Raspberry Pi GPU unlocking.

[Stephen Hill]

The processor is a RISC architecture, so the performance is closer to a 300 MHz PIII/PII than anything else. 256 MB of RAM is shared between the CPU and GPU and try getting an OS written in the last 10 years to run within 128 MB.

How about Tiny Core Linux: http://en.wikipedia.org/wiki/Tiny_Core_Linux
It won't run on the RPi but it only requires 48Mb of RAM and was written within the last 10 years. Latest release was 8 days ago!

[Stephen Hill]

I personally see the RPi as a toy - nothing more. I have one and will only be using it as a cheap and compact media player for my bedroom TV using OpenElec/XBMC.

[Mechatrommer]

I personally see the RPi as a toy - nothing more. I have one and will only be using it as a cheap and compact media player for my bedroom TV using OpenElec/XBMC.

look at the bright side! its 700MHz and 256MB RAM. not like my 8MHz only few KB RAM ARM-M3 dev kit here. it should be able to do 100X more (guesstimate) than what mine can. but if you find it slower, then something wrong, probably a "monkey" programmed that thing

[T4P]

The phillips has a 400Hz panel and it has those "Smart/Stupid TV things" and a 14bit video DAC

can anyone explain me what the purpose of a DAC is in a flatpanel TV ? especially since the pixels are digitally controlled... why do you need a dac ? marketing bullshittery ?

Yeah... marketing

I personally see the RPi as a toy - nothing more. I have one and will only be using it as a cheap and compact media player for my bedroom TV using OpenElec/XBMC.

look at the bright side! its 700MHz and 256MB RAM. not like my 8MHz only few KB RAM ARM-M3 dev kit here. it should be able to do 100X more (guesstimate) than what mine can. but if you find it slower, then something wrong, probably a "monkey" programmed that thing

O_o PIC that is, not ARM-M3, the lowest i've seen them go is 24MHz

[Mechatrommer]

O_o PIC that is, not ARM-M3, the lowest i've seen them go is 24MHz

its HY-Mini STM32V 1hunglo dev kit i got from ebay. it has 8MHz crystal connected to STM32F103VCT6. i dont know, havent looked at ARM architecture/datasheet in detail, maybe it has clock multiplier inside. anyway, if rasberry is a toy, then this 1hunglo is "toy of the toy".

[Mechatrommer]

On the downside, software engineering concepts like structured programming and modularity were not part of the lesson plans. Would they even be today, I wonder?

software engineering should be taught from ground up. first with conventional "old style" procedural programming. and then structured and modularity, and then OOP and then that "foundation class" or that c# things. its like mathematics you cant teach kids algebra where they cannot count 5+5 by fingers. otherwise in effect, the new breed of "java-code-shitter" (as free_electron termed it) will grow. i dont know i havent met one, but my senses tell me, they are growing out there... they dont have much sense what assembly codes look like (maybe only a few who can find their true self). my 2cnts.

ps: during my early kid... i just play with woods and sticks. until i'm old enough to remember then i met computer. todays kids... they play with touchscreen.

[Stephen Hill]

I personally see the RPi as a toy - nothing more. I have one and will only be using it as a cheap and compact media player for my bedroom TV using OpenElec/XBMC.

look at the bright side! its 700MHz and 256MB RAM. not like my 8MHz only few KB RAM ARM-M3 dev kit here. it should be able to do 100X more (guesstimate) than what mine can. but if you find it slower, then something wrong, probably a "monkey" programmed that thing

Not sure I get the jist of your reply. My point was that if you want to really learn how to program a processor then get yourself a proper development kit.

I think students would learn alot more with something like a PIC Starter Kit/Board than the RPi.

[Gall]

Or even just writing Pascal code on a PC.

99% of programming is just mathematics. Nothing else.

-- Professional software engineer, "Yandex"

[T4P]

O_o PIC that is, not ARM-M3, the lowest i've seen them go is 24MHz

its HY-Mini STM32V 1hunglo dev kit i got from ebay. it has 8MHz crystal connected to STM32F103VCT6. i dont know, havent looked at ARM architecture/datasheet in detail, maybe it has clock multiplier inside. anyway, if rasberry is a toy, then this 1hunglo is "toy of the toy".

Some i know bought this HY-Mini thing for 41$ and while using i find it pretty fast, the 8MHz is just RTC or the Watchdog i believe, as the STM32VLDisco has the 8MHz crystal too
the STM32F103VCT6 it has is no low performance bugger though, it's 72MHz
That explains why it can run the touchscreen and the built in OS

[T4P]

I personally see the RPi as a toy - nothing more. I have one and will only be using it as a cheap and compact media player for my bedroom TV using OpenElec/XBMC.

look at the bright side! its 700MHz and 256MB RAM. not like my 8MHz only few KB RAM ARM-M3 dev kit here. it should be able to do 100X more (guesstimate) than what mine can. but if you find it slower, then something wrong, probably a "monkey" programmed that thing

Not sure I get the jist of your reply. My point was that if you want to really learn how to program a processor then get yourself a proper development kit.

I think students would learn alot more with something like a PIC Starter Kit/Board than the RPi.

Such a board probably isn't meant for PIC starter kit or something along the lines
But anyway the PIC boards are too expensive, we need something even cheaper then the arduino
I wouldn't recommend the STM32F4Disco for a first time developer either, it's got too much performance to cheat in the coding resulting in horrible code (And that is my people, why apps these days are awful)
But at the same time i might be 18, i have written assembler for windows ... AKA Hacking a app with assembler

[Stephen Hill]

But anyway the PIC boards are too expensive, we need something even cheaper then the arduino

Your right about PIC development boards being expensive, but on the other hand if your serious about microcontroller development, 50 to 100 GBP for a development board is a worthwhile investment.

Thinking about it a bit more, I think the RPi probably isn't targeted at hardware development. It's focus will be application level software development.

[T4P]

A cheap one i can think of is PIC DM164120-2 and DM164120-1

The expensive one is this one, it's so expensive but yet it can't do much and it needs another 100$ to break out the I/O
http://i492.photobucket.com/albums/rr283/DarkShadower/_DSC0323.jpg

[Hypernova]

IMO if one is serious about getting into a MCU it's better to get a programmer and some DIP versions of it (or QFP100 on a breakout) and bread board it from the get go. It gets you into the habit of choosing for yourself how you'll connect the buttons and shit earlier and let you try out functions that aren't made unavailable because the guy to designed the dev board decided not to route it out or it got taken up by other buttons.

[Gall]

+1 for DIP.

THE easiest way to learn is to use something like ATmega8515 in DIP-40 on a solderless breadboard. One can easily probe each pin individually and change the circuit very quickly.

[nitro2k01]

The phillips has a 400Hz panel and it has those "Smart/Stupid TV things" and a 14bit video DAC

can anyone explain me what the purpose of a DAC is in a flatpanel TV ? especially since the pixels are digitally controlled... why do you need a dac ? marketing bullshittery ?

What am I missing here? What exactly does "digitally controlled" mean in this context? In the end, each pixel has an analog value, and this is what the digital signal will control. CGA was also digitally controlled, in the truest sense of the word. I.e. 1 bit per primary color, plus an intensity bit, 16 colors in total. I suspect modern flatpanel TVs have a slightly higher color depth than that, so they need a DAC of some kind.

[Gall]

This DAC is always integrated into TFT panel. The panel itself always has digital interface.

[T4P]

like HDMI, which is commonly used now

Is there really a qualititive advantage of a 14bit over a 8bit?
By the way, there's no definitive method this 14bit DAC accesses a something that has a larger bit resolution for the phillips screen i'm talking about here has a something that is next inline that has 30bit because of their claimed 14million something display colours
47PFL6007H/12 is the model i'm talking about, the 3D version with ambilight

But i believe it's not neccesary at all!

[nitro2k01]

DaveXRQ: Please use punctuation! I can't figure out where one sentence ends and the next one starts.
My original comment was for free_electron, who seemed to imply that there are no DACs to be found anywhere in an LCD panel.

Now, as to whether a 14 bit DAC makes any difference... It may not do for a TV, but I think it absolutely does for a high contrast monitor to be used in graphical design, for example. The consideration to make there is that you may want to calibrate the colors in the display. Even if not every one of 16384 values are actually being used, the values that come out after the gamma curve is applied, will be closer to the target value. And even if 14 bit may be overkill, it may be that it's so cheap to make a 14-bit video DAC there's no point in using 8 bit DACs anymore.

As for the panel in question, I can't seem to find any online source that actually says that it has 14 bit DACs, or anything at all about the video DACs. 

[free_electron]

My original comment was for free_electron, who seemed to imply that there are no DACs to be found anywhere in an LCD panel.

And that is correct. The LCD is driven using a stepped control voltage. A master step generator feeds the step voltages to the column drivers. The column driver selects a range in the step and potentially subdivides it. The steps are not linearly spaced but follow an S-curve tuned to the lcd material used. the light intensity follows a sinusoid behavior as it is a 90 degree rotation angle of the crystal that blocks the light. As temperature changes the step generator adapts the step voltages as well.

Once the base step voltage is switched in ( note that all these steps are a big bus with over a thousand independent , static , voltage levels. There is no DAC involved. ) using a multiplexer under control of the pixel data word the selected step voltage is loaded on a capacitor .

Depending on the phase of the backplane the cap is connected forward or reverse to the pixel. So when the backplane changes phase we reverse the polarity of the capacitor (lcd pixels need to toggle, they cant remain static....) there is a simple buffer opamp between cap and pixel line.

There are two capacitors. One holds the current 'value' of the pixel being applied while another is being loaded with the charge for the next line ( this example assumes the active drivers are column based ) . When the next line begins all caps have been charged and we do it all over again ..

No dacs have been harmed in driving the LCD panel. It's all pure digital.

Every column driver chip controls maybe 256 columns. For advanced panels that have matrixed backlighting that is controllable per tile ( whether white backlight or full spectrum RGB backlight ) they can actually reprogram the Step-curve to be in sync with the background RGB setting. This way they can 'boost' pixels and create larger dynamic range. The DCD technology from Faroudja is the absolute king in this. They now have full RGB backlight controllable per tile.

You could consider the step-generator and multiplexer a form of switched-cap DAC , but it really isn't .. the base principle is totally different. using a true dac per column driver would require a massive amount of silicon area , be a nightmare to trim them all identically, keep them tracking under temperature , and they would need fast settling times. the required logic would be immense... it's not feasible.

But, for the laymen the marketing bullshitters use the terminology 'DAC' because this is the highest level the 'layman' understands (they have been pavlovianly conditioned to drool when they hear technobabble) and 12 is higher than 10 so it must be good ... no ?
In reality it is nonsense.

[T4P]

I have no idea because i haven't torn a LCD down before, but when i get to i will most certainly reverse engineer it.
just for fun 

[nitro2k01]

free_electron: Ok. Thank you for the detailed description. I learned something new. Since you seem to know a thing or two about LCD technology, and while we're at it, maybe you have the answer to a thing I've observed. There seems to be a change in green bias every frame in small groups of pixels.

Code: [Select]

.X.X.X.X
.X.X.X.X
X.X.X.X.
X.X.X.X.
.X.X.X.X
.X.X.X.X
X.X.X.X.
X.X.X.X.X = white and . = black. The pattern needs to be aligned.
If you draw this pattern and invert the color of the pattern every frame, you get a very slight either green bias or magenta (anti-green) tint. Me and a friend of mine have observed this on two completely different panels. I'm thinking this may be some trick where the panel electronics are changing the green value by one or two, every frame, to increase green resolution by one bit. Does this phenomenon ring a bell with you?

[free_electron]

most lcd's use the bayer filter.
there is 2 green pixels in an RGB set. contrary to a classic picture tube where they work with triplets

Code: [Select]

B/G\R B/G\R B/
/R B\G/R B\G/

an lcd uses a quadruplet

Code: [Select]

RG|RG|RG|RG|
GB|GB|GB|GB|
RG|RG|RG|RG|
GB|GB|GB|GB|

so yes, the panel needs to compensate if you switch from whit to black...  as there are two greens switching they may be overdoing the compensation ...

[nitro2k01]

Neither of my panels (laptop, Sony LCD display) have the quadruplet subpixel layout. This is an image taken while drawing the pattern in question. You can't see the effect in question on a static image taken with a 1/500 s exposure time, but you can see that the pixels are laid out simply as RGB.

[T4P]

BTW, sharp claims a RGB+Y colour setup
Real or not?

[_Sin]

most lcd's use the bayer filter.

Most?