ekianjo
Hardcore Member
I agree. People on reddit are close to be the worst asses out there.
So, What's the DragonBox Pyra Like ? [ FOSDEM 2014 ]
- Thread starter ekianjo
- Start date
Hồng Thất Công
Đả Cẩu Bổng Pháp
Nope. They are just normal iPeople
cuetzpalin
Member
Maybe, but with the internet, you can't hide it either. I think ED did it very professionally. Now, we must bear in mind that the people that will sit for an hour watching a FOSDEM presentation is not the average user, but a person that most likely would dig the web looking for all the info about the project's and team's past. So they would find about it anyway. For what I can hear in the presentation, I think the attendants were very interested and trusted the project entirely. I think it's likely that some of them were familiar with the pandora and had some idea of the problems and delays.
For the main website, I agree with you that Craig and the Pandora issues shouldn't be addressed this way, maybe just briefly in the FAQ, with a link to the FOSDEM presentation video, for example.
Neelix
Insecticidal Maniac
You missed my point. I wasn't suggesting trying to hide what happened, just that it could be addressed without naming names.
It's all about presentation. Consider a statement like the following:
"There were multiple companies involved in the production of the Pandora. The various production issues, combined with some bad management decisions, meant that one of the companies was unable to complete delivery of many of the units they had sold. That company has since folded, and has no involvement with the planning or production of Pyra."
Do you see what I'm getting at? In this example the issue has been addressed without naming names. It tells people what they need to know without slinging mud on any individuals, which gives a much more professional impression. If people want to know specifics they are easy enough to find, but those details don't become associated with the new product.
- Neelix
It's all about presentation. Consider a statement like the following:
"There were multiple companies involved in the production of the Pandora. The various production issues, combined with some bad management decisions, meant that one of the companies was unable to complete delivery of many of the units they had sold. That company has since folded, and has no involvement with the planning or production of Pyra."
Do you see what I'm getting at? In this example the issue has been addressed without naming names. It tells people what they need to know without slinging mud on any individuals, which gives a much more professional impression. If people want to know specifics they are easy enough to find, but those details don't become associated with the new product.
- Neelix
Last edited by a moderator:
ekianjo
Hardcore Member
Well yeah, you could do it this way, but since it's a single-person company, the name == the company anyway, and the issues of the company are very much the issues of that person, no matter how you want to fold things up.
You're making the reality all rosy. Let me fix that for you:
Neelix
Insecticidal Maniac
What part of the phrase "without slinging mud" was unclear? As I said, it's all about presentation. You could say it that way, but it reflects badly on you for bringing it up when what you should be focusing on is the the product you are trying to sell and the concerns of your potential customers.
- Neelix
- Neelix
cuetzpalin
Member
I see. Well, I partially agree with you, however, if we consider that just as they did with the Pandora, they are presenting the project very openly: they are not claiming or trying to present themselves as a company or a brand, but rather as a team of developers that went on to create a product. Which, actually, is nothing but the truth.
I know that even that way, naming names could be avoided, but again, I don't think it was done in an harmful nor negative way, especially judging the reaction of the attendants as no one asked more about the issues.
ekianjo
Hardcore Member
Indeed, therefore, mentioning "no distraction/parasite (Craig)" is actually a very effective way to mention the past issue, without going into too many details or trying to hide anything, and without transition moving to the actual product.
Honestly, if there was one thing that would prevent me from spending one penny on a Pyra would be to know that Mr C is involved. I knew he wouldnt, but I'm actually very glad that ED went ahead and mentioned it clearly. After the ICP2 disaster?/fraud? (choose the one which applies), there's no way people are going to trust anything involving him. At least in the Pandora community.
Exactly. There were no questions, therefore, problem closed.
ExquisiteCorpse
Member
I asked on reddit why PowerVr was garbage and got this response. Could someone explicate on this?
PowerVR as an architecture is just fine, it's the politics surrounding it that make it garbage for embedded Linux. The drivers that Android uses are compiled with Android's libc and are closed source so while the PowerVR is a great GPU for Android with these binary drivers, you can't easily use these drivers to run embedded GNU/Linux distributions like Debian, Ubuntu, etc. There's libhybris, but it's kind of a hack. Also, Android drivers only support OpenGL ES while it may be possible to implement some desktop OpenGL commands with a given GPU architecture. Also, X11 based Linux distributions need xorg compatible graphics drivers which Android drivers are not.
There are various projects out there reverse engineering mobile GPU's, figuring out how they work internally, and then writing new open source drivers from scratch. The most complete one is Freedreno, a driver for the Qualcomm Adreno series of GPU's. That driver is able to run accelerated X11 desktops like GNOME Shell already. The nouveau reverse engineered nVidia driver has added Tegra support to its list of GPUs. There's also a project called lima driver to reverse engineer the ARM Mali GPUs but there haven't been any updates in months. There is a project to do the same for PowerVR but they haven't shown any real progress at all. That's why I'd avoid PowerVR like the plague when looking for embedded Linux GPUs.
PowerVR as an architecture is just fine, it's the politics surrounding it that make it garbage for embedded Linux. The drivers that Android uses are compiled with Android's libc and are closed source so while the PowerVR is a great GPU for Android with these binary drivers, you can't easily use these drivers to run embedded GNU/Linux distributions like Debian, Ubuntu, etc. There's libhybris, but it's kind of a hack. Also, Android drivers only support OpenGL ES while it may be possible to implement some desktop OpenGL commands with a given GPU architecture. Also, X11 based Linux distributions need xorg compatible graphics drivers which Android drivers are not.
There are various projects out there reverse engineering mobile GPU's, figuring out how they work internally, and then writing new open source drivers from scratch. The most complete one is Freedreno, a driver for the Qualcomm Adreno series of GPU's. That driver is able to run accelerated X11 desktops like GNOME Shell already. The nouveau reverse engineered nVidia driver has added Tegra support to its list of GPUs. There's also a project called lima driver to reverse engineer the ARM Mali GPUs but there haven't been any updates in months. There is a project to do the same for PowerVR but they haven't shown any real progress at all. That's why I'd avoid PowerVR like the plague when looking for embedded Linux GPUs.
ExquisiteCorpse
Member
Last edited by a moderator:
Pickle
Mega GP Mania
powervr has a bad reputation for the bad support for the intel poulsbo products. There was only a binary driver and it didnt work well at all. I think many observers correlate that with the mobile line too. So any labeled powervr gets a bad wrap.
the experience with powervr on the pandora hasnt been pleasant, but it works. The driver comes in 2 parts, theres a open source layer and the binary blob. Problems arise with the open source part as it isnt maintained across kernels. I believe notaz at one point worked on this to get it working with newer kernels.
Another issue with the driver has been X support. Early drivers supported DRI1, which is what we use. Later only supported DRI2, which is not supported. But it can be avioded as the driver supports a RAW mode and that works regardless.
So choosing powervr isnt ideal (but there arnt many options), but it can be worked with.
Im on the side of functionality over ideals. I dont care for binary blobs, but if it does what i intended to use the gpu for then im ok. For instance ill choose nvidia blob over noueveu for performance every time.
the experience with powervr on the pandora hasnt been pleasant, but it works. The driver comes in 2 parts, theres a open source layer and the binary blob. Problems arise with the open source part as it isnt maintained across kernels. I believe notaz at one point worked on this to get it working with newer kernels.
Another issue with the driver has been X support. Early drivers supported DRI1, which is what we use. Later only supported DRI2, which is not supported. But it can be avioded as the driver supports a RAW mode and that works regardless.
So choosing powervr isnt ideal (but there arnt many options), but it can be worked with.
Im on the side of functionality over ideals. I dont care for binary blobs, but if it does what i intended to use the gpu for then im ok. For instance ill choose nvidia blob over noueveu for performance every time.
ExquisiteCorpse
Member
Cheers Pickle, the guy who wrote the above also said. I know tegra was out due to Nvidia strong arming their contractors for hardware design but why not go with the Adreno driver?
So because the Lima driver isn't complete that's a good reason to choose something that will never in a million years have a proper driver? Sounds like BS to me, especially how he just ignored Adreno (the GPU that is actually well supported, both the 2xx and 3xx series) and Tegra (the nouveau project has Tegra GPU support in progress). I don't get how he intends to make a gaming device and not focus on having a properly supported GPU.
So because the Lima driver isn't complete that's a good reason to choose something that will never in a million years have a proper driver? Sounds like BS to me, especially how he just ignored Adreno (the GPU that is actually well supported, both the 2xx and 3xx series) and Tegra (the nouveau project has Tegra GPU support in progress). I don't get how he intends to make a gaming device and not focus on having a properly supported GPU.
Pickle
Mega GP Mania
I dont know if ED looked at adreno or not. I dont keep up on all the confgurations for soc's. There are a lot other things to consider when choosing the soc, like maybe the CPU is older on the adreno. Maybe there isnt a distributor, or maybe they didnt respond. I imagine there isnt a perfect answer that would make everyone happy.
I disagree with him saying powervr is not a "properly supported GPU". There will be a driver and it can be made to work and provide the functions that are expected.
If the driver was known not to work then he would be right.
I disagree with him saying powervr is not a "properly supported GPU". There will be a driver and it can be made to work and provide the functions that are expected.
If the driver was known not to work then he would be right.
WizardStan
Mega GP Mania
- Joined
- May 24, 2008
- Messages
- 16,731
The driver shim is completely open source. I don't really understand how the closed source blob called by the shim makes porting the driver to other kernels and versions of Android more difficult than if it were a full blown open source driver.
I mean, it's a "library" of sorts, right? You write the "driver" to call this "library". If the kernel interface changes you rewrite the driver to make the same library calls in a different way.
I mean, it's a "library" of sorts, right? You write the "driver" to call this "library". If the kernel interface changes you rewrite the driver to make the same library calls in a different way.
Yes, I also looked at the Adreno, and no, I couldn't get any other SoCs.
It's awesome how some "smart" guys complain that we didn't choose this and that one.
If he is able to set up the contracts with these companies so I can buy these SoCs, I'd be happy.
It's not like going into a supermarket and put the SoC you like into your basket... then a Qualcomm would probably have been the winner.
Companies releasing OpenSource drivers are not helpful if they're not selling their SoCs to companies that try to produce an OpenSource product.
I also don't really see the issue with the PowerVR.
They are talking about that there are only Android drivers available - but for each and every OMAP TI has released Linux drivers to the public as well.
And the drivers ARE already available for OMAP5 as well.
The kernel-source is open, so it can be used with more recent kernels as well - which is what notaz already showed with the Pandora.
One PND with various SGX driver versions running on our kernel...
It's awesome how some "smart" guys complain that we didn't choose this and that one.
If he is able to set up the contracts with these companies so I can buy these SoCs, I'd be happy.
It's not like going into a supermarket and put the SoC you like into your basket... then a Qualcomm would probably have been the winner.
Companies releasing OpenSource drivers are not helpful if they're not selling their SoCs to companies that try to produce an OpenSource product.
I also don't really see the issue with the PowerVR.
They are talking about that there are only Android drivers available - but for each and every OMAP TI has released Linux drivers to the public as well.
And the drivers ARE already available for OMAP5 as well.
The kernel-source is open, so it can be used with more recent kernels as well - which is what notaz already showed with the Pandora.
One PND with various SGX driver versions running on our kernel...
Last edited by a moderator:
Linux-SWAT
Forum Addict!
- Joined
- Feb 13, 2010
- Messages
- 9,187
When i'm happy with some product, i usually keep using the same brand.
Actually our OP OMAPs have proven to be pretty reliable, so i'm happy to stick with them.
Also, about horsepower concerns, the 5432 is a beast, so i can't regret not having a few more Hz.
I'd say the Destiny led ED to the wiser choice.
Actually our OP OMAPs have proven to be pretty reliable, so i'm happy to stick with them.
Also, about horsepower concerns, the 5432 is a beast, so i can't regret not having a few more Hz.
I'd say the Destiny led ED to the wiser choice.
Last edited by a moderator:
ExquisiteCorpse
Member
Ah fair enough, that guy was basically wrong then.
- Joined
- Dec 8, 2005
- Messages
- 6,603
- Age
- 40
- Location
- UK
- Website
- pokeparadox.itch.io
- WEBSITE
- https://github.com/pokeparadox
- YOUTUBE
- pokeparadox
Looking at the Pandora as a basis... yeah the guy is wrong. Plus we don't just have to stick with GLES thanks to the GLShim which will more than likely be ported.
WizardStan
Mega GP Mania
- Joined
- May 24, 2008
- Messages
- 16,731
I understand from a political and longevity perspective.When people say something like "well I could just hack a PSP" and my response is "yes, but you'd still be giving money to a company actively working against you instead of at least trying to give you what you wanted from the start"
And there's the concern about bugs and what happens if IT decides they don't want to continue support which is a valid "what if" scenario but all the doomsaying seems a little premature.
I'm pretty sure the people complaining are doing so strictly from a pure FOSS perspective: free or nothing.
levi
Still fresh, damnit!
Yes, it's a shame we never got a completely reliable binary blob out of TI for the Gigahertz Pandora, but it has taken us a long time to discover its issues, and actually we've not done too badly out of it.
