量之單價(以 CPUCore-Day 或 GPUBoard-Day 計算)。 (Last Updated: Mar. 2024. Pricing model will be reviewed and revised annually.)
Computing Pricing 計算服務
Model Computing Usage * NTD$ 國內非學術單位使用者 國外學術單位使用者
CPU-FDR5 CPUCore-Day * NTD$0.24 加計 50% 加計 50%
CPU-HDR1 CPUCore-Day * NTD$1
CPU-EDR1 CPUCore-Day * NTD$1.2
GPU-1080Ti GPUBoard-Day * NTD$1
GPU-P100 GPUBoard-Day * NTD$8
GPU-RTX3090 GPUBoard-Day * NTD$40
GPU-V100 GPUBoard-Day * NTD$35
GPU-A100 GPUBoard-Day * NTD$120
Storage Pricing (TB) 儲存與擷取服務
Storage Unit Cost(NTD$) 國內非學術單位使用者 國外學術單位使用者
YEAR NTD$1000 * per TB-year 加計 50% 加計 50%
DAY NTD$3 * per TB-day
Data Transmission 資料傳輸
目前未納入計費
Customization Service 進階服務
依據所需人時計算。額外需開發之軟體、系統或使用介面等,
將另按工時計費(每 168 man-hr 為 NT$ 100,000)
ASGC服務使用單位
本院冷凍電顯核心設施與網格中心(ASGC)合作提供冷凍電顯設施使用者穩定且高效能的大數據處理與運算資源。依據研究需求,客製化從取得數據到完成分析與資料備份保存整體資料處理流程,如同使用專屬設施般,便利的使用 CryoSparc, Relion, CisTEM, JupyterLab 等工具,於整合的高效能雲端計算與儲存資源完成所需分析。使用者可依據需要選擇分析環境與不同效能的 GPU (V100, 3090, P100) 或 CPU 計算資源。每位使用者於計算環境中可使用最大儲存空間為 60TB,另並提供預設儲存二份的備份與長期保存空間。 為維持整體雲端系統的穩定與效能,計算與儲存資源均依據用量列計使用費,每年一月需繳付前一年的使用費。所收取費用均優先改善冷凍電顯大數據分析資源與效能。冷凍電顯計算與儲存資源之管理均以群組為單位,PI 為該群組負責人。各群組間資料均獨立。網格中心每月將提供每位主持人前一月份的使用量與費用報表,並於每年一月寄送繳費通知。服務內容與計費模式均將透過冷凍電顯定期的使用者委員會議中,持續檢討改善。 使用運算資源者,應於學術發表中致謝 TPP、ASCEM、及 ASGC。 使用帳號申請、計費模式、服務內容以及任何問題與建議,均請參見網格中心網站,或逕與網格中心聯繫。

ASGC provides big data computing and storage resource and services for ASCEM users in collaboration with ASCEM. According to the research needs, customized data analysis pipeline is provisioned over the science cloud integrating high-performance GPU, CPU clusters as well as data storage system. CryoSparc, Relion, CisTEM, and JupyterLab are all available over the cloud right as dedicated computing environment. User could choose a variety of GPU or CPU resource for his analysis based on performance requirements. Each user account has 60TB disk space quota. Services for backup and long-term storage with two-copy by default are also provided. In order to maintain the reliability and efficiency of the ASCEM computing system evolutionarily, resource usage fee is a must according to the consumption. Resource usage fee is paid based on PI group annually in January. These payment will be used to enhance the reliability and efficiency of ASCEM computing system precedently. Monthly accounting report will be delivered to each PI including usage of every member account. Any improvement of the services, pricing model and resource policy will be discussed in regular ASCEM User Committee meeting. In addition to ASCEM (Academia Sinica Cryo-EM Facility) and TPP (Taiwan Protein Project (Grant No. AS-KPQ-109-TPP2)), please also include ASGC (Academia Sinica Grid-computing Center) in the acknowledgement when research outcomes that relied on ASGC resources, services or expertise are presented in your research. User account application, pricing model, services and any comment or question, please refer to ASGC web site or contact ASGC directly.