possible soc updates.


monstercameron said:
os options maybe be limited by pvr gpu ip but baytrail may be different.
Click to expand...
Sounds like Bay Trail is indeed different.

Any thoughts from our hardware gurus as to the suitability of Bay Trail as a possible SOC for P2?

From my perspective it appears that Intel / x86 are finally starting to ask some interesting questions now that (from the link) :

'Bay Trail’s power consumption under heavy CPU load ranges from 1W - 2.5W, putting it on par with other mobile SoCs that we’ve done power measurements on.'
 
Aw heck, seeing how baytrail gets about the same graphics performance as AMD e-350 and having played Aquanox on the latter with nearly maxed out settings makes me want this.

/me sobs, X86 Pandora would surely kick ass - and it could be sold to companies if they could integrate it into their infrastructure. But the thought behind having ARM again is: plug your P1 SD in your p2, all PNDs work. And I guess the standby consumption of X86 is still too high.
 
Last edited by a moderator:
Honestly, if the P2 was x86, I'm not even sure if I would buy it. The only ties I have to x86 is Windows (because of 2 key apps and commercial games), and I've got a PC for that. I don't need my handheld to do that. ARM has a much nicer instruction set and has always been a more reliable product than any x86 processor I have ever used.

-God Ginrai
 
Last edited by a moderator:
True, but you could port x86 games more easy to such a system, and who knows what else could run onto it that not even need a port. :)    No compromises anymore when bringing PC games to a mobile device (ok, NVidia was faster there, even without X86 compatibility) but I'm sure the homebrew community could take huge advantage of an powerful X86 based mobile device . ^^

However, the main problem still is to find a Company that sells their Chips to small open source projects. Are there new infos about which SoC's and architectures are sold in small batches?
 
Last edited by a moderator:
Fusion_Power said:
True, but you could port x86 games more easy to such a system, and who knows what else could run onto it that not even need a port. :)    No compromises anymore when bringing PC games to a mobile device (ok, NVidia was faster there, even without X86 compatibility) but I'm sure the homebrew community could take huge advantage of an powerful X86 based mobile device . ^^

However, the main problem still is to find a Company that sells their Chips to small open source projects. Are there new infos about which SoC's and architectures are sold in small batches?
Click to expand...
I honestly don't see the benefit. The majority of things people actually want on x86 is closed source, so we will have no way to tweak it to work better on a handheld. As for the open source things, most that people would want from that would already have people interested in playing it on Linux, so it would already be converted to OpenGL. There is unlikely to be heavy x86 reliance elsewhere in the code, so it would not be that hard to get running on ARM anyways. It's not like all these applications are tossing around a ton of x86 assembly. Not to mention, optimizing any of the more demanding applications to work on the handheld would be more work if we need to use assembly. x86 assembly is a nightmare compared to ARM assembly.

EDIT: Not to mention that tossing out the chance of compatibility with the large library of software for the Pandora that generally has been optimized for mobile is just a stupid move.

-God Ginrai
 
Last edited by a moderator:
Mr_Loon said:
monstercameron said:
os options maybe be limited by pvr gpu ip but baytrail may be different.
Click to expand...
Sounds like Bay Trail is indeed different.

Any thoughts from our hardware gurus as to the suitability of Bay Trail as a possible SOC for P2?

From my perspective it appears that Intel / x86 are finally starting to ask some interesting questions now that (from the link) :

'Bay Trail’s power consumption under heavy CPU load ranges from 1W - 2.5W, putting it on par with other mobile SoCs that we’ve done power measurements on.'
Click to expand...
The price on the 'flagship' Z3770 is likely to be high - but the Z3740... that looks like it could be 'the' SoC to me.

Of course we know so little at this point - but it may be enough for ED to talk to someone at Intel to find out if it is even possible.

Reading the article Mr_Loon linked to - The OpenGL ES 2.0 that they referenced as being available - is that the same OpenGL that is used in all of the steam games I play on my Linux workstation?

Baytrial is 64bit, but all of the OS (Windows & Android) that they're talking about putting on it is 32bit.  In the hypothetical world that this was used as a Pandora successor - what would stop us from putting 64bit Debian, Ubuntu or even Mint on it?

The 4GB RAM limit in the chart seems dubious to me for a 64bit CPU...  It's plenty for most uses, but I'm wondering if we could go to 8GGB RAM - which puts it in desktop/notebook capability ranges.

If they're really coming in at the ~2W under load dissipation levels - I think it may be workable using my heat spreader + exposed fins idea to help get the system heat out without a fan.

ED - any interest?
 
Mr_Loon said:
Sounds like Bay Trail is indeed different.

Any thoughts from our hardware gurus as to the suitability of Bay Trail as a possible SOC for P2?

From my perspective it appears that Intel / x86 are finally starting to ask some interesting questions now that (from the link) :

'Bay Trail’s power consumption under heavy CPU load ranges from 1W - 2.5W, putting it on par with other mobile SoCs that we’ve done power measurements on.'
Click to expand...
I think it's a very attractive solution for a whole lot of reasons. The graphics won't be top of the line but surely "good enough", and they wouldn't be able to get a Qualcomm SoC anyway. The CPU performance and perf/W is very strong, the x86 compatibility is nice (even if just for WINE) and the open-source GPU drivers also supporting DX11 is icing on the cake. I don't like the idea of writing more x86 optimized code alongside the ARM optimized code I've written but that's going to have to happen regardless of what Pandora 2 uses.

It'd be good if ED got in touch with Intel on this. I remember Oak Trail (two generations back) was supposed to appear on some niche gaming handheld running Windows, so it's not out of the question that they'd provide it.

If it's chosen I still wouldn't want 8GB of RAM in it. Going 64-bit is still a good idea even with 4GB of RAM.
 
God Ginrai said:
Fusion_Power said:
True, but you could port x86 games more easy to such a system, and who knows what else could run onto it that not even need a port. :)    No compromises anymore when bringing PC games to a mobile device (ok, NVidia was faster there, even without X86 compatibility) but I'm sure the homebrew community could take huge advantage of an powerful X86 based mobile device . ^^


However, the main problem still is to find a Company that sells their Chips to small open source projects. Are there new infos about which SoC's and architectures are sold in small batches?
Click to expand...
I honestly don't see the benefit. The majority of things people actually want on x86 is closed source, so we will have no way to tweak it to work better on a handheld. As for the open source things, most that people would want from that would already have people interested in playing it on Linux, so it would already be converted to OpenGL. There is unlikely to be heavy x86 reliance elsewhere in the code, so it would not be that hard to get running on ARM anyways. It's not like all these applications are tossing around a ton of x86 assembly. Not to mention, optimizing any of the more demanding applications to work on the handheld would be more work if we need to use assembly. x86 assembly is a nightmare compared to ARM assembly.

-God Ginrai
Click to expand...
No benefit?  You're casually dismissing the vast majority of the Linux gaming that I do.

Neverwinter Nights native Linux X86.

Steam games (Half Life 1, 2, variants, Killing Floor, long list from there on out...)  http://steamdb.info/linux/

I'm really looking forward to playing Worlds of Magic on my Linux workstation - to play it on a Pandora successor would be amazing.  http://www.kickstarter.com/projects/304725524/worlds-of-magic-a-new-classic-4x-fantasy-game

Then there are all of the normal free Linux games that you can get from the repositories.  Yes - most of those can, with effort, be re-compiled and ported to ARM then tweaked for a handheld, etc...

How about this - Consider a Pandora successor with an X86 platform with OpenGL graphics and a 1080p display that from a software perspective looks just like a 'normal' laptop and can install Debian or Ubuntu right out of the box with no customization - and can do the same with the games meant for that 'normal' X86 computer.  That allows us to skip the game by game optimizations by not needing to do any in the first place via using hardware that the game/program sees as native and knows how to deal with.

The current repository of PNDs...- 90+% of that is already available on the primary Linux distributions.  The PND system is cool - very cool, but exists only in the microcosm of 5000 or so Pandora users.  Do we really have to tie a successor to it?

Exophase said:
Mr_Loon said:
Sounds like Bay Trail is indeed different.


Any thoughts from our hardware gurus as to the suitability of Bay Trail as a possible SOC for P2?


From my perspective it appears that Intel / x86 are finally starting to ask some interesting questions now that (from the link) :


'Bay Trail’s power consumption under heavy CPU load ranges from 1W - 2.5W, putting it on par with other mobile SoCs that we’ve done power measurements on.'
Click to expand...
I think it's a very attractive solution for a whole lot of reasons. The graphics won't be top of the line but surely "good enough", and they wouldn't be able to get a Qualcomm SoC anyway. The CPU performance and perf/W is very strong, the x86 compatibility is nice (even if just for WINE) and the open-source GPU drivers also supporting DX11 is icing on the cake. I don't like the idea of writing more x86 optimized code alongside the ARM optimized code I've written but that's going to have to happen regardless of what Pandora 2 uses.

It'd be good if ED got in touch with Intel on this. I remember Oak Trail (two generations back) was supposed to appear on some niche gaming handheld running Windows, so it's not out of the question that they'd provide it.


If it's chosen I still wouldn't want 8GB of RAM in it. Going 64-bit is still a good idea even with 4GB of RAM.
Click to expand...
How about we split the difference and go with 6GB RAM?   :D
 
I think that if we didn't find a valid ARM solution, we should go for Intel...and we could made an agreement for a sponsored console.
 
Grench said:
God Ginrai said:
Fusion_Power said:
True, but you could port x86 games more easy to such a system, and who knows what else could run onto it that not even need a port. :)    No compromises anymore when bringing PC games to a mobile device (ok, NVidia was faster there, even without X86 compatibility) but I'm sure the homebrew community could take huge advantage of an powerful X86 based mobile device . ^^


However, the main problem still is to find a Company that sells their Chips to small open source projects. Are there new infos about which SoC's and architectures are sold in small batches?
Click to expand...
I honestly don't see the benefit. The majority of things people actually want on x86 is closed source, so we will have no way to tweak it to work better on a handheld. As for the open source things, most that people would want from that would already have people interested in playing it on Linux, so it would already be converted to OpenGL. There is unlikely to be heavy x86 reliance elsewhere in the code, so it would not be that hard to get running on ARM anyways. It's not like all these applications are tossing around a ton of x86 assembly. Not to mention, optimizing any of the more demanding applications to work on the handheld would be more work if we need to use assembly. x86 assembly is a nightmare compared to ARM assembly.

-God Ginrai
Click to expand...
No benefit?  You're casually dismissing the vast majority of the Linux gaming that I do.

Neverwinter Nights native Linux X86.

Steam games (Half Life 1, 2, variants, Killing Floor, long list from there on out...)  http://steamdb.info/linux/

I'm really looking forward to playing Worlds of Magic on my Linux workstation - to play it on a Pandora successor would be amazing.  http://www.kickstarter.com/projects/304725524/worlds-of-magic-a-new-classic-4x-fantasy-game

Then there are all of the normal free Linux games that you can get from the repositories.  Yes - most of those can, with effort, be re-compiled and ported to ARM then tweaked for a handheld, etc...

How about this - Consider a Pandora successor with an X86 platform with OpenGL graphics and a 1080p display that from a software perspective looks just like a 'normal' laptop and can install Debian or Ubuntu right out of the box with no customization - and can do the same with the games meant for that 'normal' X86 computer.  That allows us to skip the game by game optimizations by not needing to do any in the first place via using hardware that the game/program sees as native and knows how to deal with.

The current repository of PNDs...- 90+% of that is already available on the primary Linux distributions.  The PND system is cool - very cool, but exists only in the microcosm of 5000 or so Pandora users.  Do we really have to tie a successor to it?
Click to expand...
where were you when I was making these points?
 
Last edited by a moderator:
monstercameron said:
where were you when I was making these points?
Click to expand...
I'm sorry, what? Are you talking to me or are you talking to Grench here. If you're talking to Grench you should strip your quotes more. If you're talking to me then:

Silvermont is attractive because the perf vs perf/W is good for the power profile of the device. If you don't fulfill that you're instantly off the table as far I'm concerned. Also as far as I'm concerned, none of your precious Temash or Kabini fulfill that. They're either two weak (2x1GHz) or too high power. I've been totally consistent with this the entire time. I even said CT+ is a somewhat viable choice, and BT-T is clearly much better. I haven't been quiet about how good I think Silvermont is, the above post is far from the first thing I said about it, and I've said before that it's very good in spite of the lack of ARM compatibility, which I have NOT retracted as a downside.
 
Last edited by a moderator:
You must log in or register to reply here.

Similar threads

_wb_
Overview of SoC options
16 17 18
Replies
348
Views
40K
urisma
U
M
soc power draw [rough numbers]
Replies
3
Views
2K
_wb_
_wb_
Back
Top