Joe Schmoe
Posts: 4277
Joined: Sun Jan 15, 2012 1:11 pm

Re: Cubieboard vs RaspberryPi B

Sat Apr 06, 2013 9:17 pm

You guys need to get a room.
And some folks need to stop being fanboys and see the forest behind the trees.

(One of the best lines I've seen on this board lately)

pygmy_giant
Posts: 1562
Joined: Sun Mar 04, 2012 12:49 am

Re: Cubieboard vs RaspberryPi B

Sat Apr 06, 2013 9:33 pm

* unsuccessfully stifled laugh resulting in uncomfortable nasal snort *

I confused the cubie board with the odroid in my previous post because the heatsink on the odroid makes it look 'cubie' shaped:

Image
http://www.bit-tech.net/news/hardware/2 ... odroid-2/1

I suspect that there will be an enevitable trend towards computers getting smaller and cheaper paralell to that of computers getting faster and more expensive.

W. H. Heydt
Posts: 11097
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: Cubieboard vs RaspberryPi B

Sat Apr 06, 2013 10:27 pm

pygmy_giant wrote:* unsuccessfully stifled laugh resulting in uncomfortable nasal snort *

I confused the cubie board with the odroid in my previous post because the heatsink on the odroid makes it look 'cubie' shaped:

http://www.bit-tech.net/news/hardware/2 ... odroid-2/1
I can see why your confusion. Interesting article. It's also interesting that the Odriod boards require a heatsink, which *neither* the Cubie nor the Pi require. Also note that, even without the need to provide power for an attached SATA device, the Odroid requires a 10W power supply. Makes me wonder if we're going to see a succession of larger and larger capacity power-block-with-USB connectors come on the market as companies attempt to get into the "tiny SBC" market with increasingly more power hungry processors. (One almost shudders to imagine the 20W power block for an SBC using an Intel chip...) Also note the prices on the Odroid machines ($70/$135).
I suspect that there will be an enevitable trend towards computers getting smaller and cheaper paralell to that of computers getting faster and more expensive.
On the full desktop, the trend has been to faster and *cheaper* in the mainstream. There is always the custom/semi-custom market at the top, but even there the price trends are down, not up.

What I would see in the Pi world is a two way split. On the one hand, there will be an effort to increase flexbility while holding a price point. That flexibility may be in the form of processor capability (number of cores and clock frequency) or it may be in the form of increased access options (more USB ports, SATA--or future equivalent). The other fork will be towards sheer power, with less regard to price (such as the referenced Odroids) and less regard for additional peripheral access.

On both forks, I think there will be a concern with power. On the Pis approach, that will suggest a move to better power regulators, which could be possible due to scale of manufacturing (much easier to afford in the bill of materials when you're ordering in lots of 100K than when ordering in lots of 10K). On the Odroids, they may have to refrain from adding higher power ports, like SATA.

The Pi will pretty clearly go down the road of holding the price point and only "upgrading" as the price of better components comes down.

Hard to say which approach the Cubie will take. I'd tend to bet that they'll try to move to a multicore processor next, but that'll give them some power budget issues, I suspect. Alternatively, they might go to a faster single core processor. In either case, I don't think they can let the final price get away from them. At $50, they are competitive with the Pi in *some* applications, but by no means all such.

rvalles
Posts: 25
Joined: Tue Jun 12, 2012 3:51 pm

Re: Cubieboard vs RaspberryPi B

Sun Apr 07, 2013 3:42 am

The problem I'm aware of with the current odroid is it uses Exynos4. That's the one with the major security hole used in the Samsung Galaxy S3, if I'm not wrong. It's also Cortex-A9, which is a poor choice right now that there's A15 and A7 out. The A7 is basically about as fast as the A9 per clock but using just a fraction of the power (even less than the A8).

There's some cheap SoCs based on the Cortex-A7, like the "Allwinner A20|A35" (which succeed the Allwinner A10 used in the Cubieboard) but I'm sadly not yet aware of any devices using them. While the "A20", dual core, uses an arm mali GPU (with a promising ongoing free driver effort from reverse engineering), the A35, quad core, uses a PowerVR (which there's nothing for) instead. Interestingly, the A20 is "pin-compatible" with the A10 (Cortex-A8 based), meaning that devices that actually use the A10 (like the cubieboard) could easily be upgraded to A20 without much design effort.

I'm looking forward to seeing those in the market (will definitely grab something A20-based at some point), as conveniently the linux-sunxi people are succeeding at getting support for these chips merged into upstream Linux, making them very nice to work with.

As for the Raspberry Pi, I suspect they'll continue holding the price and improving components as they get cheap enough to make it possible, like they've done already with the RAM. Broadcom has recently licensed some ARMv7/ARMv8 stuff, meaning that the current SoC being replaced with something newer is feasible. However, for compatibility reasons, they're probably not going to bother for a few years. Once they do, it's probably gonna be Cortex-A53 or a derivative of that.

W. H. Heydt
Posts: 11097
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: Cubieboard vs RaspberryPi B

Sun Apr 07, 2013 5:50 am

rvalles wrote:The problem I'm aware of with the current odroid is it uses Exynos4. That's the one with the major security hole used in the Samsung Galaxy S3, if I'm not wrong. It's also Cortex-A9, which is a poor choice right now that there's A15 and A7 out. The A7 is basically about as fast as the A9 per clock but using just a fraction of the power (even less than the A8).
Is the security hole a software problem or a hardware problem...and if the later, can it be fixed in software?
There's some cheap SoCs based on the Cortex-A7, like the "Allwinner A20|A35" (which succeed the Allwinner A10 used in the Cubieboard) but I'm sadly not yet aware of any devices using them. While the "A20", dual core, uses an arm mali GPU (with a promising ongoing free driver effort from reverse engineering), the A35, quad core, uses a PowerVR (which there's nothing for) instead. Interestingly, the A20 is "pin-compatible" with the A10 (Cortex-A8 based), meaning that devices that actually use the A10 (like the cubieboard) could easily be upgraded to A20 without much design effort.
Yabbut....are they cheap enough to go into a less than $50 single board computer and retain the interfaces that the Cubie has? (This really strikes to the reason the Pi is using what is now a relatively elderly chip...the processes are so well established, and enough have been sold to earn back the development cost that they can be sold very cheaply. Going for a new, let alone "cutting edge" chip *isn't* going to result in a very low cost system, and that is the characteristic of the small, single board systems. One might as well go mooning after a $50 Ivy Bridge system....just ain't gonna happen.)
As for the Raspberry Pi, I suspect they'll continue holding the price and improving components as they get cheap enough to make it possible, like they've done already with the RAM. Broadcom has recently licensed some ARMv7/ARMv8 stuff, meaning that the current SoC being replaced with something newer is feasible. However, for compatibility reasons, they're probably not going to bother for a few years. Once they do, it's probably gonna be Cortex-A53 or a derivative of that.
*If* a newer pin-compatible SoC were to be made, *then* as prices came down, it could become a new version of the Pi. However, since there is no indication that Broadcom has made any other chip that could just be dropped in, and every indication that they never will, one would be looking at a completely new board design to make a version of the Pi with another chip.

A completely new board, while a lot of work and no end of headaches is not, on the whole a bad thing in and of itself. Such a board could incorporate lessons learned from the Pi....all those things where the designers say to themselves, "if only I'd done THAT instead of THIS..." But it's still a big undertaking.

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 24162
Joined: Sat Jul 30, 2011 7:41 pm

Re: Cubieboard vs RaspberryPi B

Sun Apr 07, 2013 7:51 am

W. H. Heydt wrote: *If* a newer pin-compatible SoC were to be made, *then* as prices came down, it could become a new version of the Pi. However, since there is no indication that Broadcom has made any other chip that could just be dropped in, and every indication that they never will, one would be looking at a completely new board design to make a version of the Pi with another chip.

A completely new board, while a lot of work and no end of headaches is not, on the whole a bad thing in and of itself. Such a board could incorporate lessons learned from the Pi....all those things where the designers say to themselves, "if only I'd done THAT instead of THIS..." But it's still a big undertaking.
I doubt that Brcm would make a pin compatible chip to the 2835. Just no point. It's pretty much as easy just to redesign a board to take whatever chip they make. Just as an example, a very popular mobile phone maker can knock out a new PCB for a phone AND get it manufactured AND working (and they are very complex devices) in a few weeks. With that sort of turnaround time there's no point in pin compatibility.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I think it’s wrong that only one company makes the game Monopoly.” – Steven Wright

pygmy_giant
Posts: 1562
Joined: Sun Mar 04, 2012 12:49 am

Re: Cubieboard vs RaspberryPi B

Sun Apr 07, 2013 3:28 pm

That's impressive!

Growing a comunity and software/user base takes longer.

In relation to an apples vs oranges / chalk vs cheese grudgematch smackdown I guess you have to ask 'better at what'. There are different ways of evaluating a board - you can look at how much bang you get for your buck or you can look at bang per gramme or bang per mA; the physical laws of the universe require trade-offs between these. For my money/purposes the Pi is the best all-rounder and has the most acessible GPIO and best add-on/software support. I also suspect that the Model A is the best in the bang per mA category.

The only proper way to settle this dispute though would be through trial by combat, which would involve soaking all the boards in vinegar and threading them on string.

Personally though I enjoy seeing a diversity of products and look forward to advances in technology.

I suposed like all hardware the Pi is destined to eventually become out of date but because it has tapped into an open source software ecosystem, the hardware is not really the Pi's most significant feature.

W. H. Heydt
Posts: 11097
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: Cubieboard vs RaspberryPi B

Sun Apr 07, 2013 9:08 pm

pygmy_giant wrote:That's impressive!

Growing a comunity and software/user base takes longer.

In relation to an apples vs oranges / chalk vs cheese grudgematch smackdown I guess you have to ask 'better at what'. There are different ways of evaluating a board - you can look at how much bang you get for your buck or you can look at bang per gramme or bang per mA; the physical laws of the universe require trade-offs between these. For my money/purposes the Pi is the best all-rounder and has the most acessible GPIO and best add-on/software support. I also suspect that the Model A is the best in the bang per mA category.

The only proper way to settle this dispute though would be through trial by combat, which would involve soaking all the boards in vinegar and threading them on string.

Personally though I enjoy seeing a diversity of products and look forward to advances in technology.

I suposed like all hardware the Pi is destined to eventually become out of date but because it has tapped into an open source software ecosystem, the hardware is not really the Pi's most significant feature.
I agree with you in general. When asking "which board is best?" the proper reply is "best at *what*?"

While the Foundation certainly didn't invent the single board computer, small form factor (SFF) computers, or really cheap computers, they certainly made the combination of very small size, very low cost, and (reasonably) complete single board well known and--inadvertently--very popular.

Has anyone else managed to get a tenth as many general purpose small, cheap single board systems on the market? I certainly don't know of any. For that alone the Foundations deserves notice and accolades.

It is somewhat ironic that other boards are now using the software that Pi developers have sweated getting working well, which is another sign of the success of the Pi.

pygmy_giant
Posts: 1562
Joined: Sun Mar 04, 2012 12:49 am

Re: Cubieboard vs RaspberryPi B

Sun Apr 07, 2013 9:21 pm

Android would open up more horizons on the Pi - I wonder how thats going....?

http://www.raspberrypi.org/phpBB3/viewt ... 73&t=38158

I guess the Pi is also champeen in the most OSeses category for good or for ill.

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 24162
Joined: Sat Jul 30, 2011 7:41 pm

Re: Cubieboard vs RaspberryPi B

Sun Apr 07, 2013 10:23 pm

pygmy_giant wrote:Android would open up more horizons on the Pi - I wonder how thats going....?

http://www.raspberrypi.org/phpBB3/viewt ... 73&t=38158

I guess the Pi is also champeen in the most OSeses category for good or for ill.
I personally see no relevance in having Android on the Raspi. It's a great consumer OS, but bugger all use in proper teaching, which of course is the Raspi's purpose.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I think it’s wrong that only one company makes the game Monopoly.” – Steven Wright

Joe Schmoe
Posts: 4277
Joined: Sun Jan 15, 2012 1:11 pm

Re: Cubieboard vs RaspberryPi B

Sun Apr 07, 2013 10:29 pm

jamesh wrote:
pygmy_giant wrote:Android would open up more horizons on the Pi - I wonder how thats going....?

http://www.raspberrypi.org/phpBB3/viewt ... 73&t=38158

I guess the Pi is also champeen in the most OSeses category for good or for ill.
I personally see no relevance in having Android on the Raspi. It's a great consumer OS, but bugger all use in proper teaching, which of course is the Raspi's purpose.
Roger that!

(in the military sense)
And some folks need to stop being fanboys and see the forest behind the trees.

(One of the best lines I've seen on this board lately)

ddv2005
Posts: 23
Joined: Fri Jul 20, 2012 2:17 am

Re: Cubieboard vs RaspberryPi B

Mon Apr 08, 2013 5:51 pm

jamesh wrote: Sounds like you don't have Gordon's FIQ fix which fixes the split transactions (this is currently still in testing).
Sounds like you don't test this patch because it crashes on USB audio. But any way FIQ fix can't fix the split transactions. The problem not only with scheduling. Several months ago I found that in USB debug mode split transactions work much more better than in release mode. I found that it is because of very small delay (about 1-5 usec) in hc_start_transactions. I did this small delay in release mode and it work much better. I assumed (because of no documentation at all) that it is because sometimes next transaction has more priority and controller sends it to the USB bus before split transaction...this results that the split transaction cannot be completed in same frame. I assumed again that transaction's priority can depends of host channel number and I found that If driver schedule split transactions on the highest host channel then it works more stable.

Code: Select all

diff --git a/drivers/usb/host/dwc_otg/dwc_otg_hcd.c b/drivers/usb/host/dwc_otg/dwc_otg_hcd.c
index 91eefec..b0a089b 100644
--- a/drivers/usb/host/dwc_otg/dwc_otg_hcd.c
+++ b/drivers/usb/host/dwc_otg/dwc_otg_hcd.c
@@ -1045,7 +1045,7 @@ static void dwc_otg_hcd_reinit(dwc_otg_hcd_t * hcd)
  */
 static void assign_and_init_hc(dwc_otg_hcd_t * hcd, dwc_otg_qh_t * qh)
 {
-	dwc_hc_t *hc;
+	dwc_hc_t *hc,*hc_c;
 	dwc_otg_qtd_t *qtd;
 	dwc_otg_hcd_urb_t *urb;
 	void* ptr = NULL;
@@ -1061,6 +1061,22 @@ static void assign_and_init_hc(dwc_otg_hcd_t * hcd, dwc_otg_qh_t * qh)
 
 
 	hc = DWC_CIRCLEQ_FIRST(&hcd->free_hc_list);
+	if(qh->do_split)
+	{
+		DWC_CIRCLEQ_FOREACH(hc_c,&hcd->free_hc_list,hc_list_entry)
+		{
+			if(hc_c->hc_num>hc->hc_num)
+				hc = hc_c;
+		}
+	}
+	else
+	{
+		DWC_CIRCLEQ_FOREACH(hc_c,&hcd->free_hc_list,hc_list_entry)
+		{
+			if(hc_c->hc_num<hc->hc_num)
+				hc = hc_c;
+		}
+	}
 
 	/* Remove the host channel from the free list. */
 	DWC_CIRCLEQ_REMOVE_INIT(&hcd->free_hc_list, hc, hc_list_entry);
After that patch usb packets lost less than 0.1% (3.2 % before). But I still hear audio clicks that mean that DEVICE still drop the packets from Pi.
I said it half year ago and I say it now that to try to fix the problem we need full USB documentation and very expensive hardware USB analyzer. But still no any documentation.

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 24162
Joined: Sat Jul 30, 2011 7:41 pm

Re: Cubieboard vs RaspberryPi B

Mon Apr 08, 2013 6:19 pm

ddv2005 wrote:
jamesh wrote: Sounds like you don't have Gordon's FIQ fix which fixes the split transactions (this is currently still in testing).
Sounds like you don't test this patch because it crashes on USB audio. But any way FIQ fix can't fix the split transactions. The problem not only with scheduling. Several months ago I found that in USB debug mode split transactions work much more better than in release mode. I found that it is because of very small delay (about 1-5 usec) in hc_start_transactions. I did this small delay in release mode and it work much better. I assumed (because of no documentation at all) that it is because sometimes next transaction has more priority and controller sends it to the USB bus before split transaction...this results that the split transaction cannot be completed in same frame. I assumed again that transaction's priority can depends of host channel number and I found that If driver schedule split transactions on the highest host channel then it works more stable.

Code: Select all

diff --git a/drivers/usb/host/dwc_otg/dwc_otg_hcd.c b/drivers/usb/host/dwc_otg/dwc_otg_hcd.c
index 91eefec..b0a089b 100644
--- a/drivers/usb/host/dwc_otg/dwc_otg_hcd.c
+++ b/drivers/usb/host/dwc_otg/dwc_otg_hcd.c
@@ -1045,7 +1045,7 @@ static void dwc_otg_hcd_reinit(dwc_otg_hcd_t * hcd)
  */
 static void assign_and_init_hc(dwc_otg_hcd_t * hcd, dwc_otg_qh_t * qh)
 {
-	dwc_hc_t *hc;
+	dwc_hc_t *hc,*hc_c;
 	dwc_otg_qtd_t *qtd;
 	dwc_otg_hcd_urb_t *urb;
 	void* ptr = NULL;
@@ -1061,6 +1061,22 @@ static void assign_and_init_hc(dwc_otg_hcd_t * hcd, dwc_otg_qh_t * qh)
 
 
 	hc = DWC_CIRCLEQ_FIRST(&hcd->free_hc_list);
+	if(qh->do_split)
+	{
+		DWC_CIRCLEQ_FOREACH(hc_c,&hcd->free_hc_list,hc_list_entry)
+		{
+			if(hc_c->hc_num>hc->hc_num)
+				hc = hc_c;
+		}
+	}
+	else
+	{
+		DWC_CIRCLEQ_FOREACH(hc_c,&hcd->free_hc_list,hc_list_entry)
+		{
+			if(hc_c->hc_num<hc->hc_num)
+				hc = hc_c;
+		}
+	}
 
 	/* Remove the host channel from the free list. */
 	DWC_CIRCLEQ_REMOVE_INIT(&hcd->free_hc_list, hc, hc_list_entry);
After that patch usb packets lost less than 0.1% (3.2 % before). But I still hear audio clicks that mean that DEVICE still drop the packets from Pi.
I said it half year ago and I say it now that to try to fix the problem we need full USB documentation and very expensive hardware USB analyzer. But still no any documentation.
The FIQ patch is still in alpha testing - it's not surprising there are still issues. As for full USB documentation and a hardware USB analyser - Gordon has/had both of those (he also has the RTL for the Synopsis hardware). He doesn't have to make the assumptions you have had to make in your analysis.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I think it’s wrong that only one company makes the game Monopoly.” – Steven Wright

pygmy_giant
Posts: 1562
Joined: Sun Mar 04, 2012 12:49 am

Re: Cubieboard vs RaspberryPi B

Mon Apr 08, 2013 7:26 pm

Jamesh said:

I personally see no relevance in having Android on the Raspi. It's a great consumer OS, but bugger all use in proper teaching, which of course is the Raspi's purpose.
True - but as it is a great consumer OS and popular it could generate more sales and 'mindshare' which would provide more revenue to directed at charitable aims.

I wonder what % of Pi's in the wild are currently used for education and what % are used for something else? Both have provided the Foundation with revenue.

I mention it here as 'runs android' seems to be promoted as a main feature by those selling other mini ARM computers like the Rikomagic jobby.

Cant believe I said 'mindshare'

Good luck with the FIQ fix.

User avatar
johnbeetem
Posts: 945
Joined: Mon Oct 17, 2011 11:18 pm
Location: The Mountains
Contact: Website

Re: Cubieboard vs RaspberryPi B

Mon Apr 08, 2013 8:34 pm

W. H. Heydt wrote:Has anyone else managed to get a tenth as many general purpose small, cheap single board systems on the market? I certainly don't know of any. For that alone the Foundations deserves notice and accolades.
I believe BeagleBoards have sold approx 100K, which meets your one-tenth. Granted it took them far longer to sell 100K than RasPi's 1M, IMO mostly because Beagle was marketed primarily as an embedded development platform for high-end TI ARMs rather than as "a GNU/Linux box". Another reason RasPi has been more successful is that IMO it's a lot easier to figure out which RasPi OS to run and from whence to download it.

My nominee for pioneer cheap motherboard-on-a-chip development board is the one Cirrus Logic made for their EP9315 ARM9 SoC. I remember the board being introduced for US$99. It was obviously too far ahead of its time -- bad luck for Cirrus.
Last edited by johnbeetem on Mon Apr 08, 2013 8:36 pm, edited 1 time in total.

pygmy_giant
Posts: 1562
Joined: Sun Mar 04, 2012 12:49 am

Re: Cubieboard vs RaspberryPi B

Mon Apr 08, 2013 9:00 pm

Didn't realise how man there are: http://www.armkits.com/Product/sbc2416g.asp

I could not find the one you mention but there are quite a few on the link above - some at $99

Makes the Pi's success all the more remarkable

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 24162
Joined: Sat Jul 30, 2011 7:41 pm

Re: Cubieboard vs RaspberryPi B

Mon Apr 08, 2013 10:16 pm

pygmy_giant wrote:
Jamesh said:

I personally see no relevance in having Android on the Raspi. It's a great consumer OS, but bugger all use in proper teaching, which of course is the Raspi's purpose.
True - but as it is a great consumer OS and popular it could generate more sales and 'mindshare' which would provide more revenue to directed at charitable aims.

I wonder what % of Pi's in the wild are currently used for education and what % are used for something else? Both have provided the Foundation with revenue.

I mention it here as 'runs android' seems to be promoted as a main feature by those selling other mini ARM computers like the Rikomagic jobby.

Cant believe I said 'mindshare'

Good luck with the FIQ fix.
But, should you spend money on trying to get Android working well, in order to sell more Raspi's, in order to make more cash for educational aims.

Or just spend the cash on educational aims. (and getting Linux working really really well, which sort of goes hand in hand)
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I think it’s wrong that only one company makes the game Monopoly.” – Steven Wright

pygmy_giant
Posts: 1562
Joined: Sun Mar 04, 2012 12:49 am

Re: Cubieboard vs RaspberryPi B

Tue Apr 09, 2013 12:37 am

Dunno - are they exclusive? I suppose it depends on the amounts involved, how much effort is required and who's doing it. You know better than me - I'm guessing from your response that Android can't be realised without foundation input as non disclosure agreements are involved? I understand why it is not a priority and its not as if the Pi lacks alternatives.

W. H. Heydt
Posts: 11097
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: Cubieboard vs RaspberryPi B

Tue Apr 09, 2013 5:08 am

johnbeetem wrote:
W. H. Heydt wrote:Has anyone else managed to get a tenth as many general purpose small, cheap single board systems on the market? I certainly don't know of any. For that alone the Foundations deserves notice and accolades.
I believe BeagleBoards have sold approx 100K, which meets your one-tenth. Granted it took them far longer to sell 100K than RasPi's 1M, IMO mostly because Beagle was marketed primarily as an embedded development platform for high-end TI ARMs rather than as "a GNU/Linux box". Another reason RasPi has been more successful is that IMO it's a lot easier to figure out which RasPi OS to run and from whence to download it.
I've already mentioned my bemusement at finding that the way you set up a Cubie is to load Berryboot to a micro-SD card (no full sized card slot here!). Once you boot a Cubie with Berryboot, you get a menu of possible OS's. Since I'd already decided I wanted a Debian system, that locked me into Raspbian. This led to a 45+ minute downlaod and install (it is nice that it allows you to install on a variety of different devices...such as a SATA device). There then followed about 2 hours of update and upgrade activity. Getting to a reasonably current, running system on the Cubie is NOT a fast process. If I add in the time to add the additional software packages I wanted (such as PuTTY and MySQL), figure between 3 and 4 hours. And I still haven't added the accounts I want and ported over the historical data and application code that I plan to run on it.

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 24162
Joined: Sat Jul 30, 2011 7:41 pm

Re: Cubieboard vs RaspberryPi B

Tue Apr 09, 2013 6:58 am

pygmy_giant wrote:Dunno - are they exclusive? I suppose it depends on the amounts involved, how much effort is required and who's doing it. You know better than me - I'm guessing from your response that Android can't be realised without foundation input as non disclosure agreements are involved? I understand why it is not a priority and its not as if the Pi lacks alternatives.
Not quite exclusive - the port probably wouldn't be done by the Foundation themselves - would have to be a volunteer effort from someone at Broadcom - the version on the front page a while back was done that way. I do wonder if the required low level acceleration libraries could be released, then the community could finish the port off, but there are probably some nasty licencing issues related to that.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I think it’s wrong that only one company makes the game Monopoly.” – Steven Wright

zwerggalaxie
Posts: 1
Joined: Mon Oct 14, 2013 4:31 pm

Re: Cubieboard vs RaspberryPi B

Mon Oct 14, 2013 5:04 pm

ddv2005 wrote:
I just got my Cubie and it is very fast and stable versus Pi. My web camera streaming application got HD 15fps with 0.7% cpu usage. On Pi the same web camera can't start with HD settings and it got 3 fps in VGA resolution with 10% cpu usage. USB audio on Cubie is FATNTASTIC...10 ms frames and NO drifting at all. On Pi the same usb audio device got 200ms audio drifting (about 3000 samples on 16Khz) and audio clicks even on 20ms frames!!!

Could you share more details about this? I have to use the Cubie for a web cam streaming application and I would need any further help.

Thanks!

corb555
Posts: 1
Joined: Sun Apr 06, 2014 1:13 am

Re: Cubieboard vs RaspberryPi B

Sun Apr 06, 2014 1:48 am

Correcting some misinformation about Cubie (I have both an Rpi and a Cubieboard2 and a half dozen Arduinos).

- One comment was that installing Linux on Cubie is difficult - It is not. Simply download Cubieboard's Lubuntu or Debian and image to SD card (using DD or WIn32DiskImage). It takes a few minutes.
- Another comment was about Cubie power supply - that it is difficult to find 2A supplies and Cubie uses a proprietary connector. Neither is true. CB2 is a standard power connector and there are tons of 2A brands available.
- The main difference between Rpi and Cubieboard/CB2 is that the Cubie is more expensive, but faster (Dual core at 1Ghz),and has more memory (1 Gb). CB2 has Mali GPU and Neon instructions (not sure what Rpi has for that). And Rpi has obviously sold orders of magnitude more systems.

All three platforms (rpi, arduino, and multi-core ARM) have their purpose. If RPI has enough power, use it. If you need more power, use a multi-core ARM like Cubie2, and if you want a crazy cheap system with RT support, use Arduino.

User avatar
Montala
Posts: 675
Joined: Mon Mar 05, 2012 11:54 pm
Location: Herefordshire (U.K.)

Re: Cubieboard vs RaspberryPi B

Wed May 14, 2014 9:26 pm

Hi,

Although the majority of the posts in this thread are now well over 12 months old, I see that there has been one recent one!

Although the first Cubieboard used the Allwinner A10 processor, its successor, the Cubieboard2 comes with a dual-core Cortex A20 chip (an ARMv7 platform) boasting dual Mali-400 graphics processors, although hardly any of the Linux operating systems currently available take full advantage of these.

I had never even heard of the Cubieboard until a two page review by Gareth Halfacree in the current issue (139) of Linux User & Developer magazine encouraged me to buy one. I would like to add here that this particular magazine claims to be "No.1 for Raspberry Pi", with at least 14 pages in every issue dedicated to the Pi.

Just in case anyone thinks that this post is an advertisement for the Cubieboard, it isn't! Developer support has already virtually dried up, although the fact that downloads are also offered by Fedora, who use it as a test bed for their ARM editions certainly helps... even it it does require the use of a second Linux computer to configure a usable card, which in this case is a micro SD rather than a 'full size' one.

Some of the other operating systems I have tried refuse to recognise my Logitech K400 wireless keyboard, others require me to edit obscure configuration files just to obtain internet access, while one just gave me an unreadable screen display... definitely a development board I think... even though it is now readily available, here in the UK.... even from Amazon!

There is certainly much more third party support available for the Raspberry Pi, and many books have already been written about it, whereas there is not even a single Kindle offering for the Cubieboard!

Yes, I am sure it will give me quite a few hours of 'fun', but it has also helped to rekindle my interest in the Raspberry Pi, which has to be a good thing! :)

riklaunim
Posts: 265
Joined: Tue Apr 22, 2014 7:34 pm

Re: Cubieboard vs RaspberryPi B

Wed May 14, 2014 10:09 pm

There is a lot of ARM single board computers out there. Most of them aren't very hot and active. There is just limited use for such board with plain Linux. Android is more handy for it and that’s why HDMI dongles with that OS got very popular. You plug it and it works (and takes most from closed source ARM drivers - like for Mali GPU). Or HDMI output from a tablet and some phones.

I have odroid X2 and I use it from time to time. Now it's used as a staging server for an app we are developing at the company, but 2GB or RAM is getting low for half gig Postgres DB and Nginx/Supervisor/Celery/Django setup :) And that's one of the highest specs you can get in those SBC (keeping price sane). Intel BayTrail or AMD Mullins SoCs could be more effective (good GPU drivers, and it's x86), but still I don't think SBC shouldn't try to replace more common PCs (let say Thin Mini ITX mini PC or nettop).

What would be cool is a SBC that has software APIs (like in Python) for many hardware parts. Like let say we take an average Android phone (2/3G, Bluetooth, WiFi, batter, touch screen, 2-4 core CPU, 512MB-1GB RAM, maybe GPS). There is SL4A - scripting layers for Android. You can send SMS, connect to WiFi, do calls even, do stuff with bluetooth, but still you don't get the library/software freedom of a plain Linux distribution. SL4A is mostly constrained to what Android exposes via API.

So would be cool if some cheap (yet hardware rich as described above) Android Chinese phones or tablets could run a mix or plain Linux (full system Python, other libraries and apps) and Android (for the API and like for running the phone features, screen...) - like Linaro ARM Ubuntu spin on backend with Android on the frontend. If it would have USB Host then it could also connect to USB-based microcontroller board (and drive a robot for example, maybe even power low power motors and alike). Bluetooth allows Bluetooth-UART communication (would work with serial-interfaced Arduino).

Return to “Off topic discussion”