Caine
Hardcore Member
Well, it's not the only way 
Able to upgrade not just the speed but manufacturer of CPU?
- Thread starter second exodous
- Start date
God Ginrai
Godmaster
It's more often how big companies leave a startup long enough for it to grow into something worth purchasing, which they do once it looks like it finally will start to become a threat.
-God Ginrai
Are you sure they don't have NDA? The impression I got from Coreboot/libreboot hearsay is that at least for desktop or server CPUs getting the needed information to code BIOS/UEFI for intel CPUs
implied some time even with NDAs.
Anyway. Beware of Intel and AMD.
I'm new and don't know the community much, so sorry if this is already known or something, but I think Intel (and more recently AMD) too, do not only require someone to program BIOS/UEFI code. They require it to include fixed binary parts there that control important functions, and even could act like a kind of hypervisor that keeps control of the computer no matter how much more free code you load it with. If your BIOS/UEFI does not run this binary signed code by the chip manufacturer the CPU might not even access RAM and certainly not boot. This has been so for years, more years with Intel than AMD GPUs. More recently intel, nvidia and amd GPUs require signed blobs, I believe ? I think the Tegra K1 did not, not sure about the X1, maybe not either.
I'm not aware that OMAP CPUs have these restrictions, so for me Intel or current AMD would be much worse than TI. I agree that Allwinner alleged GPL violations and their public stance about them, make Allwinner inadequate too. In fact ARM itself is not very free software friendly in my eyes, so there isn't really much good stuff in the market currently. I thought i.MX and OMAP was the less bad. But I might not be up to date.
Btw, I heard some time ago that all implementations of USB3 required proprietary code. Is it still so ?
Also, just a sidenote. Having the cpu in a separate board should also allow for replacing it with a replacement board of identical design, in case it breaks, not only upgrading it with a better CPU. That's already good for your pocket and the environment. It is much more likely that there will be spare units available in case it breaks than however likely it is that a different board will be produced someday by someone.
TheOldOne
Fallen Paladin
From what I understand the ATOM CPUs didn't require the signed binary blobs in the UEFI/BIOS to work but the "core i#" lines do.
levi
Still fresh, damnit!
It's not really possible to say yet whether the CPU board of the peripheral board will be more fragile in use. The CPU board is well away from the edges of the case, where failures tend to occur on the Pandora's main board (albeit mainly due to the placement of a heavy inductor in my experience), but the CPU board is sandwiched between the peripheral board and the case (and then the battery), so maybe be subject to crush mode failures (but hopefully the strength of the polycarbonate case will reduce that risk). But yeah, having both be replaceable can only be a good thing if one or the other breaks.
I'm not sure how tight intel's restrictions are given they let Lenovo introduce easily subvertable code, but yeah, binary blobs are something to be wary of after our experiences with the Pandora. Thanks for the info.
I'm not sure how tight intel's restrictions are given they let Lenovo introduce easily subvertable code, but yeah, binary blobs are something to be wary of after our experiences with the Pandora. Thanks for the info.
I'm not sure about ATOMs, but at least this port of coreboot to the Minnowboard, which has an Atom E3825, carries the FSP blob (and cpu microcode). Yet libreboot has or had some support for
intel d510mo motherboard, with a 2009 Atom D510. I guess it depends on the model, and I suspect the more recent atoms must be worse than the old atoms. You are right it does not seem to have ME,
so it may not have signed blobs, but it still looks hard to reverse engineer.
ekianjo
Hardcore Member
Nice try. I can tell you it ain't going to happen anytime soon for hardware companies. You must be talking about SOFTWARE-companies,
Caine
Hardcore Member
Well, yes. You got me there. That is indeed an important distinction to make.
WizardStan
Mega GP Mania
- Joined
- May 24, 2008
- Messages
- 16,731
I worked with an automotive board prototype (that may or may not have actually gone into a car) that used a Tegra running Linux. It's entirely likely that Tegra SoCs running Linux could be used all over in places you wouldn't even think about, like automotive, or manufacturing systems.
ible
professional vim user
well, by the time the next CPU board comes out, ED will probably have his own fabrication plant, so he'll be making the next SoC custom by himself with some licensing from ARM. no need to worry about openness/closedness.
levi
Still fresh, damnit!
Hah!
Why doing licensing from ARM ? Just hire hns to design a full SoC from scratch !
- Joined
- Jan 18, 2010
- Messages
- 11,418
There is a world of difference between being able to design a device using a SoC and designing a SoC from scratch even using existing IP.
ible
professional vim user
oh so we should talk to you about doing it?
I know, that's why I used an exclamation mark.
rygD
Nihilistic Mystic
- Joined
- Jan 18, 2010
- Messages
- 11,418
no we're not uptight...
Opens rygD profile... hovering mouse cursor over delete account button....
rygD
Nihilistic Mystic
I just checked my profile. There is no such button.
