Bestand wählen
Merken

openPower - the current state of commercial openness in CPU development

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
and thank you
thank him if it was on stage is a wonderful person will talk to us about CPU development power of open source and how this all fits together and how this will inflate hopefully influential features CPU architecture and development so please leave a very warm round of applause so what you michel few Fourier thanks very new today uh and I and reading for my from German victims mixture of German and English so so that some because I'm coming from Mar 6 need actually living chemists still wanna chemists across gamut comic steps where robot knows about that at all and that's why decks and centers to mixture of sex and German but still want to show the there's still smart people living in Saxony and not just people into Tristan leaving some strange all time and I will live on text from I just want to give a brief induction today about all the power I got inspired I think to user goes by a talk by David Kaplan who was talking about CPU developments altogether 1 from scratch to final product and all the steps for its own sake but also but the touch that pp today but I want to go more into open power and the just tell you something about this because I'm not really sure if nobody knows about that so now Fundación and that might also be a little off topic and in this conference year but maybe still of interest so this is my agenda for the day so
1st of all I like to just talk a little bit of Bobby then as mentioned about deceive you don't info all and Y or power about results
on emissions of silicon but I'm my this is look at 0 this is the level of IBM Germany on I'm there in the year verification Department of CPU development so it's not just located there it's the worldwide this you proteome Büchel teams of arms and working to get people and then the law US fonts and and so on so what is the I got all the continents are included the the my main focus is verification I'm also that an expert of all power of but it is still involved in some public so it's so that I I think I can at least talk about
some steps here and this is all about mean this
goes into it you take the agenda and this is the year CPU so as we said looking at today at IBM and so we may see it's now a symbol of you you start at the left of by either coding or generating or some specifying some hot the descriptions of the maybe the still maybe the log of whatever you think you might work of this is also this the large arrow pointing to it so also I'm using usually for verification of the also multiple different versions of each field of that are possible but just assume it's busy 1 thing about that is the specifying the obvious how it should work and yeah there was some kind of twist stream that they can growth so all you different things so to define a product of the point right on so all of details so you get an estimate that list of all those logic opponents being generated so also or some other specific verification teams working on that 1 on then you can generate a T a schematic sort of it's not used imagine that it's a source and there on the you know forming those resistance and the layout and the the label at you busy gets to define a product DAC at the end of the separation process on I will also have some numbers later today but so assume this process so you can go through all of them on a daily basis but from our 1st line of these time to define a product that just assume intellectual 4 years until you have a real good product the but
know there's some nasty year of flow chart from some of my colleagues that our justice that is traits and hold the all flow is sum over k and what you just saw the diversion coming from my heart that the script language into the product of this CT broke flow of insect itself so you have this year some kind of product so you believe what you is so far it's all the previous versions yes and customer that's some holes has some much needed for certain things to run that of and to be the disk some kind of medication that you have some teams cervical high-level design and then this is busy you your input sources of sometimes very specific sometimes a very loose and they have different teams Buckingham Palace of trying to to get this but i put together was we have some kind of system components of so that that looking at the Baltimore did and it could come component let's have a team that looking at this big box so that the put it at the end of the variety Mustafa busy design teams of writing that the shall shellcode of our approach and then you have a few against verification team has this busy making sure on that this also brought against before you go to that and get it produced and then you have also other teams began looking at timing looking at circuit assigned integration so that's all going more at the physics of which is really important that opposes the side because you want to have the maximum performance coming out of the product and now the other busy you push the button get fabricated so that doesn't this going on and if that was well then you have a good machine on sometimes you find box and that's why this small red dollar sign is coming in and if you find box true basically the will define them as early as possible to the tournament and turn around genetic and it really works before you get to the fact around if you miss the point of view have to spend so much models and it and then you yeah I have to explain mother decided on that so I just show you don't want to be that person being responsible for having to be spin of food effect because that process going to that as not like 2 months two-and-a-half months and you you don't want to be the person who 1st of all spending a binding also causing the delay for the customers against and if you are really bad prose and then at the end you get into manufacturing all these machines of being put to the customer sites they push a button and the fingers and so this bundle but that's but I didn't see that in my life so far and this is
another picture some old trying to illustrate the same thing again but now we are trying to decide sides of the left more to design verification on the right side more to of development physical stuff that has been done to receive the resides in the basement faulting stuff that you of and I know it needs to be done but if the idea because we don't want to be that area is always critical and you will always be at the end of the process so everybody's waiting for you to push a button but that's why I chose the left side of the modification find also dealing very close 50 % teams um indices it's it's too so those of circling around because you all this looking from all on the same source at different times and that's been able to withdrew all the time but never going more into kind of continues immigration OK and are 2 numbers that I
promised and some will try to compare the year of our numbers from when building a dam larger ships of compared to process or of and it's quite interesting because of some or at the end of and comparing our U for both fixed you need that you didn't for dual system about 2 thousand people and also about the construction time as summer-long select years the ship like a 45 just for people of opposes own system of full 5 fuels for just 1 shot is basically a model of the terms of the process so into the machine is busier ready to ship to the customer the and also received yet because at the end of its of more than 1 billion dollars that is we spent for 1 generation of those systems and just to mention and there are 2 new architectures currently still being worked on and again the a architecture in DC architecture those numbers are for the process all of that you really can't compare those are pretty similar for me because part of your so that's
already it's from the subdued over the side and so that's jumping into a ball around and start looking at that so that that's fine but also a white overpower because so sometimes it looks strange but why should something change in that kind of popular stuff it's always be there was used you work on your PC's assume some all getting smaller you have your phones but some of a similar kind of this stuff in there but also from this of part of its own died some owing some in building in the center and it should just work on but as some common component to it so all we use this kind of process but very low along players like IBM on but but the all the time the market shifted drastically and so you have an was law which is provided we use very constant flow of performance improvements to didn't have to do much that you just had to move to the next generation of your technology and surprisingly you get 10 20 whatever % all of new performance of change much the same the less you add new features to its own to satisfy customer needs but that is more than what the number 2 anymore but also on the outside you have some other friends like you kind of look at coming out of just the boats of some kind of become mining which is so troubled trending today all these days but also kind of consumption models of those so farms are changing and in previous years it was like or somebody from some companies that are called the I need a new machines and the customer is in the there and now sold something which was staying there for the next couple of years of the states of clouds as something of very 1st agreed changing the operation model of all of this summer's day IBM and that also but he was something that the you felt because at the end so just this products were sold on the the cloud infrastructure was not up and running from open a few because we didn't get on that what other companies like Amazon and they're very fast and bring it up so as you can see today most of the cult infrastructures x 86 based and with I think it's hard to find the power law of syllables for the public open public clouds providers on the other side you have seen a also G. stuff take system changed a little so the room to more open and mature open providers and that's 1 part of it and on the other side you have some kind of of strategy changes that you don't need to know on but more to that later on but just back to adhere to Moore's law of on that chart is seen as they seek about performance but below that something of the customers look at home for the the look at how much might be dispensed for which kind of performance and that's the number they compare between 2 different providers and then only if you can match all be below year the other companies the new job are able to to sell stuff to them of and as I mentioned before that was looking quite well for the past couple of years but now busily Moore's Law slowing down and doesn't hurt anymore so what you now try to do is of you need to have some way enhanced performance by other means so what would it be that so you could block something to use systems we you can't move everything on the process a because you just can't fabricated so use you need to do something closer poses so you can block some excellent writers on that you can some all the storage systems to be fast or you can look at the Iowa if you can move to a new generation there or you can look at the memories of if they can provide you something but but it's ever get everything comes together and a data point is also there does that thing that the sum was proved to be the issue of next yourself because of the different friends so you have in the things providing I'm sure sometimes too much data you just some holy 2 and so on every device is some operating data at the end you need to have some that bone of looking at NBC some computes centers that can deal with all those data the and you see and what kind of data out of the year traders of those charts busy emissions were coming from different fields the but in coming model power of the and busier of there was or something I already talked about the availability to other things that force this kind of for the creation of this or the our community you the 2nd thing of most will of who lost is known to be a large pride of something like search or other stuff and and some 0 they will not really happy with the AIX 86 infrastructure of sort of some of the but torques so the the other companies and also if IBM and and there is here and some stuff on the year then poets machines and there the surprise of fall different patterns so that those pricey goods and I went outside they also of this even looking right in the eye of a monopoly of inter and that was a 2nd thing that Google was not being afraid I was afraid of and busy seeing that this was creating a very costly infrastructure obviously been bound to monopolistic on and then on the open to to any movement some else and that's my idea that he felt to to do something else and also of at the end of I and the Bird fetal but at to day want to build their own of computers and that's why they did this search for some popular in uh and again to came diagram and you have looked at us if you can do something together as well as the pointers and extend the reach of Paul architecture of which before and in the past develop all perceives pocket of lost some will be stronger than before the Powell architecture is the getting market shall gaining market share in the last couple of years but the overall the the Unix market is decreasing that fast at it's busy that really broken out well for keeping that architecture how just how is so you need to find other means to to a class strengthened the if a structure in the architecture of all of those boundaries as you how you can work with other parties to the getting got different markets and that was the of the root cause of they get this of Polynesia being done in 2015 but if you look at the the other part of this chart it back then that looked like all those companies some of losing market shell wall of try to come together and found something but busy looking at it said this year's dates and some old looks like a good move and let me show you a whole that dropped out but on the 1st steps that then around the busy done by the end of 1st of all this year investing a lot of money into the next of Linux being able to run and power but that was a complaint even done by engineers internally and all of the the last decades basically that's even all education spent at finding really good on all machines because we all don't all compute pool was made based on x 86 machines of because we go of selling them all before the that was a song complained but it didn't help but to complain internally about so to customer so this year showed us that the policies developed of able to beat the market so there was some kind of mind shift going on and this the result in investing money in the of structuring the codes run clinics and this even the little-endian big-endian complications to sort them out and get them resolved at the end the 2nd thing is this kind of Ohm's all source model for hardware and I think I will talk about that in a minute and basically it is the 1st step is to use a lot of from a coat of arms really make it available at what myself was even surprised so sometimes I just Googled of my name and some kind of wrote on and stuff surprisingly I the model of the 1st 5 matches was just some the file that would of difficulty going on so just try to the search for it so you should so find it easily on that now the this mission statements I think nothing
to really of we through completely a bunch of and the idea behind this is it's not just a new company and notify upon but so it's some kind of partnership was on some open community year or that to be installed to get people together that I want to improved upon architecture as well as to to look and the whole stack so that includes space here the basics of the the heart the development of well as you itself that it didn't go and they use the same tuning the same infrastructure using the boundaries on the sensor pixel to to illustrate that on the left side of the FTB the component of that busy consists basically of serviceable the process of itself and but still deposit doesn't help much you need to have some connection points and this year in the phases to accelerators and do not just those in the phases but have been rather open that other companies can also you we use them also about maps from over and that's what I learned also very early on in my work on that the customers really want to see a roadmap over the next couple of years so that they are theory can be to invest in your product and the architecture that this continues over the next to that noted that the motorcycles at of the disaster 2nd above all it's so to make sure that this is busier common and being used by many different parties on embedded busier forget the depot I at busier gets us to media ecosystem as of today is onto the top before their short you have seen the A 5 Parties funding this community and you have the whole stack that is busier starting this year the basics at the bottom of now with the new Meyer M as well as that included Mr. Chips in Baldwin manufacturers I was storage share companies system integration companies are also but as well you have so many research teams included some of them and that's basically year something that is also a key 1st of that group too much us to do something with companies and try to sell stuff but also to to keep the people involved in research and trying to get best whole method enhanced but to just a map of like a fund of 20 members so as of today and in the 2 countries and it these days I think most of that as the standard around the US because there's BCD headquarters of 5 mn demand a lab still located and that's why also of many of the universities their joint in those areas but so this year I try to get members over and even those partners also started to bring out 1st products but I think I have 1 chart for that lead us well that's busy and so on just to illustrate also how open Paul some whole matches all the other edition of the initiatives once we have this open computer group that this I think more of concentrating on mechanical electrical synchronization and specifications on dual bulk loops some all is trying to move to fill that gap between what I was coming from top from now on software of Linux or 2nd on and over computer at the bottom and all the gap here is somewhat part of that old Apollo initiative but it's basically just 1 fact off at a busy and as you can see the green box in there but the limits itself to the particles and I O in the connections so internally want that's how it should sit on or how it's intended to be the um and know what special about its and just to show you that the at the current vote map of the true if this is you still accurate but it is it gives you an overview of how that there is so was be dropped out over the last years of so I suppose is an ineffective you try to to have real roots of a sense of regular of push of products out and commodity onto that different needs and if you look at Paul and there different circumstances so because you have a different that markets that you will detect if you look at the architecture this look much different as the process of the OGI each and every time at the same point in time so there's a very regular pattern by unipolar site that needs to be much more flexibility at to ETD customer changes on Market changes at end but as you can see it the next years so 10 7 nanometers art products that need to come out and this you will be you of specified and pushed this departments but but as the do really existing thing about the job Polish differ so that you have some try to connect at the end of the submission before of external accelerators into the system is not just to plant them into a but to make them part of 2 systems to make them part of the at the address map of process so so it's not just an ionic of some details also but it's a politically that makes it really fast but compared to other customers also this this Paul aid enhanced products of put together with the media and the and the link for was introduced as an it's own Lincoln built into the CPU and to make that uh and therefore much faster into those GP use the mature if you wrote about this announcement quite some weeks ago so that those so that in US that of the the distribute if those 2 systems and there was some kind of government contracts on 1 by media to get of IBM to distributed systems and busy day consistently it's hot there of of Apollo poses a connected to it is to be used and this year but a room full of those services at and bt make possible some but the competitions and that it uses the news generation of Guinea link and so busy that provides a Bentley you obit able to get the farthest abuse but that's the same thing and but you see that you have found this in the face is not just the way we choose to have PCI generations available on to get the cost of just 52 cap EU extensions that you have and then your you can get so excited spots attached also referred can be overcome here on this just basically this new particle and hot they're built into process the but this is just a numbers not full of marketing and they see the variable of falsehood exible by was able to a busy reduce 20 thousand so into 31 pulsar was just to to get an idea of what that makes the at the end of 1 thus for anomalous that's regarding dt community itself and the idea is basically to get debate test the possibility to get everybody involved but that means there are people that the rights of the people that use of fear or people against that that you still systems and to try to a base you get those differed in the in the groups together and have a place for them in that all communities and that's why I the creates a different groups and to get more into technical discussions and that the technical solutions at the end and Anastasio what because you have today on doubt different different 13 different go groups available and there are some that I'm even both the some colleagues so so and as busier talking about real problems and solutions at the end of the trying to solve them at different parts of the 2nd and that's busy for the last slides of regarding what that is at the end of this is something something of that costs money and that's why companies need to to to to spend by to be part of power on but if you are coming from academia and that you are able to spend pseudo loss and so part of the community and as different catalysts between them on so it needs to be some a little to predict that way have obviously but but you able to get involved so especially in the coming from research and have even check today and I'm you mention should be involved so the year part of 1 universe you have seen the charts so that there's a there are already some German universities joining that the community here but most of them still located in US and that was all I know from my side
just read all you so it was not intended to be it in that also the complete if you have more questions just come to me and I like if you also context to other people if you want to talk more details regarding and interfaces or the context and those of groups of on but with those you should check all bolded or condone that's it upside so that you can get more information about that and turned out that gave you some inspiration and some information about what was going on and you have a pretty at the thank you very much material we do in term for the questions and the someone already lined up my 1 please I as a hacker have a power of 1 of these days was of and I think if you have connected to universities of 2 possible that some of their own yeah but would right so much and for you if you are in now and you have to do is do you have that that that is the use of the about once the deal with all the time OK the overview of the field will not use the power and all of the all of this is that the international covenant has a question yes so we talked about could companies and academia joining the open call and warn the user wants to know on that learns from the inside to offer affordable systems to consumers so they can join I think that's the key and this is an idea fall for companies to to get cheap systems available on 40 you you and you so I think it's a to comparative into the wall somebody to market IBM is looking into the so it's more at this and related issue at the moment like 4 of 1 question about what's the firm areas such as 400 thousand lines of code why is that they can work when you just for you score which already supports quark and power the older generations that would be this is because I I think it's just a state that is coming out with machines already of this from and identified and decision is coming from real data you have additional lines of code so written for their stuff and it just use what they all get along that would be considered abilities of likely to reduce its you suspect I like to of course in the lumen of the law it's never been about performance has been about mobile phones use those solutions of some numbers for people over to the I don't have numbers you on the need to develop some ideas of the again performances that what counts for the customers will be a good performance and the ones that you can so I lost of the food to improve performance over time and now with the view so that each you know more display different just of it's unfortunate that the quantity questions lining up unfortunately of time and were reserved for that the thing until you still around so I grabbed him and ask him in person so please give another round of applause from until thank you very much
so what the
and is compared to the the PEP
Maschinenschreiben
Statistische Schlussweise
Open Source
Mikroarchitektur
Unrundheit
Information
Zentraleinheit
Biprodukt
Farbverwaltungssystem
Roboter
Zusammengesetzte Verteilung
Dämpfung
Strom <Mathematik>
Softwareentwickler
Zentraleinheit
Leistung <Physik>
Fitnessfunktion
Resultante
Expertensystem
Bit
Font
Programmverifikation
Information
Softwareentwickler
Fokalpunkt
Gesetz <Physik>
Zentraleinheit
Übergang
Leistung <Physik>
Schätzwert
Umwandlungsenthalpie
Prozess <Physik>
Punkt
Programmverifikation
Versionsverwaltung
Zahlenbereich
Mailing-Liste
Symboltabelle
Quellcode
Biprodukt
Zentraleinheit
Mathematische Logik
Quick-Sort
Arithmetisches Mittel
Deskriptive Statistik
Streaming <Kommunikationstechnik>
Datenfeld
Rechter Winkel
Codierung
Leistung <Physik>
Zeitrichtung
Zentraleinheit
Gerade
Web Site
Subtraktion
Mathematische Logik
Prozess <Physik>
Gewichtete Summe
Punkt
Quader
Extrempunkt
Desintegration <Mathematik>
Physikalismus
Programmverifikation
Versionsverwaltung
Zahlenbereich
Dienst <Informatik>
Systemprogrammierung
Komponente <Software>
Hardwarebeschreibungssprache
Physikalisches System
Virtuelle Maschine
Informationsmodellierung
Softwaretest
Vorzeichen <Mathematik>
Code
Mini-Disc
Skript <Programm>
Zusammenhängender Graph
Indexberechnung
Softwareentwickler
Hardware
Implementierung
Soundverarbeitung
Videospiel
Sichtenkonzept
Programmablaufplan
Logische Schaltung
Programmverifikation
Firmware
Quellcode
Physikalisches System
Ein-Ausgabe
Biprodukt
Integral
Datenstruktur
Flächeninhalt
Digitaltechnik
Zentraleinheit
Faserbündel
Varietät <Mathematik>
Resultante
Offene Menge
Prozess <Physik>
Punkt
Gewichtete Summe
Nabel <Mathematik>
Gemeinsamer Speicher
Gesetz <Physik>
Service provider
Strategisches Spiel
Spezialrechner
Mooresches Gesetz
Prozess <Informatik>
Code
Trennschärfe <Statistik>
Mustersprache
Mixed Reality
Wurzel <Mathematik>
Verschiebungsoperator
Internetworking
Konstruktor <Informatik>
Nichtlinearer Operator
Befehl <Informatik>
Hardware
Physikalischer Effekt
Gebäude <Mathematik>
Güte der Anpassung
Speicher <Informatik>
Biprodukt
Arithmetisches Mittel
Heimcomputer
Konstante
Randwert
Software
Generator <Informatik>
Datenfeld
Datenverarbeitungssystem
Strategisches Spiel
Mooresches Gesetz
Zentraleinheit
Aggregatzustand
Subtraktion
Multimedia
Klasse <Mathematik>
Mathematisierung
Zahlenbereich
Kraft
Maßerweiterung
ROM <Informatik>
Term
Data Mining
Unternehmensarchitektur
Open Source
Physikalisches System
Virtuelle Maschine
Informationsmodellierung
CMM <Software Engineering>
Mobiles Internet
Zusammenhängender Graph
Strom <Mathematik>
FLOPS <Informatik>
Speicher <Informatik>
Datenstruktur
Hilfesystem
Hardware
Leistung <Physik>
Kollaboration <Informatik>
Architektur <Informatik>
Matching <Graphentheorie>
Open Source
Datenmodell
Paarvergleich
Beanspruchung
Physikalisches System
Elektronische Publikation
Cloud Computing
Gerade
Quick-Sort
Coprozessor
Moment <Stochastik>
Diagramm
Offene Menge
Mereologie
Codierung
Leistung <Physik>
Computerarchitektur
Modelltheorie
Streuungsdiagramm
Offene Menge
Kernel <Informatik>
Domain <Netzwerk>
Einfügungsdämpfung
Desintegration <Mathematik>
Systementwurf
Benutzerfreundlichkeit
Pi <Zahl>
Raum-Zeit
Softwaretest
Field programmable gate array
Wellenpaket
Standardabweichung
Mustersprache
Protokoll <Datenverarbeitungssystem>
Gasdruck
System-on-Chip
Phasenumwandlung
Softwaretest
Kraftfahrzeugmechatroniker
Strömungsrichtung
Störungstheorie
Biprodukt
Gruppenoperation
Randwert
Generator <Informatik>
Software
Datenverarbeitungssystem
Rechter Winkel
Zellularer Automat
Elektronischer Fingerabdruck
Prozessfähigkeit <Qualitätsmanagement>
Hypercube
Eindeutigkeit
Simulation
Schnittstelle
Subtraktion
Mathematisierung
Dienst <Informatik>
Identitätsverwaltung
Supercomputer
Unternehmensarchitektur
Graphikprozessor
Chiffrierung
Loop
Open Source
Variable
Parallelrechner
Reelle Zahl
Datennetz
Maßerweiterung
Pixel
Architektur <Informatik>
Gerichtete Menge
Systemintegration
Datenmodell
Binder <Informatik>
Summengleichung
Auswahlaxiom
Whiteboard
Offene Menge
Bandmatrix
Leistung <Physik>
Partikelsystem
Mustererkennung
Data Mining
Abstimmung <Frequenz>
Konfiguration <Informatik>
Punkt
Prozess <Physik>
Gemeinsamer Speicher
Element <Mathematik>
Adressraum
Gruppenkeim
Befehl <Informatik>
Resonanz
Synchronisierung
Maßstab
Regulärer Graph
Gruppentheorie
Prozess <Informatik>
Minimum
Wurzel <Mathematik>
Druckertreiber
Inklusion <Mathematik>
Umwandlungsenthalpie
Softwareentwickler
Perspektive
Prozess <Informatik>
Machsches Prinzip
Abstimmung <Frequenz>
Applet
Quellencodierung
Globale Optimierung
Speicher <Informatik>
Übergang
Vorzeichen <Mathematik>
Firmware
Kugelkappe
Rechenschieber
Framework <Informatik>
Gruppenkeim
Verschlingung
ATM
Dualitätstheorie
Zentraleinheit
Server
Web Site
Quader
Zahlenbereich
Maschinelles Lernen
Kugelkappe
ROM <Informatik>
Physikalische Theorie
Physikalisches System
Systemprogrammierung
Task
Software
Fokalpunkt
Inverser Limes
Zusammenhängender Graph
Biprodukt
Speicher <Informatik>
Softwareentwickler
Grundraum
Hilfesystem
Leistung <Physik>
Hardware
Einfach zusammenhängender Raum
Streuungsdiagramm
Kollaboration <Informatik>
Elektronische Publikation
Integritätsbereich
Sprachsynthese
Informationsverarbeitung
Beanspruchung
Physikalisches System
Migration <Informatik>
Coprozessor
Design by Contract
Keller <Informatik>
Mapping <Computergraphik>
Fundamentalsatz der Algebra
Puffer <Netzplantechnik>
Flächeninhalt
Hypermedia
Mereologie
Ganze Funktion
Computerarchitektur
Term
Momentenproblem
Datensichtgerät
Zahlenbereich
Unrundheit
Gesetz <Physik>
Term
Code
Eins
Virtuelle Maschine
Hacker
Grundraum
Gerade
Leistung <Physik>
Schnittstelle
Addition
Quarkmodell
Physikalisches System
Kontextbezogenes System
Entscheidungstheorie
Generator <Informatik>
Datenfeld
Flächeninhalt
Information
Schlüsselverwaltung
Aggregatzustand
Hypermedia
Medianwert
Systemprogrammierung
Pauli-Prinzip

Metadaten

Formale Metadaten

Titel openPower - the current state of commercial openness in CPU development
Untertitel is there no such thing as open hardware?
Serientitel 34th Chaos Communication Congress
Autor Michel, Matteo
Lizenz CC-Namensnennung 4.0 International:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen.
DOI 10.5446/34893
Herausgeber Chaos Computer Club e.V.
Erscheinungsjahr 2017
Sprache Englisch

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract How does developing future processors with yesterdays capabilities work out today? CPU development is something out of focus these days. In this lecture I would like to show the state-of-the-art processor development flow of POWER processors from the first initial ideas to post-silicon testing. Apart from x86 Intel products there have been initiatives across the hardware industry to form some alternative business model. I would like to show if and how this compares to real open principals.
Schlagwörter Hardware & Making

Zugehöriges Material

Video wird in der folgenden Ressource zitiert

Ähnliche Filme

Loading...
Feedback