View Full Version : Android... that's Leenux isn't it?
SS2win
April 20th, 2014, 10:23 AM
I'm researching DVD/Navigation head units and see a bunch of them are running ATOM processors and 'droid. I guess I could dual boot to Windows but it sure would be neat to have a native EFILive for scanning/flashing tuning. Has there been any internal interest in an Android port?
Chevy366
April 20th, 2014, 10:35 AM
I'm researching DVD/Navigation head units and see a bunch of them are running ATOM processors and 'droid. I guess I could dual boot to Windows but it sure would be neat to have a native EFILive for scanning/flashing tuning. Has there been any internal interest in an Android port?
Ha, several of us have harped on Linux and would love to see a build for it, but alas Winders is the OS they choose to stay with.
Maybe one day when Winders get to be recognized as the old OS it is and finally goes to the bottom rung, the crew at EFILive will finally think about Linux. :anitoof:
SS2win
April 20th, 2014, 10:46 AM
I know at least one of the two original devs is a Linux fan/user. Times change and 'droid would be a great platform (though it could conceivably replace v2 hardware and that might be bad? :anitoof: )
Chevy366
April 21st, 2014, 04:15 AM
I know at least one of the two original devs is a Linux fan/user. Times change and 'droid would be a great platform (though it could conceivably replace v2 hardware and that might be bad? :anitoof: )
I don't think you could ever replace the V2 hardware/software with an Android platform, but it would be nice for us Linux users to have a install of EFILive software for it.
I have not revisited EFILive on Linux through Wine, but last time I tried you could do almost everything except flash (couldn't get USB to work).
GMPX
April 21st, 2014, 10:20 AM
Pretty sure Paul spent a lot of his time on Unix machines years ago.
However, for all the effort to support a non Windoze OS, is it worth it?
http://en.wikipedia.org/wiki/Usage_share_of_operating_systems
swingtan
April 21st, 2014, 10:30 AM
Android is "based on Linux"...... don't confuse the two.
The first big issue is that many mobile device run limited or no USB controllers. So plugging in a V2 may not be possible even if you had working software and drivers.
The Android OS has been modified and optimised for mobile / tablet based devices and in particular, the CPU's used in them. This means the code base is architectured for "ARM" hardware which is significantly different to the general "Linux X86" operating systems. Many of us have had used EFILive under Linux using applications like "wine" to emulate Windows. This works because the underlying hardware is still Intel X86 compatible and the EFILive code will run fine.
There is no "wine" equivalent for Android at the moment. OK, there is an Alpha in the works but I don't think it's released at all. The biggest issue you have is that the ARM processors are specifically designed for small devices and power efficiency. To convert the X86 code to run on ARM, the new version of wine will need to first convert the Android OS to a "Windows like" shell, then convert the "Windows like" shell to run on ARM instead of X86. This double conversion is going to need some serious CPU power, which ARM is not designed for.....
However! Allowing for technology improvements, it wouldn't be too long before mobile devices had sufficient CPU power to do the required work. You can get Windows 8 tablets that "should" work already, but a move to wider mobile devices is an interesting business idea. Just think, buying EFILive licenses via Google Play or the App Store....
Given that there is little focus on getting EFILive to run under X86 Linux, I doubt there's much thought going into Android.
GMPX
April 21st, 2014, 10:41 AM
Swingtan brings up a good point on the CPU architectures. I read somewhere that was part of the deciding factor for the XboxOne and PS4 to move to Intel x86 CPU's so the game developers only had to code for one CPU core type that would work on both consoles and current PC's (Win and OSX), makes perfect sense.
I think that because something 'could' run on a mobile device doesn't mean it would actually translate well with the limited screen space and somewhat susceptible security issues.
Chevy366
April 22nd, 2014, 01:35 PM
http://www.intel.com/content/www/us/en/smartphones/smartphones.html
There are Intel based Android devices now, quit fast in fact. I do agree Linux is a desktop/laptop more than Android but all are monolithic kernels (there is a Android install for Intel laptops now), I do think Android would be hard to code a .apx for EFILive.
But as I said Winders still has the lions share which is quickly eroding as tablets replace Laptops and PCs become a cumbersome clump on the desktop.
One day EFILive will have to move to Linux. :nixweiss:
Google has influenced a lot of the kernel development lately so Linux is becoming more Android than you think.
Oh and the day Winders becomes hardened to Linux standard let me know.:crash:
Graphene look it up, cool stuff.
GMPX
April 22nd, 2014, 02:35 PM
There are Intel based Android devices now, quit fast in fact.
x86 sucks which is why majority of phones are ARM based.
One day EFILive will have to move to Linux. :nixweiss:
I hope the new owners enjoy porting it because I can't see Paul EVER doing that, and anyway for what distro? There is only what 1,000 to choose from now :pokey:
Google has influenced a lot of the kernel development lately so Linux is becoming more Android than you think.
Oh and the day Winders becomes hardened to Linux standard let me know.
Common sense plays a big part in how safe your computer is too and how much market share something has. If (a big if) OSX or Linux had a 89%+ market share on desktops like Windows does then hackers would be all over them too.
I'm not an MS fan boy but the alternatives offer no real reason to switch.
swingtan
April 22nd, 2014, 02:55 PM
A couple of issues though.....
The Atom CPU is at best about 50% as powerful as a normal CPU. If you had the top of the range Atom CPU, it would be the equivalent of a duel core 1GHZ desktop / laptop CPU. The next CPU down, which is likely to be the one in most middle of the road tablets, is going to match a 300MHz cpu..... That's very old and slow.
The fact still remains that USB connectivity is still not supported. If you don't need the connect directly to the V2, then "Wine" will work for nearly all other functions, but there is no Wine for Android..... Also currently, the only Android native USB driver for the chipset used in the V2, is a beta and still requires "rooting" of the device. No commercial software developer I know of will release a product that requires you to void warranty.
Simon.
Chevy366
April 22nd, 2014, 03:28 PM
x86 sucks which is why majority of phones are ARM based.
No ARM Is cheaper, money dictates a lot of the choices.
I hope the new owners enjoy porting it because I can't see Paul EVER doing that, and anyway for what distro? There is only what 1,000 to choose from now :pokey:
Google has influenced a lot of the kernel development lately so Linux is becoming more Android than you think.
There is not a thousand, Ubuntu, Fedora, opensuse, are your top 3, Debian and Slackware main OS, now many winders is there now Vista, 7, 8,8.1-- :anitoof:
Common sense plays a big part in how safe your computer is too and how much market share something has. If (a big if) OSX or Linux had a 89%+ market share on desktops like Windows does then hackers would be all over them too.
I'm not an MS fan boy but the alternatives offer no real reason to switch.
That logic is so outdated it is funny, Android has a lot of the share of the market and OSX is Unix based, oh how I weep for the uninformed. :unsure: I don't use one line of code for anti-virus, you?
So stay stuck in the mud I have moved on from Winders, enjoy the basking in Gates quagmire. :thumb_yello:
Only thing that keeps me on Winders is EFILive, I have to have Winders to run it properly, see my market share point.:nixweiss:
GMPX
April 22nd, 2014, 04:08 PM
That logic is so outdated it is funny, Android has a lot of the share of the market and OSX is Unix based, oh how I weep for the uninformed. :unsure:
Are we talking about desktops/laptop OS's or any electronic device? You aren't trying to tell me a program written for an Android smart phone will 'just run' on a desktop PC running Linux because they are both Linux based?
Re the Linux choices, really, is this necessary?......
http://upload.wikimedia.org/wikipedia/commons/1/1b/Linux_Distribution_Timeline.svg
x86 isn't a good CPU architecture, Apple made a wise choice to go PowerPC years ago until they bowed to marketing GHz pressure too, or maybe because it was (god forbid) easier to run Windows on an Intel Mac :music_whistling_1:
I don't use one line of code for anti-virus, you?
Sure I do, it is Windows after all, I installed it about 3 years ago and it just runs in the background and never annoys me, not sure what all the fuss is about but this is the point Linux fan boys like to bang on about. It is not like Windows users have to install anti-virus software each time they boot the computer.
So stay stuck in the mud I have moved on from Winders, enjoy the basking in Gates quagmire. :thumb_yello:
I have on my PC VM's of Linux Mint 15, Ubuntu 12.04 and PCBSD 9.0, I'm not just being a MS fanboy (not that I am one anyway), I really have used the alternatives and saw no convincing reason to totally abandon Windows.
Only thing that keeps me on Winders is EFILive, I have to have Winders to run it properly, see my market share point.:nixweiss:
You did see that Windows still has 89% + market share on PC's right? Anyway, I'm glad you like EFILive enough to continue to have to infect your computer with Windows.
kangsta
April 22nd, 2014, 10:19 PM
out of interest, what makes x86 not good CPU architecture compared to PowerPC?
joecar
April 23rd, 2014, 01:20 AM
out of interest, what makes x86 not good CPU architecture compared to PowerPC?It is way more complex with its segmented memory scheme (altho now you can make a segment as big as the whole memory)...
back in the day when segments were 1MB (iAPX86/88), you had to split your program into multiple segments (by manipulating the segment registers which held the upper 16-bits of the 20-bit segment address... you can already see the complexity)... early compilers did not do this for you, you had to mess with it yourself.
By comparison, back in the day, the 68000 had a flat 32-bit architecture (even tho externally it had pins for 16 bit; it automatically performed 2 external accesses for 32 bit words), so you simply just wrote your program.
If the x86 architecture was simpler, it would be much faster than it is today.
SS2win
April 23rd, 2014, 01:31 AM
I've done some virtualization on Atom CPU's and they are fairly powerful processors. I've had WindowsXP, Linux/Asterisk and pfSense running all at the same time on a 4GB proxmox machine. While I haven't load tested it fully, first impressions is that it has enough power to do the job. It seems to have more "server" power than a C1000 Celeron does.
I think I'll try loading EFILive and doing USB passthrough. heh. you never know, it could work. Wine OTOH, nevermind.
GMPX
April 23rd, 2014, 08:02 AM
out of interest, what makes x86 not good CPU architecture compared to PowerPC?
One of the other issues with the x86 is they were quite limited in how many working registers you had, I think 8 but really only 2 for data manipulation use. PowerPC has 32 General Purpose Registers and 32 Floating Point Registers. Much of this comes back to the x86's origin, it was quite common at the time to not have the luxury of lots of working registers.
What this means is if you are doing a complex calculation (like the Virtual VE) on a PowerPC CPU if you have say 15 variables to add, multiply etc they are all stored and manipulated without shuffling that data around.
So for example, this is a snippet of the final Virtual VE calculation on an E78 ECM.
efsmul r12, r29, r7
efsadd r12, r22, r12
efsmul r11, r26, r6
The 'r' is the register number in the CPU, so the first operation multiplies the contents of r7 with r29 and stores it in r12. Then r12 is added with r22 and stored back in r12, then r6 is multiplied with r26 and stored in r11 and so on. The 'r' registers could contain anything, they are just scratch pads I guess you could say.
With the x86 that would take so many more steps because of it's architecture. For each operation you would have to store the result somewhere then go fetch it again next time you needed that variable, what might take one step on PowerPC might take 3 on x86. So a PowerPC CPU at 1GHz completed the same task as an x86 at 3GHz. Clock speed sells computers but doesn't mean faster is better when comparing different CPU's.
Cheers,
Ross
swingtan
April 23rd, 2014, 08:35 AM
Good old RISC vs. CISC..... I remember doing a CPU power comparison of a 1GHZ HP PA-RISC CPU vs an Athlon 3GHZ X64. I compressed multiple 1GB data files and the PA-RISC was at least twice as fast. Intel certainly pulled the wool over the worlds eyes back in the 80's and we are still having to manage that choice.
Anyway, technology aside, there are still issues with running a commercial application on Linux. For starters, Linux is not owned by a company who provides support for their product, unlike Windows. Yes, you can get support from Redhat or Novell (SLES), but that's not support for the core OS code and they only really do server based products on specific infrastructure. EFILive won't be doing Linux support, which is significantly more complex for the non-tech people. So who helps these guys out when things don't work?
Don't get me wrong, I'm a Linux fan and have worked in many versions of Unix for the past 20 years. But I've also working in enterprise support and can see that there is little business sense in working on a Linux port, at least at this time. I'd rather see more details in calibrations for tuning than a Linux port.
kangsta
April 23rd, 2014, 01:18 PM
Thanks for the excellent explanation Ross & Joe
Powered by vBulletin® Version 4.2.5 Copyright © 2024 vBulletin Solutions Inc. All rights reserved.