Тёмный
CXL Consortium
CXL Consortium
CXL Consortium
Подписаться
Compute Express Link (CXL) is a new breakthrough high-speed CPU interconnect that enables a high-speed, efficient performance between the CPU and platform enhancements and workload accelerators.

Industry leaders Alibaba, Cisco, Dell EMC, Facebook, Google, Hewlett Packard Enterprise, Huawei, Intel Corporation and Microsoft have teamed up to form an open industry standard group to develop technical specifications that facilitate breakthrough performance for emerging usage models while supporting an open ecosystem for data center accelerators and other high-speed enhancements.
Introducing the CXL 3.1 Specification
54:10
7 месяцев назад
Introduction to CXL
3:14:23
Год назад
CXL Type 3 Memory Device Demo - Meta
4:22
2 года назад
Комментарии
@JulieLowman-j8p
@JulieLowman-j8p День назад
88145 Rutherford Meadows
@OccamHelen
@OccamHelen День назад
64319 Bobby Fall
@TimothyDuke-w2h
@TimothyDuke-w2h 4 дня назад
442 Cole Underpass
@HarringtonBartholomew-u9d
@HarringtonBartholomew-u9d 4 дня назад
Williams William Harris Patricia Hernandez Kevin
@KatePetiet
@KatePetiet 6 дней назад
5612 Audrey Knoll
@DouglassFrederic
@DouglassFrederic 12 дней назад
0925 Ratke Parkways
@LenSuazo-r6m
@LenSuazo-r6m 13 дней назад
Marc Lights
@DorisChristopher-t9o
@DorisChristopher-t9o 15 дней назад
Leilani Squares
@VilmaRichoux-k2w
@VilmaRichoux-k2w 16 дней назад
Hermiston Island
@CaseyCain-r7s
@CaseyCain-r7s 17 дней назад
Garcia Kimberly White Larry Martinez Helen
@EdithLakey-k5m
@EdithLakey-k5m 18 дней назад
Cummings Parkways
@DonaldMoore-i5p
@DonaldMoore-i5p 19 дней назад
54731 Kshlerin Shores
@HillRachel-w4q
@HillRachel-w4q 20 дней назад
034 Ricky Point
@CaseyCain-r7s
@CaseyCain-r7s 23 дня назад
Davis Brian Lewis Christopher Harris Eric
@HeadingWatch
@HeadingWatch 24 дня назад
Anderson Deborah Hall Jennifer Rodriguez Edward
@PatrickChilde-y1w
@PatrickChilde-y1w 24 дня назад
Norene Plains
@JosePerez-w1m
@JosePerez-w1m 25 дней назад
Wanda Plains
@BetsyGodfery-l3n
@BetsyGodfery-l3n 27 дней назад
Leannon Harbors
@LonnieThomas-v4s
@LonnieThomas-v4s Месяц назад
Jones George Martin Helen Thompson Nancy
@Dollar_CoinYT
@Dollar_CoinYT Месяц назад
Wow... Spelled FPGAs wrong im not gonna trust this
@ZezhouWang
@ZezhouWang 6 месяцев назад
So many details in the slides. It's easy to get lost
@AndyAvera
@AndyAvera 7 месяцев назад
I'd love the presentation material as well. Great session!
@jleehere
@jleehere 7 месяцев назад
Is there any chance for me to get the presentation material? :)
@RTL_Design_Enthusiast
@RTL_Design_Enthusiast 8 месяцев назад
How can be the credits exchange mechanism work without LLCRD flit in CXL 3.0? I mean which flit should carry credit information after INIT.Pram Flit being sent.
@solidreactor
@solidreactor 9 месяцев назад
Will these Micron memory expansion cards be available for Workstations platforms as well, like the Threadripper and Threadripper Pro? Or will there only be support for the Epic 9004 series? If there will be workstation support, are there plans for these cards being available as PCIe5 x16 slot variant add-in boards? With Intel Optane gone, I really hope CXL and Micron will include support for Workstation users as well.
@AbhishekRamesh1
@AbhishekRamesh1 9 месяцев назад
I think it'll be a while before we see CXL support for HEDT's.
@rondafasano8226
@rondafasano8226 Год назад
Promo-SM
@sergioacuna1772
@sergioacuna1772 Год назад
and the gpu processor, can control a cxl port, how caché memory?
@j_m_b_1914
@j_m_b_1914 Год назад
I'm trying to learn what problem CXL is designed to solve? How is this any different than standard system memory?
@roflmagister5
@roflmagister5 Год назад
why blur the text, it's pretty much obvious this is a 96-thread × 2 cpu system.
@ronaldwoofer5024
@ronaldwoofer5024 Год назад
how do i install cxl to help reduce Texture Memory usage in RAM ?
@asdfghjkznsnsj
@asdfghjkznsnsj Год назад
Why do you think the CXL devices' cache size should be around 1MB? I'm curious about this because there are many GPUs that have a cache larger than 1MB
@varaprasadch4340
@varaprasadch4340 Год назад
Can I have a pdf of this video please since your explanation is very I want to go through this
@Mr_ST_720
@Mr_ST_720 Год назад
Will it go point to point over chassis
@etp4379
@etp4379 2 года назад
I wonder if CXL 2.0 would be able to be used to create a virtual machine that runs a single kernel that runs on all hardware as though it were one giant system. Not that I can think of a use case for this, it is just fascinating.
@_____alyptic
@_____alyptic 2 года назад
I wonder when the CXL 3.0 Drafts would be allowed to be previewed by the public :/
@juanpabloochoa9418
@juanpabloochoa9418 11 месяцев назад
Last month they released the first public version for 3.0 and efforts are now on 3.1
@springer9406
@springer9406 2 года назад
The graphic in the beginning misspells "FPGA".
@yizhoushan6333
@yizhoushan6333 2 года назад
Very informative!
@sjlee944
@sjlee944 2 года назад
Thanks for the great video. I have a few questions about the video. In the case of Multiple Logical Device (MLD), one CXL Memory Node is likely to consist of 1 ~ 4 channels. Is the host assigned to each channel? Or, if this memory node is E3.S form factor, there will be 20 or 40 DRAM packages, are the hosts allocated in units of packages?
@aurodeepta
@aurodeepta 2 года назад
Does a CXL Device support hotplug?
@luisbertranalvarez6769
@luisbertranalvarez6769 2 года назад
32:51
@alannair44
@alannair44 Год назад
29:50 The hosts and devices are attached and removed from the CXL switch via hotplug. More detail is given from 32:50 onward
@zhengxudong7294
@zhengxudong7294 2 года назад
How about the memory latency behind the CXL?
@alannair44
@alannair44 Год назад
= Latency of the attached memory device + latency of the PCIe interconnect.
@IvanHunglin
@IvanHunglin 2 года назад
It seems the fabric manager manages system memory for all those hosts. In this case, the fabric manager has higher exception level than hypervisors. What exactly is this piece of SW?
@alannair44
@alannair44 Год назад
It will be either be either a kernel module in the Host OS of a host (exception level = 0) like device drivers, or (for some CXL-compliant devices) implemented in hardware itself. The spec does not specify this, so it is left to the implementer. This is what I think, pls refer to spec to confirm.
@miltonmoss3069
@miltonmoss3069 2 года назад
Buy NLST = NLST = CXL = Winner
@anup9539
@anup9539 2 года назад
Which version of ubuntu is this which understands CXL ?
@rupeshkumar3949
@rupeshkumar3949 2 года назад
Very informative
@IvanHunglin
@IvanHunglin 2 года назад
Thanks a lot for the presentation. It clarifies many of my questions.
@yuriyd6839
@yuriyd6839 2 года назад
main question - what is the speed?
@大海田
@大海田 2 года назад
Excellent presentation, Rick
@antongavriliuk6548
@antongavriliuk6548 2 года назад
So what is the CXL attached DRAM idle latencies ? Please run Intel Memory Latency Checker 3.9a