Libre-SoC Open-Source GPU/VPU Project Loses Key Funding

[cb88]

Are you stupid their whole CoC was the reason they got in the mess, It's because it incredibly vague (deliberately to be “anti-woke”). They didn't want to offend you people, who somehow are offended over any comprehensive CoC lmao.

Additionally they got in this mess... because the project never had any technical merit, and the developer kept complaining that their ideas couldn't get into the RISC-V spec... again because they had no technical merit. That is when they switched to OpenPower after complaining and *whinging* for ages.

A 5GFlop GPU is an absolute joke. This is on par with something from about 15 years ago... and probably actually worse because it is literally a soft GPU.

[kurkosdr]

File this under "open-source GPUs that went nowhere", next to OpenGraphics and Manticore.

The problem with open-source hardware is that the feedback cycle enabled by a compiler doesn't exist (you have to use slow emulators to see your changes in action), so the incentive is simply not there. Open-source CPUs tend to do better because there is business interest in the datacenter sector to fund development (a form of spreading costs among big businesses).

[Old Grouch]

A 5GFlop GPU is an absolute joke.

A 5GFlop open hardware GPU is not a joke. I'll happily take a 15-year old GPU (or CPU) with no 'binary blobs' or code signed with keys I don't control.

I'm not so naïve to think that hardware trojans don't exist, but removing the easy possibility of software trojans is a good aim to have, If this project fails, I hope someone else can pick up the work and run with it. The processor design is interesting, but orthogonal (pretty much) to the security aim - which a lot of people would like to see work,

[Doomer]

A 5GFlop open hardware GPU is not a joke. I'll happily take a 15-year old GPU (or CPU) with no 'binary blobs' or code signed with keys I don't control.

well, so far every attempt turned into a joke so maybe there is really no escaping the boogeyman

[EliasOfWaffle]

A 5GFlop open hardware GPU is not a joke. I'll happily take a 15-year old GPU (or CPU) with no 'binary blobs' or code signed with keys I don't control.

I'm not so naïve to think that hardware trojans don't exist, but removing the easy possibility of software trojans is a good aim to have, If this project fails, I hope someone else can pick up the work and run with it. The processor design is interesting, but orthogonal (pretty much) to the security aim - which a lot of people would like to see work,

the only way is trying, without a simple implementation will be impossible to have a competitive, this can be evolved and in some years can be more contributors and improvements, but it's not easy.
The only way to scape from Demiurge(Big Corporations) is getting knowledge(trial and error), is impossible to design a faster thing without a slow thing to be a prototype, and this is from risc-v plan

[sophisticles]

I have to wonder if @EliasOfWaffle was one of the scammers considering he spent 5 posts one after the other to convince us that this project was not a scam.

[EliasOfWaffle]

Additionally they got in this mess... because the project never had any technical merit, and the developer kept complaining that their ideas couldn't get into the RISC-V spec... again because they had no technical merit. That is when they switched to OpenPower after complaining and *whinging* for ages.

A 5GFlop GPU is an absolute joke. This is on par with something from about 15 years ago... and probably actually worse because it is literally a soft GPU.

Obvious, nothing is real scalar because of big companys and things like sillicon consortiums restrictions, SVP64 is trying to be a new revolutionary thing, a new trying, and the most reason is because RISC-V Mailing list and things about extensions depends to be a foundation member, and for it, is necessary money, OpenPower Foudation, is a real open ISA foundation, because not have restrictions for simple things

[EliasOfWaffle]

I have to wonder if @EliasOfWaffle was one of the scammers considering he spent 5 posts one after the other to convince us that this project was not a scam.

The project is not a scam, you just need to search and read email list, and view presentations and submissions, this is not scam, and i'm not a developer or hardware designer

[EliasOfWaffle]

Escape from Demiurge(Big corporations) get into Sophia(Open Hardware)
We can be back to the Pleroma

[cb88]

A 5GFlop open hardware GPU is not a joke. I'll happily take a 15-year old GPU (or CPU) with no 'binary blobs' or code signed with keys I don't control.

I'm not so naïve to think that hardware trojans don't exist, but removing the easy possibility of software trojans is a good aim to have, If this project fails, I hope someone else can pick up the work and run with it. The processor design is interesting, but orthogonal (pretty much) to the security aim - which a lot of people would like to see work,

It absolutely is a joke... and if it fails nobody else is going to pick up thier trash and continue they'd just make something thier own good to begin with. Just being open doesn't make something good. We probably have NORMAL, RISC-V CPUs now with more than 20 gigaflop... the fact that they suck that bad is insane.

Your very average mundane 10 years behind the curve chinsese company has no issues churning out designs that are orders of magnitude faster than this crap.