The SiCS Multicore Day took place last week, for the tenth year in a row! It is still a very good event to learn about multicore and computer architecture, and meet with a broad selection of industry and academic people interested in multicore in various ways. While multicore is not bright shiny new thing it once was, it is still an exciting area of research – even if much of the innovation is moving away from the traditional field of making a bunch of processor cores work together, towards system-level optimizations. For the past few years, SiCS has had to good taste to publish all the lectures online, so you can go to their Youtube playlist and see all the talks for free, right now!
Once upon a time, when multicore processors were novelties, multicore was motivated by the simple fact that it was impossible to keep raising the clock frequency of processors. More “clocks” simply would result in an overheated mess. Instead, by adding more cores, much more performance could be obtained without having to go to extreme frequencies and power budgets. The first multicore processors pretty much kept clock frequencies of the single-core processors preceding them, and that has remained the mainstream fact until today. Desktop and laptop processors tend to stay at 4 cores or less. But when you go beyond 4 cores, clock frequencies tend to start to go down in order to keep power consumption per package under control. A nice example of this can be found in Intel’s Xeon lineup.
Continue reading “Clocks or Cores? Choose One”
I have read some recent IBM articles about the POWER8 processor and its hardware debug and trace facilities. They are very impressive, and quite interesting to compare to what is usually found in the embedded world. Instead of being designed to help with software debug, it seems the hardware mechanisms in the Power8 are rather focused on silicon bringup and performance analysis and verification in IBM’s own labs. As well as supporting virtual machines and JIT-based systems!
While I was on vacation, Wind River published a blog post I wrote about the new multicore accelerator feature of Simics 5. The post has some details on what we did, and some of the things we learnt about simulation performance.
When mobile phones first appeared, they were powered by very simple cores like the venerable ARM7 and later the ARM9. Low clock frequencies, zero microarchitectural sophistication, sufficient for the job. In recent years, as smartphones have come into their own as the most important computing device for most people, the processor performance of mobile phones have increased tremendously. Today, cutting-edge phones and tablets contain four or eight cores, running at clock frequencies well above 2 gigahertz. The performance race for most of the market (more about that in a moment) was mostly about pushing higher clock frequencies and more cores, even while microarchitecture was left comparatively simple. Mobile meant “fairly simple”, and IPC was nowhere near what you would get with a typical Intel processor for a laptop or desktop.
Today, that seems to be changing, as the Nvidia Denver core and Apple’s Cyclone core both go the route of a few fat cores rather than many thin cores.
Via the EETimes, I found a very interesting talk by Bristol professor David May, presented at the 4th Annual Bristol Multicore Challenge, in June of 2013. The talk can be found as a Youtube movie here, and the slides are available here. The EETimes focused on the idea to cut down ARM to be really RISC, but I think the more interesting part is Professor May’s observations on multicore computing in general, and the case for and against heterogeneity in (parallel) computers.
Probably thanks to the yearly Mobile World Congress, there have been a slew of recent announcements of mobile application processors recently. Everything is ARM-based, but show quite some variety in the CPU core configurations used. Indeed, I think this variety has something to say on the general state of multicore.
The 2012 edition of the SiCS Multicore Day was fun, like they have always been in the past. I missed it in 2010 and 2011, but could make it back this year. It was interesting to see that the points where keynote speakers disagreed was similar to previous years, albeit with some new twists. There was also a trend in architecture, moving crypto operations into the core processor ISA, that indicates another angle on the hardware accelerator space.
Nvidia recently announced that their already-known “Kal-El” quad-core ARM Cortex-A9 SoC actually contains five processor cores, not just four as a “normal” quad-core would. They call the architecture “Variable SMP”, and it is a pretty smart design. The one where you think, “I should have thought of that”, which is the best sign of something truly good.
By chance, I got to attend a day at the UPMARC Summer School with a very enjoyable talk by Francesco Zappa Nardelli from INRIA. He described his work (along with others) on understanding and modeling multiprocessor memory models. It is a very complex subject, but he managed to explain it very well.
Episodes 299 and 301 of the SecurityNow podcast deal with the problem of how to get randomness out of a computer. As usual, Steve Gibson does a good job of explaining things, but I felt that there was some more that needed to be said about computers and randomness, as well as the related ideas of predictability, observability, repeatability, and determinism. I have worked and wrangled with these concepts for almost 15 years now, from my research into timing prediction for embedded processors to my current work with the repeatable and reversible Simics simulator.
I have another blog up at Wind River. This one is about multicore bugs that cannot happen on multithreaded systems, and is called True Concurrency is Truly Different (Again). It bounces from a recent interesting Windows security flaw into how Simics works with multicore systems.
SCDSource ran a short but good article summarizing a few DAC talks that I would liked to attend. it mostly about the experience of long-term parallel programming research David Bailey in presenting results in the field…
Past Tuesday, I attended the Freescale Design With Freescale (DWF) one-day technology event in Kista, Stockholm. This is a small-scale version of the big Freescale Technology Forum, and featured four tracks of talks running from the morning into the afternoon. All very technical, aimed at designing engineers.
My post on SiCS multicore, as well as the SiCS multicore day itself, put a renewed spotlight on the GPGPU phenomenon. I have been following this at a distance, since it does not feel very applicable to neither my job of running Simics, nor do I see such processors appear in any customer applications. Still, I think it is worth thinking about what a GPGPU really is, at a high level.
Last Friday, I attended this year’s edition of the SiCS Multicore Day. It was smaller in scale than last year, being only a single day rather than two days. The program was very high quality nevertheless, with keynote talks from Hazim Shafi of Microsoft, Richard Kaufmann of HP, and Anders Landin of Sun. Additionally, there was a mid-day three-track session with research and industry talks from the Swedish multicore community. Continue reading “SiCS Multicore Day 2009”
Freescale has now released the collected, updated, and restyled book version of the article series on embedded multicore that I wrote last year together with Patrik Strömblad of Enea, and Jonas Svennebring, and John Logan of Freescale. The book covers the basics of multicore software and hardware, as well as operating systems issues and virtual platforms. Obviously, the virtual platform part was my contribution.
In a post from late June, Jeff Atwood at Coding Horror discusses the horrible cost of a large HP server (scaling up to 32 processor cores in eight AMD x86 sockets), compared to a bunch of simple single-socket basic servers. There are some interesting notes on relative costs of small-and-simple servers, including things like administration and power. There is an undercurrent to the post and the comments that the big HP machine is “overpriced”. I don’t think it is. If you have ever had Erik Hagersten as a teacher in computer architecture, you will know why.
About two months ago, Cavium Networks launched their second generation of Octeon chips, the Octeon II. The most obvious difference to the previous generation (Octeon, Octeon Plus) is a new MIPS64 core with much better support for hypervisors and virtualization. There are some other interesting aspects to this chip, though.
Last year in a blog post on video encoding for the iPod Nano, I complained about the lack of performance on my old Athlon. A bit later, I noted that (obviously) video encoding is a good example of an application that can take advantage of parallelism. Yesterday I put these two topics together in a practical test. And it worked nicely enough.
Yes, when does hardware acceleration make sense in networking? Hardware acceleration in the common sense of “TCP offload”. This question was answered by a very nicely reasoned “no” in an article by Mike Odell in ACM Queue called “Network Front-End Processors, Yet Again“.
The EETimes article Multicore CPUs face slow road in comms piqued my interest. There is an interesting chart in there about just how slow more-than-one-core processors will be in penetrating a vaguely defined “comms” market place. I can believe that, but I think their comments on the PowerQUICC series require some commentary…
A common question from simulation users to us simulation providers is “can I simulate a machine with N cores”, where N is “large”. As if running lots of cores was a simulation system or even a hardware problem. In almost all cases, the problem is with software. Creating an arbitrary configuration in a virtual platform is easy. Creating a software stack for that arbitrary platform is a lot harder, since an SMP software stack needs to understand about the cores and how they communicate.
Essentially, what you need is a hardware design that has addressing room for lots of cores, and a software stack that is capable of using lots of cores — even if such configurations do not exist in hardware. Unfortunately, since software is normally written to run on real existing machines, there tends to be unexpected limitations even where scalability should be feasible “in principle”.
Here is the story of how I convinced Linux to handle more than two cores in a virtual MPC8641D machine.
The best way to learn something is to try, fail, and then try again. That is how I just learned the basics of multiprocessor interrupt management. For an educational setup, I have been creating a purely virtual virtual platform from scratch. This setup contains a large number of processors with local memory, and then a global shared memory, as well as a means for the processors to interrupt each other in order to notify about the presence of a message or synchronize in general. Getting this really right turned out to be not so easy.
Jack Ganssle wrote a column about the failure of multicore to scale, based on an article in IEEE Spectrum. He makes the following claim:
Now a study in IEEE Spectrum shows that even for the classic embarrassingly parallel problems like weather simulations multicore offers little benefit. The curve in that article is priceless. As the number of cores grow from two to 64 performance plummets by a factor of five. Additional processors nullify each other.
Call it the Nulticore Effect.
I keep looking out for interesting examples of parallel software, and there is constant trickle of these. This past week I spotted a couple of new ones in the EDA field: SPICE simulation and chip timing analysis.
It is a week ago now, and sometimes it is good to let impressions sink in and get processed a bit before writing about an event like the SiCS Multicore Days. Overall, the event was serious fun, and I found the speakers very insightful and the panel discussion and audience questions added even more information.
More from the SiCS multicore days 2008.
There were some interesting comments on how to define efficiency in a world of plentiful cores. The theme from my previous blog post called “Real-Time Control when Cores Become Free” came up several times during the talks, panels, and discussions. It seems that this year, everybody agreed that we are heading to 100s or 1000s of “self-respecting” cores on a single chip, and that with that kind of core count, it is not too important to keep them all busy at all times at any cost. As I stated earlier, cores and instructions are now free, while other aspects are limiting, turning the classic optimization imperatives of computing on its head. Operating systems will become more about space-sharing than time-sharing, and it might make sense to dedicate processing cores to the sole job of impersonating peripheral units or doing polling work. Operating systems can also be simplified when the job of time-sharing is taken away, even if communications and resource management might well bring in some new interesting issues.
So, what is efficiency in this kind of environment?
The two days of the SiCS Multicore Days is now over, and it was a really fun event this year too. I will be writing a few things inspired by the event, and here is the first.
Kunle Olukotun‘s presentation on the work of the Stanford Pervasive Parallelism lab included a diagram where they showed a range of domain-specific languages (DSL) being compiled to a universal implementation language. That language is currently Scala, and in the end all applications end up being compiled into Scala byte codes, which are then optimized and dynamically reoptimized and executed on a particular hardware system based on the properties of that system. Fundamentally, the problem of creating and compiling a DSL, and combining program segments written in different DSLs, is solved by interposing a layer of indirection.
But this idea got me thinking about what the best such intermediary might be for large-scale general deployment.