Samsung Memory Solution for HPC - The leverage of right choice of DRAM in improving performance and reducing power consumption of HPC systems - 8. September 2011 Samsung Semiconductor Europe GmbH Gerd Schauss Marketing Intelligence
Samsung Memory 1 HPC: Spearhead of Computing 2 Today & Tomorrow 3 The Day After Tomorrow 4 Summary 5
Samsung, WW#1 Total Memory Solution Provider … … MEMORY for 18 years DRAM for 19 years NAND for 10 years DRAM market share („10) NAND market share („10) Samsung Samsung Toshiba Hynix 38% 40% Hynix Elpida Micron Micron Intel Others Others 3
SAMSUNG Green Memory Solutions 4
SAMSUNG Green Memory Solution SAMSUNG Green solution can save about 86% of Power consumption against DDR2 solution [W] Green Solution 1 Green Solution 2 102W 35% 25% 66W 18% 17% 17% 50W 14% 41W 42% 34W 28W 24W 14W DDR2 DDR3 DDR3 DDR3 DDR3 I/F DDR3 DDR3 DDR3 60nm 60nm 50nm 40nm 40nm 40nm D/R 30nm 30nm 1Gb 1Gb 1Gb 1Gb 2Gb 2Gb Den. 2Gb 4Gb 1.8V 1.5V 1.5V 1.5V 1.5V VDD 1.35V 1. 35V 1. 35V Considered the 8hours active and 16hours idle status in server Source: Measured by Samsung Lab. 5
Samsung announced 32GB with TSV technology Samsung samples 30nm, 32GB DDR3 RDIMMs Aug. 16 th , 2011 .. “The new 32GB RDIMM with 3D TSV package technology is based on Samsung's 30nm-class four gigabit (Gb) DDR3. It can transmit at speeds of up to 1,333 megabits per second (Mbps), a 70 percent gain over preceding quad-rank 32GB RDIMMs with operational speeds of 800Mbps.”… 6
32GB TSV RDIMM Power Evaluation Results TSV RDIMM shows -32% power decrease over LRDIMM@1333 [3DIMM/ch] [2DIMM/ch] [mW] [mW] -32% Common condition : 32GB (based on 30nm 4Gb), RST-Jump - RDIMM : RC AB, 2RCD - 3DS RDIMM : RC AB based, 2RCD Successfully developed POC in Current System 7
Samsung Memory 1 HPC: Spearhead of Computing 2 Today & Tomorrow 3 The Day After Tomorrow 4 Summary 5
Memory Performance Requirement Keeps Growing # of processor core and performance keeps growing CPU + GPU heterogeneous computing needs more fast DRAM Memory bandwidth should increase to hide data I/O time # of core per GPU # of core per CPU 1600 Core 8 (~16) Core … … …… … … 800core 6 Core 320core 4 Core 128core 56 core 6 Core 2 Core 1 Core 1 Core 1995 2000 2005 2010 2000 2005 2010 Last 10 years, # of GPU core increased by 260X and # of CPU core by 8X 9
Memory Performance Requirement Keeps Growing # of processor core and performance keeps growing CPU + GPU heterogeneous computing needs more fast DRAM In current heterogeneous, data motion thru PCIe is bottleneck • Strong movement to go towards On-die heterogeneous • Memory bandwidth should increase to hide data I/O time CPU Future 25GB/s DRAM DDR3 CPU GPU …… PCIe 12GB/s CPU 200GB/s Future GDDR5 GPU DRAM GPU Current Heterogeneous Future Heterogeneous Computing Computing 10
Memory Requirements for Exascale Computing The world is heading forward for exascale computing realized until 2018 *Source: top500.org 10X Performance/Watt is needed compared to current computing • Future computing: ~20pJ/Flop(DPFP) - 20pJ/Flop 50GFLOP/W 10 TFLOP/200W 1EFLOP/20MW (US/EU directive) • Current computing: ~200pJ/FLOP(DPFP) K-Computer (~1.000pJ/FLOP) Not just performance, but performance / watt is important for exascale 11
Samsung Memory 1 HPC: Spearhead of Computing 2 Today & Tomorrow 3 The Day After Tomorrow 4 Summary 5
DDR4 Will Keep Performance Increase Trend [GB/s] 51.2 DDR4 44.8 DDR4-2667 38.4 32.0 DDR4-2133 25.6 DDR3 DDR3-1600 19.2 DDR3-1066 12.8 DDR3-800 DDR2-667 6.4 DDR2-533 DDR-400 DDR-266 2001 2003 2005 2007 2009 2011 2013 2015 Double bandwidth over DDR3 13
Samsung‟s High -Density & High-Speed Solution High-density & High- speed memory increases system‟s value System Performance System Performance per Power (Floating point operation) +10.5% +5% Note: SPEC Power benchmark, Intel Romley platform Note: SPEC CPU benchmark, Intel Romley platform High-density component with Thermal less # of DPC is better Better system performance • 50.7’C 55.4’C Better performance per power • Better thermal environment • 42.7’C 51.0’C 14
DDR4: Optimized for Green & Performance Key value of DDR4 is efficient power with high performance • Adopted many power saving & fast power-down exit features • Saved IO power with POD interface: Suit for high speed VTT=VDDQ/2 [Watt] -30% IO SSTL (DDR3) Core VDDQ VDDQ 1.35V DDR3 1.2V DDR4 (1333Mbps) (1600Mbps) POD (DDR4) 15
How Samsung Keeps Innovation for Green Memory Samsung has been the leader of keeping innovation for higher density with less power High speed at low voltage 2400Mbps 2133Mbps 1866Mbps 1600Mbps 2.5V 1333Mbps 1.8V 1.5V 667Mbps 1.35V 1.25V 1.2V 400Mbps ‘01 ‘03 ‘05 ‘07 ‘09 ‘11 ‘13 ‘15 High capacity with low power Assumption 256 Mb 1 Gb 4 Gb 16 Gb 150 nm 80 nm 40 nm 10 nm class 16
GFX DRAM for Heterogeneous Computing Keeps Evolving Evolution of high-speed with lower-voltage solution been kept DRAM process & design improvement realized much more power/performance efficient solution 17
Samsung Memory 1 HPC: Spearhead of Computing 2 Today & Tomorrow 3 The Day After Tomorrow 4 Summary 5
New High-performance Memory is Getting Needed GPU performance keeps increasing and GFX memory performance requirement keeps growing • Current solution’s limit: 7Gbps(GDDR5) X 512 IO’s = 448GB/s GFX card memory BW trend Territory which needs new < Gbps/IO > ‘ 15: 512GB/s solution 16 (TSV, diff- IO…) 1TB/s Serial-IO 12 GDDR4 ‘11: 256GB/s SDR GDDR GDDR3 GDDR5 8 ‘ 08: 128GB ‘06: 64GB 4 Wide-IO ‘04: 32GB 1536 128 256 512 768 1024 2048 Single GPU Memory BW history < # of I/O > Existing Projection solution 19
Consideration for Next High-performance Memory Several solutions can be considered • To meet performance requirement within power budget for Exa-scale BW per Memory BW System Watt /(GB/s) DRAM pkg per Processor configuration DRAM GDDR5 0.9X of DDR3 ~28GB/s ~400GB/s Processor PCB DRAM Si Interposer Wide-IO 0.3X of DDR3 100+GB/s ~1TB/s Processor PCB DRAM Serial + 0.5X of DDR3 100+GB/s ~1TB/s Wide-IO Processor PCB 20
TSV in Memory application Can achieve more stacking & connection with thin profile • More stacking High density with less electronic loss • More connection Many IOs (Better performance) Wire Bonding Type Thru Via Type But it‟s high cost solution compared to wire-bonding • Key bottleneck: Thin wafer/die handling (50um), Drilling/Filling/Align Via Machine Filling Thinning Bonding 20 ㎛ CD 30um AR : 2 30um 50um TSV technology is promising for future DRAM‟s capacity and performance increase But, the issue of increased cost should be addressed 21
Consideration of New Memory Hierarchy Will the memory hierarchy still be the same? Current Outstanding issues & Future outlook Challenges CPU CPU Cache Memory L4$? Large Cache or Multi layer Memory Main memory Main Memory (DRAM) Emerging NVM memory NVM? Storage Storage Collaboration within End-User/Platform/CPU/Memory is Essential ! 22
New Memory Cell structures are in development Non-Volatile Memory Volatile Memory Resistance Change Charge Trap DRAM SRAM RRAM STT- PRAM MRAM NAND NOR 1 2 I 1‟ “ 0” 2‟ ” V1 V0 Charge-Based Device oxide Phase- Interface Magneto- dependent or bulk Resistance Resistance Resistance changes changes changes Resistance-Based Device Resistance change memory cells are good candidates due to DRAM compatible cell size, latency, & power On active research for these to find new memory solution 23
Samsung Memory 1 HPC: Spearhead of Computing 2 Today & Tomorrow 3 The Day After Tomorrow 4 Summary 5
Call for Action HPC is vision of future Server/PC, so Close Collaboration among End User/System/Platform level is highly important Memory in HPC has developed in evolutionary steps. • DDR1 DDR2 DDR3 … However, future of HPC Memory will face new challenges • Whole memory hierarchy including storage may need to change • Samsung invites to a dialogue and active collaboration to jointly create the next evolutionary steps and prepare for a possible paradigm shift 25
1 Samsung = sustainable leading edge technology Today‟s excellence in mass production: 2 30nm class, DDR3, 32GB based on 4Gb Tomorrow‟s cutting edge: 3 20nm, DDR4, DDR5 … and TSV The day after tomorrow: 4 „Giga -investments “ + disruptive system memory technology The future is not to be predicted. Let‟s create it together! 5
You can plant SAMSUNG Green Memory on your solution 27
Recommend
More recommend
