Author Topic: GigaDevice RISC-V microcontroller  (Read 17450 times)

0 Members and 1 Guest are viewing this topic.

Offline Pineapple DanTopic starter

  • Contributor
  • Posts: 45
  • Country: ie
GigaDevice RISC-V microcontroller
« on: April 14, 2021, 11:08:38 pm »
Anyone use them? What are they like to use?

Do you need a proprietary, sh1tty Chinese IDE to program them?

Any non-Chinese manufacturers making RISC-V uC yet? I really begrudge giving the CCP a penny more than I need to
 

Online brucehoult

  • Super Contributor
  • ***
  • Posts: 4000
  • Country: nz
Re: GigaDevice RISC-V microcontroller
« Reply #1 on: April 15, 2021, 12:15:24 am »
You can use your favourite gcc and openocd setup whether that's emacs/vim or VSCode or Eclipse with platformio or whatever.

You can buy them in the west (they ship from the EU) here:

https://www.tme.com/au/en/news/library-articles/page/42112/gigadevice-risc-v-microcontrollers/
 

Offline SiliconWizard

  • Super Contributor
  • ***
  • Posts: 14301
  • Country: fr
Re: GigaDevice RISC-V microcontroller
« Reply #2 on: April 15, 2021, 12:49:25 am »
I know a few western companies are working on RISC-V-based cores for future MCUs, but at the moment, I could not find ANY MCU available off-the-shelf that wasn't from China.
There are already a few from China, including from GD, Espressif (the new ESP32-C3), and a few others.

The only non-chinese company I know of that ships RISC-V MCUs at the moment is SiFive, with the FE310. There are a couple dev boards available, including from Sparkfun, but buying the bare MCUs can only be done from SiFive directly at the moment, at least I could not find them available from any of the usual distributors. Maybe Bruce knows more about this particular point.

If anyone knows of any other non-chinese vendor, let us know.
 

Online brucehoult

  • Super Contributor
  • ***
  • Posts: 4000
  • Country: nz
Re: GigaDevice RISC-V microcontroller
« Reply #3 on: April 15, 2021, 02:18:29 am »
Mouser have them, but not a lot.

https://www.mouser.com/ProductDetail/Crowd-Supply/cs-hifiveb-02?qs=Zz7%252BYVVL6bGRLEumsP1bRg%3D%3D

The FE310 is more or less a demo of the cores, to help get licensing customers, not a serious attempt to compete in the stand-alone microcontroller market.

ARM doesn't make and sell microcontroller chips and SiFive doesn't want to either.

I think it must be very hard to break into the stand-alone microcontroller market. There are tens of thousands of customers, none of whom buy very many, and ST, MicroChip, NXP, Renesas, TI have it sewn up.

I think I saw recently that RISC-V is now up to 15% of the market for microcontroller cores licensed to go inside some other chip. There are only ... hundreds (?) ... of such customers, and they make a lot of each chip.

NXP already dipped their toes in the water with the RV32M1 chip, with four CPU cores: one each of RISC-V RI5CY & Zero-RISCYand ARM Cortex M4F & M0+.

MicroChip of course has RISC-V hard cores in their PolarFire SoC FPGA chips.

ST right now has a job ad for a RISC-V microprocessor architect: https://euraxess.ec.europa.eu/jobs/613325


Perhaps weirdly, I think you'll be able to buy cheap Linux-capable RISC-V chips for under $10 before any of those. Allwinner just formally announced their D1, with a 1 GHz Alibaba C906 core. I've heard the initial production run is 5 million chips. Sipeed, Pine64, and Allwinner themselves have announced SBCs using this chip -- but obviously that won't use 5 million of them.

Of course all of the above are Chinese, so ...
 

Offline abraxalito

  • Newbie
  • Posts: 5
Re: GigaDevice RISC-V microcontroller
« Reply #4 on: April 15, 2021, 04:39:51 am »
Perhaps weirdly, I think you'll be able to buy cheap Linux-capable RISC-V chips for under $10 before any of those. Allwinner just formally announced their D1, with a 1 GHz Alibaba C906 core. I've heard the initial production run is 5 million chips. Sipeed, Pine64, and Allwinner themselves have announced SBCs using this chip -- but obviously that won't use 5 million of them.

https://www.cnx-software.com/2021/04/13/allwinner-d1-linux-risc-v-sbc-processor/
 
The following users thanked this post: soFPG, Pineapple Dan

Offline woofy

  • Frequent Contributor
  • **
  • Posts: 321
  • Country: gb
    • Woofys Place
Re: GigaDevice RISC-V microcontroller
« Reply #5 on: April 15, 2021, 08:34:09 am »
If you don't want to spend money on a dedicated risc v board you could get an fpga board and put one of the risc v cores on that. Clifford Wolf has the picoRV32 here: https://github.com/cliffordwolf/picorv32.
It won't have the perfomance of a dedicated chip but to learn risc v assembly its fine.

Offline mon2

  • Frequent Contributor
  • **
  • Posts: 462
  • Country: ca
Re: GigaDevice RISC-V microcontroller
« Reply #6 on: April 15, 2021, 10:20:16 am »
Hi. Have experimented a bit with the following:

https://www.efinixinc.com/products-riscv.html

using the XYLONI kit from Digikey.
 

Online brucehoult

  • Super Contributor
  • ***
  • Posts: 4000
  • Country: nz
Re: GigaDevice RISC-V microcontroller
« Reply #7 on: April 15, 2021, 10:52:25 am »
If you don't want to spend money on a dedicated risc v board you could get an fpga board and put one of the risc v cores on that. Clifford Wolf has the picoRV32 here: https://github.com/cliffordwolf/picorv32.
It won't have the perfomance of a dedicated chip but to learn risc v assembly its fine.

If you already have an FPGA board, yes.

But once the boards with these Allwiner chips hit you'll have a 1 GHz Linux 64 bit RISC-V board for a less money than an FPGA board that can run a soft core at 25 MIPS.

The cheapest route if you don't need to blink LEDs (etc) is to run the code on an emulator such as qemu, spike, or even my trivial little one-C-flle RV32I emulator at https://github.com/brucehoult/trv

Which I need to document and stuff...
 

Offline SiliconWizard

  • Super Contributor
  • ***
  • Posts: 14301
  • Country: fr
Re: GigaDevice RISC-V microcontroller
« Reply #8 on: April 15, 2021, 04:03:47 pm »
Mouser have them, but not a lot.

https://www.mouser.com/ProductDetail/Crowd-Supply/cs-hifiveb-02?qs=Zz7%252BYVVL6bGRLEumsP1bRg%3D%3D

Ah thanks. Very limited stock though.
Mouser appears to get them from CrowdSupply though, so not directly from SiFive.

The FE310 is more or less a demo of the cores, to help get licensing customers, not a serious attempt to compete in the stand-alone microcontroller market.

ARM doesn't make and sell microcontroller chips and SiFive doesn't want to either.

Yep.

I think it must be very hard to break into the stand-alone microcontroller market. There are tens of thousands of customers, none of whom buy very many, and ST, MicroChip, NXP, Renesas, TI have it sewn up.

Yeah, I think there's still very little incentive for western MCU vendors at the moment to switch to RISC-V.

Chinese companies, IMO, do this for the sake of becoming independent from ARM or any other western company. So there you have some, and it's growing.
« Last Edit: April 15, 2021, 04:11:59 pm by SiliconWizard »
 

Online dmendesf

  • Frequent Contributor
  • **
  • Posts: 318
  • Country: br
Re: GigaDevice RISC-V microcontroller
« Reply #9 on: April 15, 2021, 04:26:20 pm »
There's another problem... NVIDIA Aquisition of ARM. They're only after the high end space (to become independent of Intel and AMD) and will probably neglect the cortex-M. In the short time M4 / M7 / M33 are ok but in the long run to they'll need another horse...
 

Offline technix

  • Super Contributor
  • ***
  • Posts: 3507
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: GigaDevice RISC-V microcontroller
« Reply #10 on: April 27, 2021, 03:21:06 am »
There's another problem... NVIDIA Aquisition of ARM. They're only after the high end space (to become independent of Intel and AMD) and will probably neglect the cortex-M. In the short time M4 / M7 / M33 are ok but in the long run to they'll need another horse...
Well since ARMv8-A and Cortex-M are pretty divergent, I do see NVIDIA divesting the MCU part of ARM, assuming they can buy ARM at all. (That transaction still needs approval from UK, EU, US and China, and China has blocked the Qualcomm-NXP merger previously.)
 

Offline westfw

  • Super Contributor
  • ***
  • Posts: 4196
  • Country: us
Re: GigaDevice RISC-V microcontroller
« Reply #11 on: April 27, 2021, 05:07:16 am »
Quote
Perhaps weirdly, I think you'll be able to buy cheap Linux-capable RISC-V chips for under $10 before any of those.
Why "weirdly"?  Haven't all of the architectures (and/or ISAs) that have spanned "big" systems to "microcontroller" had the "big" choices show up first?  ARM, MIPS, PPC, 68k, x86, Z80...
The only architectures that have originated in the "small" space have pretty much STAYED small (and aren't really capable of becoming big.)

Maybe the "under $10" is new :-)  (and of course, have an ISA designed before chips is new.)
 

Offline technix

  • Super Contributor
  • ***
  • Posts: 3507
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: GigaDevice RISC-V microcontroller
« Reply #12 on: April 27, 2021, 07:44:41 am »
Why "weirdly"?  Haven't all of the architectures (and/or ISAs) that have spanned "big" systems to "microcontroller" had the "big" choices show up first?  ARM, MIPS, PPC, 68k, x86, Z80...
The only architectures that have originated in the "small" space have pretty much STAYED small (and aren't really capable of becoming big.)
I think ARM started off small and worked its way up... There was a time when ARM was only known as a low power embedded / microcontroller core (albeit a high end one) yet now we have ARM based servers. x86 never really worked as small cores.

The thing is, it is not where the core starts that matters, it is whether the design was scalable. Those processors that stayed small was not scalable.
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14076
  • Country: de
Re: GigaDevice RISC-V microcontroller
« Reply #13 on: April 27, 2021, 08:04:58 am »
The ISA is allways designed before th chip - it kind of has too unless one writes microcode later and starts with a kind of universal hardware.

The x86 have started small - kind of still 8080 compatible. The original 8086 was pretty small too and there was a 80186 µC variant too.

The ARM CPU did not start small - at the start they about competed with Motorola 68K and i80286 in desktop computers. It was 32 bits from the start.

If the GD chip takes over the periphery from there STM32 clone, they could have a very nice market. For the modern µCs the actual CPU core does not matter that much. If the periphery and pinout fit's it could be as little as to recompile the C code. Given the lead times for some ARM based µcs, there may be enough desparate customers.
Of cause GD could as well sell the ARM chips. The savings on licensing may not be that high, but for the cost sensitive markts even cents may count.
 

Offline technix

  • Super Contributor
  • ***
  • Posts: 3507
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: GigaDevice RISC-V microcontroller
« Reply #14 on: April 27, 2021, 08:11:32 am »
If the GD chip takes over the periphery from there STM32 clone, they could have a very nice market. For the modern µCs the actual CPU core does not matter that much. If the periphery and pinout fit's it could be as little as to recompile the C code. Given the lead times for some ARM based µcs, there may be enough desparate customers.
Of cause GD could as well sell the ARM chips. The savings on licensing may not be that high, but for the cost sensitive markts even cents may count.
The GD32VF103 line is pin-to-pin, spec-for-spec and peripheral-to-peripheral identical to their GD32F103 line except the CPU core. And GD32F103 is supposed to be a drop-in replacement for STM32F103 line.
 

Offline techman-001

  • Frequent Contributor
  • **
  • !
  • Posts: 748
  • Country: au
  • Electronics technician for the last 50 years
    • Mecrisp Stellaris Unofficial UserDoc
Re: GigaDevice RISC-V microcontroller
« Reply #15 on: April 27, 2021, 09:01:24 am »
If the GD chip takes over the periphery from there STM32 clone, they could have a very nice market. For the modern µCs the actual CPU core does not matter that much. If the periphery and pinout fit's it could be as little as to recompile the C code. Given the lead times for some ARM based µcs, there may be enough desparate customers.
Of cause GD could as well sell the ARM chips. The savings on licensing may not be that high, but for the cost sensitive markts even cents may count.
The GD32VF103 line is pin-to-pin, spec-for-spec and peripheral-to-peripheral identical to their GD32F103 line except the CPU core. And GD32F103 is supposed to be a drop-in replacement for STM32F103 line.

Politicians are 'supposed' to be honest, flying cars were 'supposed' to be the norm by the 1980's. Those claims weren't true, and neither is yours.

Please point to a *English* peer reviewed on-line article detailing what is 100% identical, what isn't and why between the GD32F103 and the STM32F103?

For starters, I'd like to know how the GD use of a stacked and glued SPI mirror ram in the place of on-mcu flash makes it the *same*.

It may make it better, it may make it worse, (depending what you want to do) but it won't make it a 'drop-in replacement for STM32F103 line'.

The same goes for the several other counterfit STM32F103 chips made in China by CKS, APM etc.
 

Offline tszaboo

  • Super Contributor
  • ***
  • Posts: 7307
  • Country: nl
  • Current job: ATEX product design
Re: GigaDevice RISC-V microcontroller
« Reply #16 on: April 27, 2021, 09:12:31 am »
Why "weirdly"?  Haven't all of the architectures (and/or ISAs) that have spanned "big" systems to "microcontroller" had the "big" choices show up first?  ARM, MIPS, PPC, 68k, x86, Z80...
The only architectures that have originated in the "small" space have pretty much STAYED small (and aren't really capable of becoming big.)
I think ARM started off small and worked its way up... There was a time when ARM was only known as a low power embedded / microcontroller core (albeit a high end one) yet now we have ARM based servers. x86 never really worked as small cores.

The thing is, it is not where the core starts that matters, it is whether the design was scalable. Those processors that stayed small was not scalable.
No, ARM started as Acorn RISC machine, a personal computer. After the initial failure, they had a great success for PDAs running at around 400MHz, which was blazing fast at a time. The diversification to Application\Real-time\Microcontroller came later.
 
The following users thanked this post: newbrain

Offline techman-001

  • Frequent Contributor
  • **
  • !
  • Posts: 748
  • Country: au
  • Electronics technician for the last 50 years
    • Mecrisp Stellaris Unofficial UserDoc
Re: GigaDevice RISC-V microcontroller
« Reply #17 on: April 27, 2021, 09:22:16 am »
For starters, I'd like to know how the GD use of a stacked and glued SPI mirror ram in the place of on-mcu flash makes it the *same*.

GD's intended customers don't care about real time determinism or anything fancy. They want a cheap replacement to use a fraction of capability and function of an STM32.

Most people don't use all potency of a chip. If it gets the job done on this particular case and saves money, it is a green.

The Chinese standard of compatible is simple. It functionally behaves the same, and runs at the same clock. Who cares if it is cycle accurate or not, or if it works at all in corner cases.

The fact that so many GD32F103's ended up re branded as STM32F103 in Blue Pills shows that someone in China cared.

The fact that CKS use STM only ID's in their chips shows they cared.

Do you think that Western buyers knowing that the Chinese chips and products sold internationally are made by people for whom 'near enough' is 'close enough', will actually buy any of your products ?

Personally I'd rather you reply had been, "Yes its a big problem here, we shoot all the counterfeiters we can find, and we are working tirelessly to make better chips, cheaper than the western ones, can I send you some free samples of our new mcu's and a 5000 page detailed technical manual PDFs in English ?"
 

Offline technix

  • Super Contributor
  • ***
  • Posts: 3507
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: GigaDevice RISC-V microcontroller
« Reply #18 on: April 27, 2021, 09:33:55 am »
Politicians are 'supposed' to be honest, flying cars were 'supposed' to be the norm by the 1980's. Those claims weren't true, and neither is yours.
Those are *their* claim, not mine.

Please point to a *English* peer reviewed on-line article detailing what is 100% identical, what isn't and why between the GD32F103 and the STM32F103?
1) They can not make it actually 100% identical otherwise they will be sued into oblivion. Using a built-in QSPI controller and a bonded SPI Flash is enough of a difference to fend off those lawsuits.
2) There are a lot of opinions on Chinese communities urging folks like GD to entirely drop their English documentations, even on the basis of just irking the Five Eyes citizens. Should those articles exist, they will be in Chinese, and it will have to be YOU learning Chinese in order to read those, since absolutely nobody are permitted to translate it without either getting sued or catch the absolute mother load of flak.
3) You want peer-reviewed and online article... I am sensing some internal conflicts.

For starters, I'd like to know how the GD use of a stacked and glued SPI mirror ram in the place of on-mcu flash makes it the *same*.

It may make it better, it may make it worse, (depending what you want to do) but it won't make it a 'drop-in replacement for STM32F103 line'.
You throughly underestimate what folks at GD can pull off.

The same goes for the several other counterfit STM32F103 chips made in China by CKS, APM etc.
Imitation and compatible designs =/= counterfeit. The ST legal department has learned that a long time ago, otherwise they would have sued.

The fact that so many GD32F103's ended up re branded as STM32F103 in Blue Pills shows that someone in China cared.
When did Blue Pill became that strongly associated with STM32F103? There has always been multiple versions of that - STM32F103, GD32F103, etc.

The fact that CKS use STM only ID's in their chips shows they cared.
About what?

Do you think that Western buyers knowing that the Chinese chips and products sold internationally are made by people for whom 'near enough' is 'close enough', will actually buy any of your products ?
AFAIK you are still buying. If you have some serious applications that only demands the best of the best, even if I have a project that demands the best of the best, I will take care not to use those replacement chips. However how many of those products actually needs that level of quality to begin with?

Personally I'd rather you reply had been, "Yes its a big problem here, we shoot all the counterfeiters we can find, and we are working tirelessly to make better chips, cheaper than the western ones, can I send you some free samples of our new mcu's and a 5000 page detailed technical manual PDFs in English ?"
Sorry buddy, you will only be getting Chinese documentations down the road, so start learning the world's lingua franca of the future before it is too late. Also unless there is a patent on chip pinouts (which is not patentable to begin with) there will always be compatible and imitations, which is always fully legal worldwide.
 

Online brucehoult

  • Super Contributor
  • ***
  • Posts: 4000
  • Country: nz
Re: GigaDevice RISC-V microcontroller
« Reply #19 on: April 27, 2021, 09:39:54 am »
Why "weirdly"?  Haven't all of the architectures (and/or ISAs) that have spanned "big" systems to "microcontroller" had the "big" choices show up first?  ARM, MIPS, PPC, 68k, x86, Z80...
The only architectures that have originated in the "small" space have pretty much STAYED small (and aren't really capable of becoming big.)
I think ARM started off small and worked its way up... There was a time when ARM was only known as a low power embedded / microcontroller core (albeit a high end one) yet now we have ARM based servers. x86 never really worked as small cores.

You must be young :-)

ARM's first CPU was used in PCs that were faster than the 80386 PCs and 68030 Macs of the day, at a lower price. It was called "Acorn Archimedes".

The second big use was Apple Newton and Psion PDAs. Then mobile phones.

Microcontrollers came much later.
 

Offline techman-001

  • Frequent Contributor
  • **
  • !
  • Posts: 748
  • Country: au
  • Electronics technician for the last 50 years
    • Mecrisp Stellaris Unofficial UserDoc
Re: GigaDevice RISC-V microcontroller
« Reply #20 on: April 27, 2021, 09:42:48 am »
I have my own MCU designed and made in 180nm process, tailored for SMPS applications, and it has a long TRM/DS combo in very readable English (I wrote it all by myself).

That being said, the chip is not for sale, as it contains a proprietary DSP core and I did not bother to write a compiler for it, so for now the DSP part is assembly-only, and the CPU part is an RV32E with some special instructions for accelerated interrupt handling, so it also needs some special care on the compiler side.

If I do release it, the customers will have native English document (the documents were never written in Chinese in the first place), and will have access to all convenient tools and source code level library frameworks.

But I'm not interested in this STM clone business. I'll leave that to less innovative players preferring quantity over quality. When I can put this chip in a power module that rivals Vicor's technology and sells at a comparable price, why would I sell this chip for silicon cost, fighting a myriad of cheap MCU makers?

Awesome, you are a innovator, I wish you every success with your chip :)
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14076
  • Country: de
Re: GigaDevice RISC-V microcontroller
« Reply #21 on: April 27, 2021, 09:46:13 am »
Having a seprate flash chip and use of shaddow RAM makes the CPU part behave different.  Even if the µCs are are not 100% compatible, they are so similar that porting the software from one chip to the counterpart would in most cases be relatively easy. With the difference CPU one would need to recompile at least anyway and not expect identical execution times for all details.

It is of cause bad practice to sell chips with a wrong lable on a test board, so the users don't knows and may have problems even with minor differences. It is something different for a manufacturers who uses a fixed program and knows which chips he gets for the product.
It is also not just porting a product from one chip to a replacement with better availablity. It also help with the learing curve - once you know the STM32F...   moving to a similar GD32V...  can be relatively easy. So new products may start with the new chip instead - there is always the route back, just in case.

 

Offline technix

  • Super Contributor
  • ***
  • Posts: 3507
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: GigaDevice RISC-V microcontroller
« Reply #22 on: April 27, 2021, 09:48:02 am »
I'm so fed up with all those xenophobia certain members here want to bring up when talking about Chinese brands. Yes Chinese chips aren't the most innovative thing in the world and yes they don't have the best possible quality, however they do fill a market niche most Western players has long neglected: low price low end markets where a generations-old chip will work but the chip maker want to move on.

As of those "counterfeits", as long as the chip is marked with its true part number and manufacturer logo, it is not a fake chip. One can not patent a chip ID integer or a chip pinout, even a part number, so as long as the internal implementation is not a mask-for-mask copy, as long as the HDL is not a module-for-module copy, and all IP cores used are either properly licensed or clean-room reimplementations, those chips are fully legal compatible parts. Just ask ST why they can not sue GD, their legal department has all documents ready yet they can not file the suit.
 

Offline technix

  • Super Contributor
  • ***
  • Posts: 3507
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: GigaDevice RISC-V microcontroller
« Reply #23 on: April 27, 2021, 09:52:38 am »
You must be young :-)

ARM's first CPU was used in PCs that were faster than the 80386 PCs and 68030 Macs of the day, at a lower price. It was called "Acorn Archimedes".
I know that, but after that one hurrah they don't really have much impact on the notebook and up market until Huawei's ARM based server chips and Apple M1 came along. Even Surface RT flopped.

The second big use was Apple Newton and Psion PDAs. Then mobile phones.
Those are the upper edge of what I would consider small - tables and below.

Microcontrollers came much later.
Umm LPC2103? AT91SAM7?
 

Offline technix

  • Super Contributor
  • ***
  • Posts: 3507
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: GigaDevice RISC-V microcontroller
« Reply #24 on: April 27, 2021, 10:00:50 am »
Having a seprate flash chip and use of shaddow RAM makes the CPU part behave different.  Even if the µCs are are not 100% compatible, they are so similar that porting the software from one chip to the counterpart would in most cases be relatively easy. With the difference CPU one would need to recompile at least anyway and not expect identical execution times for all details.
AFAIK GD chips have a dedicated DMA engine that copies the QSPI contents to its program cache RAM upon boot, so it will have a slow boot due to cache misses, but soon after that it will be able to match ST's embedded Flash design clock-for-clock. Since GD made their chips register-compatible with ST, as long as your application don't have critical boot-time timing requirement, it is usually binary compatible.

It is of cause bad practice to sell chips with a wrong lable on a test board, so the users don't knows and may have problems even with minor differences. It is something different for a manufacturers who uses a fixed program and knows which chips he gets for the product.
That is what I was saying. whatever product you may be selling, be it Blue Pill or others, just make sure you are honest with the markings on the chip. I am perfectly okay to receive a GD32F103 Blue Pill as long as it is clearly marketed and marked as such.

It is also not just porting a product from one chip to a replacement with better availablity. It also help with the learing curve - once you know the STM32F...   moving to a similar GD32V...  can be relatively easy. So new products may start with the new chip instead - there is always the route back, just in case.
And since GD32VF103 used the similar QSPI MCP, using GD32F103 as a stepping stone may be a good idea too.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf