Workload Description: Crysis Remastered Island integrated benchmark mode running at 1080p High Settings

Detailed Configurations:

Processor: i9 Intel® Core ™ i9-11980HK Processor; tested on a MSI GP76 Leopard 11UH; memory: 2x8GB DDR4 3200 MHz; Storage: Samsung MZVL21THCLR-00B00 (2x1TB); Display Resolution: 1920x1080; OS: Microsoft Windows 10 Professional version: 20H2 19042.928; Graphics card: RTX 3080; Graphics driver: 462.21; GPU Switch: Hybrid Graphics; Crysis Remastered (Island Benchmark) Version: 2.1.2; Game Resolution/Setting: 1920x1080/High

Processor: AMD Ryzen 9 5900HX Processor; tested on an Asus ROG Strix (G533QS); memory: 32GB 2x16 3200 MHz; Storage: HFM001TD3JX013N (976,762 MBytes (1024 GB)); Display Resolution: 2560 x 1440; OS: Microsoft Windows 10 Professional version: 20H2 19042.928; Graphics card: RTX 3080; Graphics driver: 466.11; GPU Switch: Hybrid Graphics; Crysis Remastered (Island Benchmark) Version: 2.1.2; Game Resolution/Setting: 1920x1080/High

Workload Description: Transcoding of 4:2:2 10-bit 4K content to 1080p 10-bit 4:2:0 using DaVinci Resolve

Detailed Configurations:

Processor: i9 Intel® Core ™ i9-11980HK Processor; tested on a MSI GP76 Leopard 11UH; memory: 2x8GB DDR4 3200 MHz; Storage: Samsung MZVL21THCLR-00B00 (2x1TB); Display Resolution: 1920x1080; OS: Microsoft Windows 10 Professional version: 20H2 19042.928; Graphics card: RTX 3080; Graphics driver: 462.21; GPU Switch: Hybrid Graphics; DaVinci Resolve Version: 17.2; Game Resolution/Setting: 1920x1080/High

Processor: AMD Ryzen 9 5900HX Processor; tested on an Asus ROG Strix (G533QS); memory: 32GB 2x16 3200 MHz; Storage: HFM001TD3JX013N (976,762 MBytes (1024 GB)); Display Resolution: 2560 x 1440; OS: Microsoft Windows 10 Professional version: 20H2 19042.928; Graphics card: RTX 3080; Graphics driver: 466.11; GPU Switch: Hybrid Graphics; DaVinci Resolve Version: 17.2; Game Resolution/Setting: 1920x1080/High

Workload Description: Manual gameplay of Valheim running at 1080p Medium settings on two systems utilizing the same world seed

Detailed Configurations: Processor: i7 Intel® Core ™ i7-1195G7 Processor; tested on an Intel Reference Platform; memory: 16GB (2x8GB) LPDDR4X-4266; Storage: Samsung SSD 980 PRO 1TB; OS: Microsoft Windows 10 Professional version: 20H2 19042.928; Graphics card: Intel(R) Iris(R) Xe Graphics; Graphics driver: 10.0.19042.928

Processor: AMD Ryzen 7 5800U; tested on a HP ProBook; memory: 16GB (2x8GB) DDR4-3200; Storage: Samsung MZVLQ512HALU-000H1 512GB; OS: Microsoft Windows 10 Home version: 10.0.19042.844; Graphics card: AMD Radeon Graphics; Graphics driver: 27.20.14022.3003;

3DMark Sampler Feedback Feature Test

Workload Description: Release Candidate version of 3DMark’s Sampler Feedback Feature test running with Sampler Feedback enabled on supported platforms, and disabled on all platforms

Detailed Configurations:

Processor: i7 Intel® Core ™ i7-1195G7 Processor; tested on an Intel Reference Platform; memory: 16GB (2x8GB) LPDDR4X-4266; Storage: Samsung SSD 980 PRO 1TB; OS: Microsoft Windows 10 Professional version: 20H2 19042.928; Graphics card: Intel(R) Iris(R) Xe Graphics; Graphics driver: 10.0.19042.928; 3DMark Sampler Feedback Feature test RC build

Processor: AMD Ryzen 7 5800U; tested on a HP ProBook; memory: 16GB (2x8GB) DDR4-3200; Storage: Samsung MZVLQ512HALU-000H1 512GB; OS: Microsoft Windows 10 Home version: 10.0.19042.844; Graphics card: AMD Radeon Graphics; Graphics driver: 27.20.14022.3003; 3DMark Sampler Feedback Feature test RC build

Workload Description: Running Nero Score benchmark application which analysis the contents of 50 photos using AI Inference

Detailed Configurations:

Processor: i5 Intel® Core ™ i5-1135G7 Processor; tested on a Microsoft Surface Pro 7+ with LTE Model 1961; memory: 8 GB LPDDR4X-4267; Storage: Samsung MZ9LQ256HBJQ-00000 256GB; Display Resolution: 2736x1824; OS: Microsoft Windows 10 Pro Insider Preview version: 21370.1; Graphics card: Intel® Iris® Xe Graphics; Graphics driver: 27.20.100.9030

Processor: SQ2 processor; tested on a Microsoft Surface Pro X H Model 1876, memory: 16 GB LPDDR4x-4267; Display Resolution: 2880x1920; OS: Microsoft Windows 10 Home Insider Preview version: 21370.1; Graphics card: Microsoft SQ2 Adreno 690, Graphics driver: 27.20.1460.0

Workload Description: Windows-native game Valheim running under virtualization via the Parallels application on MacOS.

Workload Description: Windows-native game Valheim running under virtualization via the Parallels application on MacOS.

Detailed Configurations:

Processor: Apple M1 Processor; tested on an Apple MacBook Pro 13 (Late 2020); memory: 16GB LPDDR4X-4266; Storage: Apple SSD AP0512Q; OS: Big Sur 11.2.3; Graphics card: Integrated; Display Resolution: 2560x1600

Workload Description: Transitioning from game play in Valorant to content creation in Adobe After Effects on a single 11^th Generation H-series based system

Detailed Configurations:

Processor: i7 Intel® Core ™ i7-11800H Processor; tested on a MSI GP76 Leopard 11UH; memory: 2x8GB DDR4 3200 MHz; Storage: 1TB Samsung Evo 980 Pro; Display Resolution: 1920x1080; OS: Microsoft Windows 10 Professional version: 20H2 19042.928; Graphics card: RTX 3080; Graphics driver: 462.21; GPU Switch: Hybrid Graphics;

11th Generation Intel® Core™ with Iris® Xe graphics delivers the world's best processor for gaming on a thin and light laptop

In thin & light Windows-based devices, based on unique features and performance testing (as of May 21, 2021) on industry benchmarks and Representative Usage Guides of Intel® Core™ i7-1195G7, including in comparison to AMD Ryzen 7 5800U. Visit 11th Gen Mobile

for details. Results may vary.

Thin and light gaming leadership: 11th Gen U-series offers casual gaming experiences that eclipse the competition.

Processor: Intel® Core™ i7-1195G7 processor PL1=28W with Intel® Dynamic Tuning Technology (Intel® DTT) enabled at 35W, 4C8T, Memory: LPDDR4-4267MHz, 16GB (2x8GB), dual channel and dual rank, Storage: Samsung MZVL21T0HCLR-00B00, Display Resolution: 1920x1080p, OS: Microsoft Windows 10 20H2-19042.928 Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: Intel® Xe Graphics, Graphics driver: 27.20.100.9219, Bios version: 309 measured on MSI Prestige 14 pre-production laptop.

Performance with Intel® DTT will vary based on chassis design choices, chassis temperature thresholds, cooling solutions, form factors (xyz dimensions), air flow, and ambient air temperatures.

Vs.

Processor: AMD Ryzen 7 5800U processor, 8C16T, Memory: 2x8GB DDR4-3200MHz, Storage: Samsung MZVLQ512HALU-000H1, Display Resolution: 1920x1080p, OS: Microsoft Windows 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: AMD Radeon(TM) Graphics, Graphics driver: 27.20.21002.112, Bios version: 01.02 measured on HP Probook 455 G8.

Testing done using the system's out-of-box performance profile, with Windows EPP slider set to "better performance". Observed 15W sustained power level with bursts at ~25W from AMD's μProf tool

Processor: Intel® Core™ i7-1195G7 processor PL1=28W, 4C8T, Memory: LPDDR4-4267MHz, 16GB (2x8GB), dual channel and dual rank, Storage: Samsung PM9A1 SSD, Display Resolution: 1920x1080p, OS: Microsoft Windows 10 20H2-19042.844 Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: Intel® Xe Graphics, Graphics driver: prod-hi-releases_​2021-01-26-ci-master-7075-revenue-pr-1009268-TGL, Bios version: ucode 16 measured on Intel reference board.

Vs.

Processor: AMD Ryzen 7 5800U processor, 8C16T, Memory: 2x8GB DDR4-3200MHz, Storage: Samsung MZVLQ512HALU-000H1, Display Resolution: 1920x1080p, OS: Microsoft Windows 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: AMD Radeon(TM) Graphics, Graphics driver: 27.20.21002.112, Bios version: 01.02 measured on HP Probook 455 G8.

Testing done using the system's out-of-box performance profile, with Windows EPP slider set to "better performance". Observed 15W sustained power level with bursts at ~25W from AMD's μProf tool

Testing as of May 21, 2021

The future of light content creation: 11^th Gen U-series provides the right compute for the right workload for modern content creation.

Processor: Intel® Core™ i7-1195G7 processor PL1=28W, 4C8T, Memory: LPDDR4-4267MHz, 16GB (2x8GB), dual channel and dual rank, Storage: Samsung PM9A1 SSD, Display Resolution: 1920x1080p, OS: Microsoft Windows 10 20H2-19042.844 Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: Intel® Xe Graphics, Graphics driver: prod-hi-releases_​2021-01-26-ci-master-7075-revenue-pr-1009268-TGL, Bios version: ucode 16 measured on Intel reference board.

Vs.

Processor: AMD Ryzen 7 5800U processor, 8C16T, Memory: 2x8GB DDR4-3200MHz, Storage: Samsung MZVLQ512HALU-000H1, Display Resolution: 1920x1080p, OS: Microsoft Windows 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: AMD Radeon(TM) Graphics, Graphics driver: 27.20.21002.112, Bios version: 01.02 measured on HP Probook 455 G8.

Processor: Intel® Core™ i7-1195G7 processor PL1=28W with Intel® Dynamic Tuning Technology (Intel® DTT) enabled at 35W, 4C8T, Memory: LPDDR4-4267MHz, 16GB (2x8GB), dual channel and dual rank, Storage: Samsung MZVL21T0HCLR-00B00, Display Resolution: 1920x1080p, OS: Microsoft Windows 10 20H2-19042.928 Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: Intel® Xe Graphics, Graphics driver: 27.20.100.9219, Bios version: 309 measured on MSI Prestige 14 pre-production laptop.

Vs.

Processor: AMD Ryzen 7 5800U processor, 8C16T, Memory: 2x8GB DDR4-3200MHz, Storage: Samsung MZVLQ512HALU-000H1, Display Resolution: 1920x1080p, OS: Microsoft Windows 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: AMD Radeon(TM) Graphics, Graphics driver: 27.20.21002.112, Bios version: 01.02 measured on HP Probook 455 G8.

Testing as of May 21, 2021

6GHz Wi-fi 6E: Consumer Wi-Fi's biggest advancement in 20 years

Nearly 3X faster (Gig+) Wi-Fi speeds

Based on IEEE 802.11ax specification. 160 MHz channels and Wi-Fi 6/6E technology advantages enable significantly higher maximum theoretical PC client speeds vs. standard Wi-Fi 5 products.

Wi-Fi 6 Wi-Fi 6E are based on IEEE 802.11ax and Wi-Fi 5 is based on the IEEE 802.11ac. Wi-Fi 6 and Wi-Fi 5 utilize legacy 5 GHz vs. only Wi-Fi 6E can utilize the new, larger 6GHz Wi-Fi spectrum. Wi-Fi 6E products operate in Wi-Fi 6 mode when used in legacy 5 GHz spectrum.

Intel® Wi-Fi 6/6E (Gig+) products support optional 160 MHz channels which enable the fastest possible maximum speed for typical 2-stream PC client Wi-Fi products. Intel® Wi-Fi 6/6E (Gig+) (802.11ax 2x2 160 MHz) products enable a 2402 Mbps maximum theoretical data rate, ~3X (2.8X) faster than the standard Wi-Fi 5 (802.11ac 2x2 80 MHz) products capable of a theoretical max data rate of 867 Mbps, as documented in IEEE 802.11 wireless standard specifications, and require the use of similarly configured 802.11ax wireless network routers.

April 2019 (Intel® Wi-Fi 6 (Gig+)) December 2020 (Intel® Wi-Fi 6E (Gig+))

World's Best Processor for Thin-and-Light Windows Laptops Just Got Better

In thin & light Windows-based devices, based on unique features and performance testing (as of May 21, 2021) on industry benchmarks and Representative Usage Guides across 5 key usages: productivity, creation, collaboration, gaming and entertainment of Intel® Core™ i7-1195G7, including in comparison to AMD Ryzen 7 5800U. Visit 11th Gen Mobile

for details. Results may vary.

Offer up to 25% overall application performance advantage vs. competition

Processor: Intel® Core™ i7-1195G7 processor at PL1= 28W, 4C8T, Memory: LPDDR4-4267MHz, 2x8GB, dual channel and dual rank, Storage: Samsung PM9A1 SSD, Display Resolution: 1920x1080p, OS: Microsoft Windows 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: Intel® Xe Graphics, Graphics driver: prod-hi-releases_​2021-01-26-ci-master- 7075-revenue-pr-1009268-TGL, Bios version: ucode 16 measured on Intel reference board.

Processor: AMD Ryzen 7 5800U processor, 8C16T, Memory: 2x8GB DDR4-3200MHz, Storage: Samsung 512GB MZVLQ512HALU-000H1, Display Resolution: 1920x1080p, OS: Microsoft Windows* 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: AMD Radeon(TM) Graphics, Graphics driver: 27.20.21002.112, Bios version: 01.02 measured on HP Probook 455 G8 laptop.

Testing as of May 21, 2021

Accelerate the number of designs with Intel® Wi-Fi 6/6E (Gig+)

Approximately 40 IA notebook, desktop, and motherboard designs in market now with more than 20 additional notebook models expected in time for holiday

Enable ISV optimizations to deliver up to 8x faster transcoding and up to double the video editing speed with AI acceleration over the competition

Processor: Intel® Core™ i7-1195G7 processor at PL1= 28W, 4C8T, Memory: LPDDR4-4267MHz, 2x8GB, dual channel and dual rank, Storage: Samsung PM9A1 SSD, Display Resolution: 1920x1080p, OS: Microsoft Windows 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: Intel® Xe Graphics, Graphics driver: prod-hi-releases_​2021-01-26-ci-master- 7075-revenue-pr-1009268-TGL, Bios version: ucode 16 measured on Intel reference board.

Processor: AMD Ryzen 7 5800U processor, 8C16T, Memory: 2x8GB DDR4-3200MHz, Storage: Samsung 512GB MZVLQ512HALU-000H1, Display Resolution: 1920x1080p, OS: Microsoft Windows* 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: AMD Radeon(TM) Graphics, Graphics driver: 27.20.21002.112, Bios version: 01.02 measured on HP Probook 455 G8 laptop.

Testing done using the system's out-of-box performance profile, with Windows EPP slider set to "better performance". Observed 15W sustained power level with bursts at ~25W from AMD's μProf tool

Testing as of May 21, 2021

Processor: Intel® Core™ i7-1195G7 processor PL1=28W with Intel® Dynamic Tuning Technology (Intel® DTT) enabled at 35W, 4C8T, Memory: LPDDR4-4267MHz, 16GB (2x8GB), dual channel and dual rank, Storage: Samsung MZVL21T0HCLR-00B00, Display Resolution: 1920x1080p, OS: Microsoft Windows 10 20H2-19042.928 Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: Intel® Xe Graphics, Graphics driver: 27.20.100.9219, Bios version: 309 measured on MSI Prestige 14 pre-production laptop.

Processor: AMD Ryzen 7 5800U processor, 8C16T, Memory: 2x8GB DDR4-3200MHz, Storage: Samsung 512GB MZVLQ512HALU-000H1, Display Resolution: 1920x1080p, OS: Microsoft Windows* 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: AMD Radeon(TM) Graphics, Graphics driver: 27.20.21002.112, Bios version: 01.02 measured on HP Probook 455 G8 laptop.

Testing done using the system's out-of-box performance profile, with Windows EPP slider set to "better performance". Observed 15W sustained power level with bursts at ~25W from AMD's μProf tool

Testing as of May 21, 2021

Show impressive 1080p gameplay on new and popular titles such as Valheim with up to 2.7x higher frame rate than the competition

Processor: Intel® Core™ i7-1195G7 processor PL1=, 4C8T, Memory: LPDDR4-4267MHz, 16GB (2x8GB), dual channel and dual rank, Storage: Samsung PM9A1 SSD, Display Resolution: 1920x1080p, OS: Microsoft Windows* 10 20H2-19042.844 Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: Intel® Xe Graphics, Graphics driver: prod-hi-releases_​2021-01-26-ci-master- 7075-revenue-pr-1009268-TGL, Bios version: ucode 16 measured on Intel reference board.

Processor: AMD Ryzen 7 5800U processor, 8C16T, Memory: 2x8GB DDR4-3200MHz, Storage: Samsung 512GB MZVLQ512HALU-000H1, Display Resolution: 1920x1080p, OS: Microsoft Windows* 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks. All benchmarks run in Admin mode. Graphics: AMD Radeon(TM) Graphics, Graphics driver: 27.20.14022.3003, Bios version: 01.01.02 measured on HP Probook 455 G8.

Testing as of May 21, 2021

Select new 11th Gen Intel Core designs will continue to accelerate this industry leadership with best-in-class Intel Wi-Fi 6E (Gig+) products which enable gigabit+ speeds, lower latencies, and enhanced reliability via unencumbered Wi-Fi performance in new 6 GHz bands (where available).

"Best-in-class Wi-Fi 6E" Based on IEEE 802.11ax specification. Wi-Fi 6/6E technology advantages related to network managed traffic enable lower latencies, more efficient operation, and higher reliability vs. random contention-based traffic of standard Wi-Fi 5 networks. Intel® Wi-Fi 6/6E (Gig+) support for optional 160 MHz channels enable faster speeds vs. standard Wi-Fi 6/6E products which only support required 80 MHz channels.

New 6 GHz wireless spectrum available to Intel® Wi-Fi 6E (Gig+) products will include many more 160 MHz channels which enable Gigabit Wi-Fi speeds

New, much larger, 6 GHz wireless spectrum is exclusively available to new Wi-Fi 6E devices. This provides protection from neighboring legacy 2.4/5 GHz Wi-Fi networks and more channel options if needed to avoid new neighboring 6 GHz networks.

6 GHz PC Wi-Fi 6E operation requires use of Intel® Wi-Fi 6E (Gig+) products in conjunction with operating systems and routers/APs/Gateways that support Wi-Fi 6E, together with regional spectrum allocation together with the required regulatory certifications.

SSDs with Intel® Optane™ memory are the fastest, enabling you to launch applications such as Adobe Photoshop up to 49% faster.

SSDs with Intel® Optane™ memory are the fastest:

As measured by Intel® Optane™ memory H20 with solid state storage prevails in the majority of tests against each NAND SSD tested. See 11th Gen Mobile for details.

* Testing by Intel as of 5/12/2021, in the following common laptop use cases on shipping configurations from major OEMs as compared to NAND SSDs:

(1) a collection of benchmarks and real-world workloads with drives at 50% prefill including PCMark 10 storage tests, CrystalDiskMark 7, email launch, PDF editor launch, video editor launch, and selected games.

(2) a series of Iometer tests at QD1 and QD2 with drives at 50% prefill. Tests were composed of various combinations of sequential and random read and write scenarios, designed to approximate real-world client PC usage based on Intel research, in both single and mixed modes with span size 8GB. System configuration: Intel® Core™ i7-1165G7 processor (4C8T, Turbo up to 4.70 GHz) on Intel reference platform. Graphics: Intel Iris Xe. Memory: 16GB (2x8GB) DDR4-3200. Storage: 32GB+500GB Intel® Optane™ memory H20 with solid state storage (with Intel® Rapid Storage Technology driver 18.1.3.1036) vs. Samsung PM9A1, Toshiba XG6, and Western Digital SN720 (all 500GB NAND with Microsoft Windows 10 in-box PCIe driver). OS: Microsoft Windows 10 (version 19042.789).

Result: Intel® Optane™ memory H20 with solid state storage prevails in the majority of tests against each NAND SSD.

Launch Adobe Photoshop up to 49% faster while multitasking on an 11th Gen Intel® Core™ processor with 512GB Intel® Optane™ memory H20 with solid state storage vs. a PCIe 3.0 TLC NAND SSD alone

As measured by Adobe Photoshop launch with local file copy workload on 11th Gen Intel® Core™ i7-1185G7 processor with 512GB Intel® Optane™ memory H20 with solid state storage vs. 512GB Western Digital SN750 (PCIe 3.0 TLC NAND SSD)

Processor: Intel® Core™ i7-1185G7 processor 28W PL1 with Intel® Dynamic Tuning Technology (Intel® DTT) enabled, 4C8T, Memory: LPDDR4-4267MHz, 16GB (2x8GB), Display Resolution: 1920x1080, OS: Microsoft Windows* 10 20H2-19042.572, Graphics: Intel® Xe Graphics, Graphics driver: 27.20.100.8431, BIOS version: TGLSFWI1.R00.3465.A01.2011140118 measured on Intel reference board.

Testing as of January 20, 2021

Intel® 5G Solution 5000 for PCs enables nearly 5X faster speeds vs. Intel 1Gbps LTE cellular based solutions.

Intel® (1Gbps) CAT 16 LTE cellular modem for PCs is based on the 3GPP Release 13 specification and supports 60-100 MHz wide aggregate bandwidth utilizing 3-5CA respectively, and utilizes 4X4/2x2 MIMO configurations to enable up to 1Gbps maximum theoretical download data rates.

The Intel® 5G NR modem for PCs is based on the Release 15 of the 3GPP standard, and supports 2x100 MHz channels, along with 4x4 MIMO, coupled with 256 QAM Modulation and can enable up to 4.7 Gbps theoretical download data rates which are nearly 5X faster (4.7X) than what is theoretically posible with Intel CAT 16 LTE products, as documented in 3GPP Rel 13 & Rel 15 standard specifications and requires the use of similarly configured 3GPP cellular networks.

The Intel® 5G Solution 5000 product for PCs can enable up to 4.7 Gbps theoretical download data rates which are nearly 5X faster (4.7X) than what is theoretically possible with the current Intel 1.0 Gbps LTE product, as documented in 3GPP Rel 13 & Rel 15 standard specifications and requires the use of similarly configured 3GPP cellular networks.

5G performance may vary and requires a carrier contract for 5G service, which is subject to regional availability and may not be available in all markets.

Test Date: 5/20/2021

3rd Gen Xeon Scalable processors deliver nearly 50 percent performance improvement, expanded memory bandwidth, faster I/O with more channels of PCI Express and advanced security capabilities with Intel Software Guard Extensions.

Visit 3rd Gen Xeon Scalable Processors

(Claim 125) for details. Results may vary.

3 weeks ago we launched our 11th Gen H-Series mobile processors - the world's best gaming laptop processors - offering desktop caliber game play from anywhere.

As measured by in-game benchmark mode performance (score or frames per second) where available, or frames per second where benchmark mode is unavailable, on the following 29 games: Ashes of the Singularity: Escalation,Assassin's Creed: Odyssey,Borderlands 3,Counter-Strike: Global Offensive, Crysis Remastered,DOTA 2,F1 2020,Far Cry New Dawn,Grand Theft Auto V,Gears 5,Grid 2019,Hitman 3,League of Legends,Metro Exodus,Player Unknown's Battlegrounds(PUBG),Red Dead Redemption 2,Serious Sam 4,Shadow of the Tomb Raider,Sid Meier's Civilization 6: Gathering Storm, Strange Brigade,Tom Clancy's Rainbow Six Siege,Tom Clancy's The Division 2,Total War: Three Kingdoms: Dynasty Mode, Troy: A Total War Saga: Siege,War Thunder,Watch Dogs: Legion,World of Tanks Encore RT, World of Warcraft: Shadowlands,World War Z. Results: The 11th Gen Intel® Core™ i9-11980HK-based system scored higher on the majority of the 29 game titles tested.

Processor: 11th Gen Intel® Core™ i9-11980HK processor, 8C16T; tested on a pre- production OEM design, memory: 2X 16 GB DDR4- 3200MHz; Storage: Samsung 970 Evo Plus 1TB ; Display Resolution: 1920x1080; OS: Microsoft Windows 10 Pro 20H2 19042.867; Graphics card: NVIDIA RTX 3080 with 16 GB GDDR6 RAM, Graphics driver: 27.21.14.6221; Max GFX power = 155W, BIOS version 30D. Power Plan = Balanced, AC power slider setting = Better performance; Game testing settings: performance optimization SW: Used highest performance setting (Performance Level = High), Fan Speed = Cooler Boost, GPU Switch = Discrete Graphics

Processor: 10th Gen Intel® Core™ i9-10980HK processor, 8C16T; tested on MSI GE76 laptop, memory: 2X 16 GB DDR4- 3200MHz; Storage: Samsung 970 Evo Plus 1TB ; Display Resolution: 1920x1080; OS: Microsoft Windows 10 Home 20H2 19042.867; Graphics card: NVIDIA RTX 3080 with 16 GB GDDR6 RAM, Graphics driver: 27.21.14.6172; Max GFX power = 155W ,BIOS version 109. Power Plan = Balanced, AC power slider setting = Better performance; Game testing settings: MSI Dragon Center: Used highest performance setting (Scenario = Extreme Performance) Gaming Mode = On, GPU Switch = Discrete Graphics Mode

Processor: AMD Ryzen 9 5900HX processor, 8C16T; tested in Lenovo Legion R9000K laptop, memory: 2X 16 GB DDR4- 3200MHz; Storage: Western Digital WDC PC SN730 SDBPNTY-1T00-1101 1TB ; Display Resolution: 1920x1080; OS: Microsoft Windows 10 Home 20H2 19042.746; Graphics card: NVIDIA RTX 3080 with 16 GB GDDR6 RAM, Graphics driver: 27.21.14.6156;Max GFX power = 165W, BIOS version GKCN34WW. Power Plan = Balanced, AC power slider setting = Better performance; Game testing settings: Lenovo PC Management: Used highest performance setting (Performance Mode) GPU Switch = Discrete Graphics Mode

Testing as of May 3, 2021

Chuck Duvall: Gaming Demo powered by 11^th Gen Intel Core H-series

Head-to-Head Gaming demo featuring Intel's first H-series product on 10nm process technology and H-series first new microarchitecture since Skylake. The new H-Series delivers an amazing leap forward for Enthusiast, Mobile Workstation and Commercial laptops.

More than 20% win vs competition

Crysis Remastered | Gameplay Settings - 1080p High

Tested as of 05/24/2021

Platform: MSI GP76 Leopard 11UH

Processor: Intel® Core™ i9-11980HK Processor (8C/16T, 45W TDP)

Memory: 2x8GB DDR4 3200 MHz SDRAM

Storage: Samsung MZVL21THCLR-00B00 (2x1TB)

Display Resolution: 1920 x 1080 display

OS: Microsoft Windows 10 Pro, Build: 19042.985

Graphics: RTX 3080

Platform: Asus ROG Strix (G533QS)

Processor: Ryzen 9 5900HX (8C/16T, 45W TDP)

Memory: 2x16GB DDR4 3200 MHz SDRAM

Storage: HFM001TD3JX013N (1024 GB)

Display Resolution: 2560 x 1440 display

OS: Microsoft Windows 10 Pro, Build 19042.928

Graphics card: RTX 3080

Maintaining a 25% overall application performance advantage vs. comp.

Processor: Intel® Core™ i7-1195G7 processor at PL1= 28W, 4C8T, Memory: LPDDR4-4267MHz, 2x8GB, dual channel and dual rank, Storage: Samsung PM9A1 SSD, Display Resolution: 1920x1080p, OS: Microsoft Windows 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: Intel® Xe Graphics, Graphics driver: prod-hi-releases_​2021-01-26-ci-master- 7075-revenue-pr-1009268-TGL, Bios version: ucode 16 measured on Intel reference board.

Processor: AMD Ryzen 7 5800U processor, 8C16T, Memory: 2x8GB DDR4-3200MHz, Storage: Samsung 512GB MZVLQ512HALU-000H1, Display Resolution: 1920x1080p, OS: Microsoft Windows* 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: AMD Radeon(TM) Graphics, Graphics driver: 27.20.21002.112, Bios version: 01.02 measured on HP Probook 455 G8 laptop.

Testing as of May 21, 2021

We are unleashing ISV optimizations to deliver up to 8x faster transcoding (than comp)

As measured by HandBrake Workload on

Processor: Intel® Core™ i7-1195G7 processor at PL1= 28W, 4C8T, Memory: LPDDR4-4267MHz, 2x8GB, dual channel and dual rank, Storage: Samsung PM9A1 SSD, Display Resolution: 1920x1080p, OS: Microsoft Windows 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: Intel® Xe Graphics, Graphics driver: prod-hi-releases_​2021-01-26-ci-master- 7075-revenue-pr-1009268-TGL, Bios version: ucode 16 measured on Intel reference board.

Processor: AMD Ryzen 7 5800U processor, 8C16T, Memory: 2x8GB DDR4-3200MHz, Storage: Samsung 512GB MZVLQ512HALU-000H1, Display Resolution: 1920x1080p, OS: Microsoft Windows* 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: AMD Radeon(TM) Graphics, Graphics driver: 27.20.21002.112, Bios version: 01.02 measured on HP Probook 455 G8 laptop.

Testing done using the system's out-of-box performance profile, with Windows EPP slider set to "better performance". Observed 15W sustained power level with bursts at ~25W from AMD's μProf tool

Testing as of May 21, 2021

Up to double the video editing speed with our AI acceleration over the competition.

As measured by Adobe Premiere Pro Scene Edit Detect Workload

Processor: Intel® Core™ i7-1195G7 processor PL1=28W with Intel® Dynamic Tuning Technology (Intel® DTT) enabled at 35W, 4C8T, Memory: LPDDR4-4267MHz, 16GB (2x8GB), dual channel and dual rank, Storage: Samsung MZVL21T0HCLR-00B00, Display Resolution: 1920x1080p, OS: Microsoft Windows 10 20H2-19042.928 Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: Intel® Xe Graphics, Graphics driver: 27.20.100.9219, Bios version: 309 measured on MSI Prestige 14 pre-production laptop.

Processor: AMD Ryzen 7 5800U processor, 8C16T, Memory: 2x8GB DDR4-3200MHz, Storage: Samsung 512GB MZVLQ512HALU-000H1, Display Resolution: 1920x1080p, OS: Microsoft Windows* 10 20H2-19042.844. Power policy set to AC/Balanced mode for all benchmarks except SYSmark 25 which is measured in AC/BAPCo mode for Performance. All benchmarks run in Admin mode. Graphics: AMD Radeon(TM) Graphics, Graphics driver: 27.20.14022.3003, Bios version: 01.01.02 measured on HP Probook 455 G8.

Testing as of May 21, 2021

Intel has the world's best CPUs.

In thin & light Windows-based devices, based on unique features and performance testing (as of May 21, 2021) on industry benchmarks and Representative Usage Guides across 5 key usages: productivity, creation, collaboration, gaming and entertainment of Intel® Core™ i7-1195G7, including in comparison to AMD Ryzen 7 5800U. Visit https://www.intel.com

for details. Results may vary.

Chuck Duvall: Evo Content Creation Demo powered by 11^th Gen Intel Core U-series

Adobe Premiere Pro: Multi-Camera Playback - 6 4K30 videos

AMD: Dropped Frames = 99%

Intel: Dropped Frames = 0.1%

Adobe Premier Pro | Version 15.20 (Build 35)

RUG1336-1.03 - Adobe Premier Pro Multicam Playback Using 6-4K

Tested as of 05/13/2021

Platform: MSI Prestige (Pre-Production)

Processor: Intel® Core™ i7-1195G7 Processor (4C/8T), PL1=28W with Intel® Dynamic Tuning Technology enabled at 35W

Memory: 16GB LPDDR4-4267 Mhz SDRAM

Storage: SAMSUNG MZVL21T0HCLR-00B00

Display Resolution: 1920 x 1080 display

OS: Microsoft Windows 10 Pro 10.0.19042.928

Graphics: Intel® Iris® Xe Graphics

Platform: HP ProBook 455 G8

Processor: AMD Ryzen 7 5800U with Radeon Graphics (8C/16T), Observed 15W sustained power level with bursts at ~25W from AMD's μProf tool

Memory: 32GB DDR4-3200 DDR4 SDRAM

Storage: MZVL21T0HCLR-00B00

Display Resolution: 1920 x 1080 display

OS: Microsoft Windows 10 Home 10.0.19042.844

Graphics card: AMD Radeon™ Graphics

Benchmarks:

• CrossMark is a benchmark from the BAPCo consortium that measures the performance across Windows, iOS or macOS platforms. CrossMark uses a combination of open source and proprietary workloads to assess system performance scores in the areas of Productivity, Creativity and Responsiveness. is a benchmark from the BAPCo consortium that measures the performance across Windows, iOS or macOS platforms. CrossMark uses a combination of open source and proprietary workloads to assess system performance scores in the areas of Productivity, Creativity and Responsiveness.
• PCMark 10 Applications is published by UL/Futuremark, a for-profit company owned by Underwriters Laboratories (UL). The PCMark 10 Applications test runs pre-installed Microsoft Office and supports UWP-ARM.
• SYSmark 25 is a benchmark from the BAPCo consortium that measures the performance of Windows platforms. SYSmark 25 tests three usage scenarios: Productivity, Creativity and Responsiveness. SYSmark contains real applications from Independent Software Vendors such as Microsoft and Adobe

Workloads:

• · 3DMark Sampler Feedback Feature Test is a test from UL Benchmarks that measures gaming performance for PCs with DirectX 12 feature - sampler feedback.
• Adobe After Effects workload measures the time in seconds it takes Adobe After Effects to perform a “Content Aware Fill” on a source video.
• Adobe Photoshop Elements workload measures the time it takes Adobe Photoshop Elements to process the “Colorize Photo” feature on 7 photos.
• Adobe Photoshop Content-Aware Fill workload measures the time it takes Adobe Photoshop to render a preview of a Content-Aware-Fill effect.

• Adobe Photoshop Radial Blur Filter workload

  measures the time it takes Adobe Photoshop to render a blur effect on one layer of a 2-layer photoshop file.

• Adobe Premiere Pro Multi-cam playback (Version: Adobe Premiere Pro - Version 15.2.0 (Build 15)) measures the number of frames dropped per second in Premiere Pro when a timeline is played back using the Multi-Cam feature. The video source has a multi camera source sequence that consists of six 30-second 4K clips. 5 clips are from a GoPro Hero8 Black camera: 4K29.97fps AVC @60Mbps; 1 clip is from a Mavic Air 2 drone: 4K29.97fps AVC @ 103Mbps.
• Adobe Premiere Pro Scene Edit Detect (Version: Adobe Premiere Pro - Version 15.2.0 (Build 15)) measures the time it takes to perform multiple operations of a video editing scenario featuring 2 operations using Adobe Premiere Pro, Scene Edit Detect and HEVC export using hardware decoding. The video source is 4K, H.265, 24FPS, 10-bit, MP4 footage, recorded at a bitrate of approximately 98.7 Mbps.
• DaVinci Resolve workload (Version: DaVinci Resolve Studio 17.2 Build 11) measures the time it takes to transcode a 4K, 10-bit, 4:2:2 video to 1080p, 10-bit, 4:2:0 format. The source file is a 30 seconds 4K@60fps, 10-bit HEVC video taken by Canon EOS R5.
• Export Microsoft PowerPoint to PDF (Software: Office 365 MSO (Version 2014 - Build 13929.20372)) workload measures the time it takes Microsoft PowerPoint 365 to convert the workload file, a 3.75 MB, 200 slide, .pptx presentation, to a ~4.0 MB PDF file. The export is created with Office's built-in capability to create PDF files. It does not use any add-ins to create the PDF.
• HandBrake Workload (Version: 20210403145615-e58b78c6f-master) measures the average FPS encoding speed for HandBrake Nightly to transcode a 4K HEVC 10bit video to 1080p HEVC 10bit. The source file is a 4K 3840 x 2160 HEVC 10bit MP4 video file. The output file is a 10bit 1080p HEVC MKV video file. This workload uses hardware encoding for the encoding process if available.
• Handbrake 1.3.3 Transcoding Workload measures the time it takes HandBrake to transcode a 30 seconds video clip from 1080p [H.264, 26Mbps, 30FPS] to 720p [H.264, 5Mbps, 30FPS]. This workload can be used to test devices enabled with QSV, VCE, NVENC, or CPU only testing. Hardware encoding is used in this testing.
• MAGIX Vegas Pro (Version 18.0 Build 527) measures the time it takes MAGIX Vegas Pro to render a 4K H.264 clip with a "Style Transfer" effect applied. The render is made using hardware encoding if available. The video is a H.264 4K MP4 file recorded at a bit rate of 72.2 Mb/s, and ~60seconds in length.
• Microsoft Word document to PDF conversion (Software: Office 365 MSO (Version 2014 - Build 13929.20372)) workload measures the time it takes Microsoft Office to export a Word Document to PDF. o The Word source file is an 801-page, 6.49 MB document. The converted PDF file is approximately 3.7 MB.
• Nero Score is a CPU AI benchmark that measures CPU-based AI inference time (i.e. how many pictures are processed per second, based on a pre-trained model). The benchmark measures a processor's multi-core power, while pushing the graphics card to its maximum limit, with real world multimedia use cases.
• Topaz Labs Gigapixel AI Photo Upscale (Version: Topaz Labs Gigapixel AI 5.5.2) workload measures the time it takes to process a 4X photo enlargement on 7 images in a batch process. The source files are 7 jpeg files with a total file size of 1.88 MB. The files are a mix of dimensions, resolutions and subject matter.

1.5x higher AI performance with 3rd Gen Intel® Xeon® Scalable processor supporting Intel® DL Boost vs. FP32 AMD EPYC 7763 (64C Milan): (geomean of 20 workloads including logistic regression inference, logistic regression fit, ridge regression inference, ridge regression fit, linear regression inference, linear regression fit, elastic net inference, XGBoost Fit, XGBoost predict, SSD-ResNet34 inference, Resnet50-v1.5 inference, Resnet50-v1.5 training, BERT Large SQuaD inference, kmeans inference, kmeans fit, brute_​​knn inference, SVC inference, SVC fit, dbscan fit, traintestsplit) 3rd Gen Intel Xeon: 8380: 1-node, 2x Intel Xeon Platinum 8380 (40C/2.3GHz, 270W TDP) processor on Intel Software Development Platform with 512 GB (16 slots/ 32GB/ 3200) total DDR4 memory, ucode X55260, HT on, Turbo on, Ubuntu 20.04 LTS, 5.4.0-65-generic/5.4.0-64-generic, 1x Intel_​​SSDSC2KG96, Intel SSDPE2KX010T8, tested by Intel, and results as of March 2021.

AMD: 7763: 1-node, 2-socket AMD EPYC 7763 (64C/2.45GHz, 280W cTDP) on GIGABYTE R282-Z92 server with 512 GB (16 slots/ 32GB/ 3200) total DDR4 memory, ucode 0xa001114, SMT on, Boost on, Power deterministic mode, Ubuntu 20.04 LTS, 5.4.0-65-generic, 1x Samsung_​​MZ7LH3T8/INTEL SSDSC2KG019T8, tested by Intel, and results as of March 2021.

ResNet50-v1.5 Intel : gcc-9.3.0, oneDNN 1.6.4, BS=128, INT8, TensorFlow 2.4.1 with Intel optimizations for 3rd Gen Intel Xeon Scalable processor, upstreamed to TensorFlow- 2.5 (container- intel/intel-optimized-tensorflow:tf-r2.5-icx-b631821f), Model zoo: https://github.com/IntelAI/models/tree/icx-launch-public/quickstart, ResNet50-v1.5 AMD : gcc-9.3.0, oneDNN 1.6.4, BS=128, FP32, TensorFlow- 2.4.1, Model zoo: https://github.com/IntelAI/models/tree/icx-launch-public/benchmarks/image_recognition/tensorflow/resnet50v1_5

ResNet50-v1.5 Training Intel : gcc-9.3.0, oneDNN 1.6.4, BS=256, FP32, TensorFlow 2.4.1 with Intel optimizations for 3rd Gen Intel Xeon Scalable processor, upstreamed to TensorFlow- 2.5 (container- intel/intel-optimized-tensorflow:tf-r2.5-icx-b631821f), Model zoo: https://github.com/IntelAI/models/tree/icx-launch-public/quickstart, ResNet50-v1.5 Training AMD : gcc-9.3.0, oneDNN 1.6.4, BS=256, FP32, TensorFlow- 2.4.1, Model zoo: https://github.com/IntelAI/models/tree/icx-launch-public/benchmarks/image_recognition/tensorflow/resnet50v1_5 SSD-ResNet34 Intel : gcc-9.3.0, oneDNN 1.6.4, BS=1, INT8,

TensorFlow 2.4.1 with Intel optimizations for 3rd Gen Intel Xeon Scalable processor, upstreamed to TensorFlow- 2.5 (container- intel/intel-optimized-tensorflow:tf-r2.5-icx-b631821f), Model zoo: https://github.com/IntelAI/models/tree/icx-launch-public/quickstart/, AMD : SSD-ResNet34, gcc-9.3.0, oneDNN 1.6.4, BS=1, FP32, TensorFlow- 2.4, Model zoo: https://github.com/IntelAI/models/tree/icx-launch-public/benchmarks/object_detection/tensorflow/ssd-resnet34 BERT-Large SQuAD Intel : gcc-9.3.0, oneDNN 1.6.4, BS=1, INT8,

TensorFlow 2.4.1 with Intel optimizations for 3rd Gen Intel Xeon Scalable processor, upstreamed to TensorFlow- 2.5 (container- intel/intel-optimized-tensorflow:tf-r2.5-icx-b631821f), Model zoo: https://github.com/IntelAI/models/tree/icx-launch-public/quickstart/, AMD : BERT-Large SQuAD, gcc-9.3.0, oneDNN 1.6.4, BS=1, FP32, TensorFlow- 2.4.1, Model zoo: https://github.com/IntelAI/models/tree/icx-launch-public/benchmarks/language_modeling/tensorflow/bert_large Python : Python 3.7.9, SciKit-Learn : Sklearn 0.24.1, oneDAL : Daal4py 2021.2, XGBoost : XGBoost 1.3.3 : Benchmarks: https://github.com/IntelPython/scikit-learn_bench

.3x higher AI performance with 3rd Gen Intel® Xeon® Scalable processor supporting Intel® DL Boost vs. NVIDIA A100 GPU: (geomean of 20 workloads including logistic regression inference, logistic regression fit, ridge regression inference, ridge regression fit, linear regression inference, linear regression fit, elastic net inference, XGBoost Fit, XGBoost predict, SSD-ResNet34 inference, Resnet50-v1.5 inference, Resnet50-v1.5 training, BERT Large SQuaD inference, kmeans inference, kmeans fit, brute_​​knn inference, SVC inference, SVC fit, dbscan fit, traintestsplit) 3rd Gen Intel Xeon: 8380: 1-node, 2x Intel Xeon Platinum 8380 (40C/2.3GHz, 270W TDP) processor on Intel Software Development Platform with 512 GB (16 slots/ 32GB/ 3200) total DDR4 memory, ucode X55260, HT on, Turbo on, Ubuntu 20.04 LTS, 5.4.0-65-generic, 1x Intel_​​SSDSC2KG96, Intel SSDPE2KX010T8, tested by Intel, and results as of March 2021.

DL Measurements on A100: 1-node, 2-socket AMD EPYC 7742 (64C) with 256GB (8 slots/ 32GB/ 3200) total DDR4 memory, ucode 0x8301038, HT on, Turbo on, Ubuntu 20.04 LTS, 5.4.0-42-generic, INTEL SSDSC2KB01, NVIDIA A100-PCIe-40GB, HBM2-40GB, Accelerator per node =1, tested by Intel, and results as of March 2021.

ML Measurements on A100 : 1-node, 2-socket AMD EPYC 7742 (64C) with 512 GB (16 slots/ 32GB/ 3200) total DDR4 memory, ucode 0x8301034, HT on, Turbo on, Ubuntu 18.04.5 LTS, 5.4.0-42-generic,NVIDIA A100 (DGX-1), 1.92TB M.2 NVMe, 1.92TB M.2 NVMe RAID tested by Intel, and results as of March 2021.

ResNet50-v1.5 Intel : gcc-9.3.0, oneDNN 1.6.4, BS=1, INT8, TensorFlow 2.4.1 with Intel optimizations for 3rd Gen Intel Xeon Scalable processor, upstreamed to TensorFlow- 2.5 (container- intel/intel-optimized-tensorflow:tf-r2.5-icx-b631821f), Model zoo: https://github.com/IntelAI/models/tree/icx-launch-public/quickstart ResNet50-v1.5 NVIDIA :A100 (7 instance/GPU), BS=1,TensorFlow - 1.5.5 (NGC: tensorflow:21.02-tf1-py3), https://github.com/NVIDIA/DeepLearningExamples/tree/master/TensorFlow/Classification/ConvNets/resnet50v1.5, TF AMP (FP16+TF32);

ResNet50-v1.5 Training Intel : gcc-9.3.0, oneDNN 1.6.4, BS=256, FP32, TensorFlow 2.4.1 with Intel optimizations for 3rd Gen Intel Xeon Scalable processor, upstreamed to TensorFlow- 2.5 (container- intel/intel-optimized-tensorflow:tf-r2.5-icx-b631821f), Model zoo: https://github.com/IntelAI/models/tree/icx-launch-public/quickstart.

ResNet50-v1.5 Training NVIDIA :A100, BS=256,TensorFlow - 1.5.5 (NGC: tensorflow:21.02-tf1-py3),https://github.com/NVIDIA/DeepLearningExamples/tree/master/TensorFlow/Classification/ConvNets/resnet50v1.5, TF32; BERT-Large SQuAD Intel : gcc-9.3.0, oneDNN 1.6.4, BS=1, INT8,

TensorFlow 2.4.1 with Intel optimizations for 3rd Gen Intel Xeon Scalable processor, upstreamed to TensorFlow- 2.5 (container- intel/intel-optimized-tensorflow:tf-r2.5-icx-b631821f), Model zoo: https://github.com/IntelAI/models/tree/icx-launch-public/quickstart/ A100 : BERT-Large SQuAD, BS=1, A100 (7 instance/GPU), TensorFlow - 1.5.5 (NGC: tensorflow:20.11-tf1-py3), https://github.com/NVIDIA/DeepLearningExamples/tree/master/TensorFlow/LanguageModeling/BERT,TF AMP (FP16+TF32) ; SSD-ResNet34 Intel : gcc-9.3.0, oneDNN 1.6.4, BS=1, INT8,

TensorFlow 2.4.1 with Intel optimizations for 3rd Gen Intel Xeon Scalable processor, upstreamed to TensorFlow- 2.5 (container- intel/intel-optimized-tensorflow:tf-r2.5-icx-b631821f), Model zoo: https://github.com/IntelAI/models/tree/icx-launch-public/quickstart/, SSD-ResNet34 NVIDIA :A100 (7 instance/GPU), BS=1,Pytorch - 1.8.0a0 (NGC Container, latest supported): A100 : SSD-ResNet34 (NGC: pytorch:20.11-py3), https://github.com/NVIDIA/DeepLearningExamples/tree/master/PyTorch/Detection/SSD, AMP (FP16 +TF32) ; Python : Intel: Python 3.7.9, Scikit-Learn : Sklearn 0.24.1, OneDAL : Daal4py 2021.2, XGBoost: XGBoost 1.3.3 Python : NVIDIA A100 : Python 3.7.9, Scikit-Learn : Sklearn 0.24.1, CuML 0.17, XGBoost 1.3.0dev.rapidsai0.17, Nvidia RAPIDS : RAPIDS 0.17, CUDA Toolkit : CUDA 11.0.221 Benchmarks: https://github.com/IntelPython/scikit-learn_bench

1.50x higher performance on Monte Carlo FSI Kernel 3rd Gen Intel Xeon:8380: 1-node, 2x 3rd Gen Intel Xeon Platinum 8380 (40C/2.3GHz, 270W TDP) processor on Intel Software Development Platform with 256 GB (16 slots/ 16GB/ 3200) total DDR4 memory, ucode 0x261, HT on, Turbo on, CentOS Linux 8.3.2011, 4.18.0-240.1.1.el8_​​3.crt1.x86_​​64, 1x Intel_​​SSDSC2KG96, App Version: v1.1; Build notes: Tools: Intel MKL 2020u4, Intel C Compiler 2020u4, Intel Threading Building Blocks 2020u4; threads/core: 1; Turbo: used; Build knobs: -O3 -xCORE-AVX512 -qopt-zmm-usage=high -fimf-precision=low -fimf-domain-exclusion=31 -no-prec-div -no-prec-sqrt

AMD: 7763: 1-node, 2-socket AMD EPYC 7763 (64C/2.45GHz, 280W cTDP) on GIGABYTE R282-Z92 server with 512 GB (16 slots/ 32GB/3200) total DDR4 memory, ucode 0xa001114, SMT on, Boost on, Power deterministic mode, NPS=4, CentOS Linux 8.3.2011, 4.18.0-240.1.1.el8_​​3.crt1.x86_​​64, 1x Samsung_​​MZ7LH3T8, App Version: v1.1; Build notes: Tools: Intel MKL 2020u4, Intel C Compiler 2020u4, Intel Threading Building Blocks 2020u4; threads/core: 2; Turbo: used; Build knobs: -O3 -march=core-avx2 -fimf-precision=low -fimf-domain-exclusion=31 -no-prec-div -no-prec-sqrt Tested by Intel and results as of March 2021

1.32x higher performance on RELION Plasmodium Ribosome 3rd Gen Intel Xeon: 8380: 1-node, 2x 3rd Gen Intel Xeon Platinum 8380 (40C/2.3GHz, 270W TDP) processor on Intel Software Development Platform with 256 GB (16 slots/ 16GB/ 3200) total DDR4 memory, ucode 0x261, HT on, Turbo on, CentOS Linux 8.3.2011, 4.18.0-240.1.1.el8_​​3.crt1.x86_​​64, 1x Intel_​​SSDSC2KG96, App Version: 3_​​1_​​1; Build notes: Tools: Intel C Compiler 2020u4, Intel MPI 2019u9; threads/core: 2; Turbo: used; Build knobs: -O3 -ip -g -debug inline-debug-info -xCOMMON-AVX512 -qopt-report=5 -restrict

AMD: 7763: 1-node, 2-socket AMD EPYC 7763 (64C/2.45GHz, 280W cTDP) on GIGABYTE R282-Z92 server with 512 GB (16 slots/ 32GB/3200) total DDR4 memory, ucode 0xa001114, SMT on, Boost on, Power deterministic mode, NPS=4, CentOS Linux 8.3.2011, 4.18.0-240.1.1.el8_​​3.crt1.x86_​​64, 1x Samsung_​​MZ7LH3T8, App Version: 3_​​1_​​1; Build notes: Tools: Intel C Compiler 2020u4, Intel MPI 2019u9; threads/core: 2; Turbo: used; Build knobs: -O3 -ip -g -debug inline-debug-info -march=core-avx2 -qopt-report=5 -restrict Tested by Intel and results as of March 2021

1.27x higher performance on NAMD STMV 1.27x higher performance on NAMD (geomean of Apoa1, STMV) 3rd Gen Intel Xeon: 8380: 1-node, 2x 3rd Gen Intel Xeon Platinum 8380 (40C/2.3GHz, 270W TDP) processor on Intel Software Development Platform with 256 GB (16 slots/ 16GB/ 3200) total DDR4 memory, ucode 0x261, HT on, Turbo on, CentOS Linux 8.3.2011, 4.18.0-240.1.1.el8_​​3.crt1.x86_​​64, 1x Intel_​​SSDSC2KG96, App Version: 2.15-Alpha1 (includes AVX tiles algorithm); Build notes: Tools: Intel MKL, Intel C Compiler 2020u4, Intel MPI 2019u8, Intel Threading Building Blocks 2020u4; threads/core: 2; Turbo: used; Build knobs: -ip -fp-model fast=2 -no-prec-div -qoverride-limits -qopenmp-simd -O3 -xCORE-AVX512 -qopt-zmm-usage=high AMD: 7763: 1-node, 2x AMD EPYC 7763 (64C/2.45GHz, 280W cTDP) on GIGABYTE R282-Z92 server with 512 GB (16 slots/ 32GB/3200) total DDR4 memory, ucode 0xa001114, SMT on, Boost on, Power deterministic mode, NPS=4, CentOS Linux 8.3.2011, 4.18.0-240.1.1.el8_​​3.crt1.x86_​​64, 1x Samsung_​​MZ7LH3T8, App Version: 2.15-Alpha1 (includes AVX tiles algorithm); Build notes: Tools: Intel MKL, AOCC 2.2.0, gcc 9.3.0, Intel MPI 2019u8; threads/core: 2; Turbo: used; Build knobs: -O3 -fomit-frame-pointer -march=znver1 -ffast-math Tested by Intel and results as of March 2021

New: 8380: 1-node, 2x Intel Xeon Platinum 8380 (40C/2.3GHz, 270W TDP) processor on Intel Software Development Platform with 256 GB (16 slots/ 16GB/ 3200) total DDR4 memory, ucode 0x261, HT on, Turbo on, CentOS Linux 8.3.2011, 4.18.0-240.1.1.el8_​​3.crt1.x86_​​64, 1x Intel_​​SSDSC2KG96, Tested by Intel between March 12, 2021 and March 29, 2021.

Baseline: 8280: 1-node, 2x Intel Xeon Platinum 8280 (28C/2.7GHz, 205W TDP) processor on Intel Software Development Platform with 192GB (12 slots/ 16GB/ 2933) total DDR4 memory, ucode 0x4002f01, HT on, Turbo on, CentOS Linux 8.3.2011, 4.18.0-240.1.1.el8_​​3.crt1.x86_​​64, 1x Intel_​​SSDSC2KG48 .Tested by Intel between February 1, 2021 to February 20, 2021

1.54x higher performance on NAMD STMV from using the AVX Tiles Algorithm vs. without AVX Tiles

2.43x higher performance on NAMD STMV with AVX Tiles Algorithm vs. prior gen without AVX Tile

1.57x higher performance on NAMD STMV without AVX Tiles Algorithm vs. prior gen

NAMD with Tiles: App Version: 2.15-Alpha1 (includes AVX tiles algorithm); Build notes: Tools: Intel MKL, Intel C Compiler 2020u4, Intel MPI 2019u8, Intel Threading Building Blocks 2020u4; threads/core: 2; Turbo: used; Build knobs: -ip -fp-model fast=2 -no-prec-div -qoverride-limits -qopenmp-simd -O3 -xCORE-AVX512 -qopt-zmm-usage=high

NAMD without Tiles: App Version: 2.15-Alpha1 (built without AVX tiles algorithm); Build notes: Tools: Intel MKL, Intel C Compiler 2020u4, Intel MPI 2019u8, Intel Threading Building Blocks 2020u4; threads/core: 2; Turbo: used; Build knobs: -ip -fp-model fast=2 -no-prec-div -qoverride-limits -qopenmp-simd -O3 -xCORE-AVX512 -qopt-zmm-usage=high -DNAMD_​​KNL tested by Intel and results as of March 2021

1. 43x higher NAMD STMV performance using the AVX Tiles Alogrithm on Platinum 8380 vs. prior gen without AVX Tiles