Thanks for posting those Gizmo,
It would be interesting to know though whether that is actually one of the first 'production models', or indeed just another developement or prototype board.
We still look forward to reading your initial thoughts and findings in due course however, so I can see you are in for a busy weekend!
From the pics posted on the front of the raspberrypi.org website, they have the bright yellow composite video jack, the same as mine, whereas the other early proto pic I have seen has a black jack, so I would guess it is from that new batch...
Whether it is the same as the final relased production units will depend on the compliance testing and whether any needs to be changed, I imagine.
Please, comeback soon.
Connect the plugs
we look forward to your first feelings
Happy for you but how can competition winners get there Pi's before us that got up at 5:30am and fought our way through the chaos to get a order in?
RaspberryPi.org has just posted the supposedly first pictures of the first boards in the country, these are either a few days old or Farnell has had it's own delivery to be able to deliver to Road Testers as well.
Does anyone have an explanation or we being mushroomed again?
Liz from Raspberry Pi here - the board that went out as a competition prize was one of element14's qualification prototype boards, not a production board.
ETA: Eben says to tell GizmoB73 that the Raspberry Pi you won is one of a few which he and Pete were working on personally to get ready for EM testing. Hope that adds some sort of frisson!
Well if it is any consolation, probably just like you and many others, I have followed the foundations progress with eager anticipation for months and I was up at 05:30 too and spent a good few hours hitting the F5 key as soon as I read the announcement at 06:00.
By an incredible stroke of good fortune, I was selected as a Road Test Winner, but as I mentioned in my first post, this is a prototype and not a final finished product.
Thanks for adding some clarity Liz, and to know that Eben and Pete have been personally working on this board is pretty mind blowing. I think frisson would be an understatement for how I am feeling.
Look forward to your experiences, now I have got over my insane jealousy it will be nice to get an independent view of it.
It would be better for mine and others blood pressure if they didn't keep dangling these carrots under our noses like webinars for boards we cant have and pictures of boxes full of them that they can't ship and either just put the whole thing on hold or send me one and put me out of my misery.
I'm off for a stiff drink, bet gizmo will be up all night playing with his new toy the lucky so and so
Well done - that's awesome! Let us know how you get on, it'll be nice to know what your first impressions are.
Nobody is frustrated or jealous here, element14 is not a kids forum.
So please, here are some tests we all would like you to do,
the boot time
the maximum display resolution
the reported available memory
some usefull commands
Ttcp - measures the point-to-point bandwidth over a network connection
Hdparm - "-t" and "-T" options can be used to measure disk-to-memory transfer rates
Dga - the "-b" option measures CPU/video memory bandwidth
Bonnie++, io throughput benchmark,
Iozone, broad filesystem analysis. The benchmark tests file I/O performance,
BYTEmark, CPU benchmark suite, reporting CPU/cache/memory, integer and floating-point performance
Cachebench, measures bandwidth of the memory subsystem (L1, L2 and main memory)
It looks like we have byte-unixbench results here:
courtesy of OmerK
[updated link per comments below]
Re your last link .... Apparently DesignSpark user pages are not public, very annoying.
Hmm! Seems that http://www.designspark.com/content/adventures-pi-initial-thoughts may be the same version except publically available. I can't tell for sure since I'm not registered on DesignSpark! Not sure how DesignSpark differentiates between 'content' and 'knowledge' ;-)
Thanks xtramural, your link worked without login. Yes, hopefully it's the same content.
Addendum: Excellent stuff from Omer -- he also has a site here http://omer.me/ .
I like Erlang a lot as a language (I did my research in concurrency, way back), so this is material that I'll be following closely.
That last post should have given you a few ideas Gizmo... assuming that you needed some, that is!
Anyway, it's getting a bit late now here in the U.K., and I'm off to bed.
I am sure though that you will be working all night with your new 'baby', and I will look forward to reading a full report in the morning!
Ideas are always welcome, time permitting. Most of this Linux is new to me, closest for me was a DRS6000 about 12 years ago, and I was only writing code on it. You'll need to be patient and type S-L-O-W-L-Y for me, but glad to answer what/when I can.
My SD Card is:
Samsung 8GB Class 6
Running the "debian6-17-02-2012" image.
From plug to terminal login prompt (not into LXDE), whilst connected to Ethernet, startup time is about 36 seconds.
Screen resolution using xrandr
Memory using free -m
hdparm -Tt /dev/mmcblk0p1
Timing Cached reads: 306MB in 2.00 seconds = 152.92 MB/sec
Timing buffered disk reads: 32MB in 3.07 seconds = 10.42 MB/sec
My very preliminary thoughts are:
Overall response feels capable, but not blisteringly fast when running LXDE. As the first Debian image, I would imagine there is scope for improvement with future tweaks and builds, but certainly not dissapointing for something so small and cheap.
So far the only real indication to me that this is such a low cost device, is the quality of the SD Card socket. Having been used to the slick metal spring loaded versions in the past, this is a rather simple plastic friction affair. If I find myself swapping cards a lot, I might decide to invest in an SD Card extender. Otherwise, the only other weak point is the micro USB socket, but that should be no surprise to anybody that has used a device with one of these before. Best to avoid heavy hands and if unplugging a lot (to disconnect the power), then use the other end of the cable, or a power switch. A careful blob of epoxy glue over the socket and surrounding board would probably strengthen things up too, but I won't be putting any near mine.
I haven't even scratched the surface, but what I like best so far is it's multiple personalities. It is just a matter of swapping the SD Card and in a few seconds it does something different. I am keeping plenty of card backups at regular intervals, it is a simple task and makes it so easy to roll back if something goes wrong. It is such a simple and obvious thing, but I hadn't anticipated how useful/empowering that would be as a linux noob.
I don't feel like I have done very much this evening, but really need some sleep now. Well, I will try anyway.
Great stuff GizmoB73, it's very cool to read your experiences.
And I echo what jacques_nil said, nobody here will get annoyed at anything you write, good or bad, newbie or expert, it's all useful input on the Pi experience and interesting. All the people who have the pettiness to get annoyed are on the official forum, which encourages intolerance.
I ran up the Fedora 14 Remix distro for the Raspberry Pi under QEMU emulation a little while back, and reported on the info revealed by running some standard commands, see here:
If you have a moment to run the same commands on your live hardware and paste the result, it would be interesting to see how it compares with the emulation. The fact that mine was Fedora and yours is Debian will make some difference, but most things should be identical.
Thanks Morgaine, I will try to get around to this for you.
Jacques_nil: here is the full Bonnie output for that card:
Version 1.96 ------Sequential Output------ --Sequential Input- --Random- Concurrency 1 -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks-- Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP raspberrypi 368M 23 99 8722 25 4698 14 360 99 13419 13 119.8 18 Latency 830ms 3339ms 940ms 30000us 69999us 610ms Version 1.96 ------Sequential Create------ --------Random Create-------- raspberrypi -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete-- files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP 16 1654 61 +++++ +++ 2739 66 1847 67 +++++ +++ 2080 50 Latency 989ms 10000us 10000us 1069ms 10000us 2014ms
As for the HD Video streaming... easier said than done.
I am currently chasing my tail trying to get something to work. VLC doesn't work because it needs a specific version of alsa-lib. It would appear that on debian this is called libasound2 and is the latest version when I checked, so not sure how to get that working. I am guessing that I probably need an older version, and that VLC isn't pulling that in.
I then thought I would try a different media player, but it looks like "debian-multimedia-keyring" is pretty fundamental and I cannot find a source for it that works on the Pi.
I guess these are just noob issues on my part, and I am learning loads each time I hit another dead-end.
OK, got "debian-multimedia-keyring" to install manually by downloading the .deb file and installing with dpkg -i.
I then installed SMPlayer and Gnome MPlayer. Tested with a standard definition file streaming from my NAS over my Gigabit backbone. Playback had no audio and was at best running at one third of the frame rate it should have been. I have no doubt that the R-Pi is capable of better than that, but something is obviously not right.
I wonder if it is the debian build?
Instead of mplayer, you might have better luck with ilplayer.
It is apparently a custom player supplied with the RPi that Paul Maunders
used (he bought beta board #8 at the January eBay auction) and blogged
about it at:
He found ilplayer to be much faster than mplayer, but also had some trouble with audio.
He was able to get sound using
He disappeared after being asked about watching videos from hulu at 1080p,
so don't try that.
Thanks coder27, that was an interesting read, I hadn't seen it before. I think the 128MB memory side of things has been sorted out now, but I would agree with the rest of his performance observations.
I haven't been given any files with the Pi other than what I can download from here, so I don't have ilplayer. ilplayer sounds like something that Broadcom have put together internally to harness the Hardware accelleration.
If anybody reading can let me have a copy, i'd like to give it a try.
As for the requests, I will see what I can do, but I am away for a week after this evening with limited access or time for the internet. I will be taking the Pi with me but doubt that I will get a lot of time to use it, and also won't have a network etc. with me. So if I stop responding for the week, which is likely, don't worry.
It's so sad that you left us, we hope you do not disappear forever, as the two other owners who never came back.
So please, post as many tests as you can before leaving.
I am sure he will come back Jacques... he does say that he will only be away for one week!
Sorry about the timing but yes, I will be back, and I will try and keep an eye on these forums from my phone when I can, but don't expect any Pi action.
Thanks for everybodys suggestions and patience.
"Playback had no audio and was at best running at one third of the frame rate it should have been."
sounds like it might not be using any hardware accelleration (GPU) at all ? If it can play 1080p MP4 files, SD should be easy...
apparently none of the standard debian packages are accelerated,
only ilplayer. Paul got VLC to load, but it was too slow.
Apparently AAC sound is supported, but not AC3.
Once again, thank you
Sequential disk write:
with putc() : 23Kbytes/sec, cpu overhead 99% <- horrible, but this is due to SD card that do not like small writes
with write() : 8.72MB/sec, cpu overhead 25%
read,seek,write : 4.7MB/sec, cpu overhead 14%
Sequential disk read:
with getc(): 360Kbytes/sec, cpu overhead 99%
Block read: 13.4MB,, cpu overhead 13% <- better than your first test
Seek time: 8.3ms ( = 1 / 119.8)
Read perfs. are not bad, but the poor write perfs.are for sure 100% due to the SD, wich are optimized for writing big block of data.
That should have no effect on applications that read a lot and write few, but when running multiple tasks, if the OS start swapping to the SD it could feel like snail driving.
I wonder if the results will be better with a SSD attached to the USB port, but is it possible to boot from usb ?
It's not possible to boot from USB. But, a cheap SD card can be your "boot disk" and then load the rest and run from the USB HDD. (Kinda like the olden days of booting Linux from a floppy.)
[quote]what I like best so far is it's multiple personalities. It is just a matter of swapping the SD Card and in a few seconds it does something different.[/quote]
Many moons ago there was a terminal type PC called an Einstein, it didn't have an OS, one purchased - 51/4" floppies that contained the emulated OS (such as they were) of your choice.
Thank you for he post Gizmo, that is interesting.
10.4 MB/s for the disk read speed looks quite low for a class 6. Should be around 20MB/s
Do you have the write speed ?
Another nice and simple test would be to try some HD video steamig over the network,
this would invalidate (or validate) the therory that the shared ethernet over usb is a huge bottleneck.
I continue to be surprised when people are suprised about the ease (or lack of) playing media on this device. The hardware acceleration is there and it makes for a great demo but the reality is that it will be some time before client libraries and applications follow. I have a little arm based board, not much bigger than the Pi but probably double the CPU performance and it is far too slow to play any sort of media with a software codec such as you would typically find in mplayer or vlc (or Chromium browser). Fortunately there is gstreamer support for my hardware so I can use an application such as totem but the Pi doesn't have anything that high level yet. In the day or two I have been looking I still haven't found a browser that will play html5 video.
The Pi is a great education computer for an amazing price but it is an engineering compromise to reach that price. It is a shame that "end users" and developers are fighting over the same limited supply as there is sure to be some negative reviews from the people who don't understand the work that still is required to get this thing ready for its main mission.
I agree, but as part-time teacher myself, I do not understand the definition of an "education computer",
what kind of education ? software or hardware, application level or sytem level, children or bachelors ...
On the other hand, the communication on the web around the Raspberry is quite focused on its possible HD video playback and accelerated graphics,
and there is a huge number who pre-ordered thinking about using it as an XBMC HTPC. Few of them seek a solution to learn python.
Watch Eben's talk.
23-30 years ago when he and I were in our teens, we got a "basic" computer that provided you with the option to start programming. Maybe there were some games available as well. Nowadays you get a slick games-console, but you can't program it. By allowing youngsters to program their computers they will learn about it.
Eben also talks about the group of primary school children who when shown a RPI and a snake-game in Python, started to modify the thing to show different colors etc etc. Those are the first steps to get children interested in programming and in the "how does a computer WORK".
THAT is what the RPI is meant to do.
What is there in that story that couldn't be done on a PC just as easily (and most UK schools have plenty of them) ?
The RPi with it's big OS and secret hardware is just as opaque (as a PC) from a hardware point of view - no way that Joe Public (regardless of age) can see how it works in the way you could with the early 8 bit machines.
If you want to learn about hardware and processing from the ground up you would do better with an Arduino.
Having said that the RPi has generated a lot of interest and if some of that ends up with more people interested in how things work that's a good thing - the disapointment from the over-hyping will be a bad thing.
Arduinos are great fun, I have several that i have used in various projects, but clearly they are not enticing people away from their PC's and encouraging them to experiment.
I think the Pi has the possibility of bridging that gap between the PC and the Arduino. It will hopefully draw people into Linux which I agree is a big OS but does require a bit more understanding of the underlaying hardware. With Gert's GPIO bolt-on, people could easily start experimenting with electronics and interfacing.
I hope that if introduced to these in school, the children could buy one and take it further at home. Most parents would prefer them hacking with a Pi rather than try to use the family PC or laptop if they have one.
There is plenty that can be done on an Rpi that can't (or won't) be done on a PC, simply because of its very low price.
I'm an example of this myself. I have lots of Intel-based machines here, and since I'm both an electronics engineer and a computer scientist I more or less know what I'm doing, yet I will not attach prototype hardware directly to any of my "real" PCs as a matter of principle. Things do go wrong occasionally when you're experimenting, and I don't relish blowing any of my gear up.
This is why the "computer hosts" for my hardware experiments are Arduinos and not Linux boxes --- if the Arduinos get blown up, the financial loss is tolerable, and there is no investment in operating system or filesystems to lose. But Arduinos are not really "home" for me. I need a host running a Unix-type system to really feel at home, but so far they have been too expensive.
The Rpi changes that totally. When they finally become stock items, I'll be at "home" on a Linux machine, yet totally willing to attach dodgy hardwre to the SPI or I2C without fear of what may happen. I may curse when the Rpi turns into a firework display, but $35 or so is not a huge barrier to overcome. For me it will open up many new applications, because a Unixy machine is so vastly more capable than an Arduino.
The point I was making was that an RPi running Python to make a Snake game using a standard keyboard and monitor is just an inconvenient substitue for the PCs which are already available in schools.
I'm sure that there will be things you can do with an RPi that you can't do with a PC that make use of it's low power and small size.
As for your concerns about blowing things up ...... hmmmm.
I do lots of development work with a target connected to a PC via a £1000+ worth of trace emulator - I don't find it inhibiting. If you need to interface to something dangerous (to you or computer) why not just use isolation in the interface ?
Because building your own isolated interface can fail as well, especially while you're commissioning it. In contrast I'm happy using the magnetic isolation of Ethernet on industry-certified boards, as it gives us the double protection of verified magnetic decoupling at both ends.
Really I can't understand where you're coming from at all. Surely your operating systems are not worthless to you so that you don't care about the risk? To err is human. I know of no human, no matter how expert and experienced, who does not make mistakes or suffer accidents. That certainly applies to me, and while you may not think so, it applies to you as well.
I agree, and as far as memory limitations etc. That is where I hope the collective genius of the masses around the world will come up with some great solutions.
I started programming when I was about 8, my dad won a Sinclair Spectrum in a competition. I then moved onto 6502 assembler with the BBC Micro's at school. I've lost count of the various languages I have used since then, but never tinkered with Perl.
Nowadays, there is ever more requirement for more memory and faster hardware. I am sure that this is because software development itself seems to become ever more abstracted. I often wonder what could be acheived with some good old back to basics code, I wait with baited breath to see what will emerge when these get into more peoples hands.
I am using this as a good excuse to get up to speed with Linux, and can't wait to get stuck into some robotics projects with it.
So you think the target is the education of children.
This is a very specific area with its own rules, more psychology, sociology, behavioral science, collaboration ...
This is not my area of work, thankfully
Or, because he speeks of a box, perhaps he thinks about something like OLPC project, but for developed countries.
I am not sure if this counts as a first unboxing, but I thought I would share it with you all anyway.
I was extremely lucky to be one of the Element14 Road Test winners and this morning and was very surprised and excited to receive a box from Farnell/Element14 via UPS.
I should say that it did come with a disclaimer that says it is an R&D Prototype and that it is not necessarily in the form that the final, finished product will be made available. I guess that is in case any changes are required as a consequence of the compliance testing.
At the risk of being deluged with questions that I cannot answer yet, or hate mail from the angry and frustrated (please don't hate me) I thought I would share a few pics before I get back to my day job.
Sorry, I just had to rip the box open before I got the camera out.
It really is very neat and small, but the really exciting part should hopefully come this evening when I get chance to power it up!
Here it is inside my first box. It is a plastic business card box that has now been recycled and put to a better use. This really brings home to me how small the R-Pi really is.
I still have to cut the holes when I get time.
I have a lot to learn and a lot of work before my Road Test Review, but I hope you like the pics...