Most information presented during the annual X.Org Developers’ Conference doesn’t tend to be very surprising or ushering in breaking news, but during today’s XDC2020 it was subtly dropped that Arm Holdings appears to now be backing the open-source Panfrost Gallium3D driver.
Panfrost has been developed over the past several years as what began as a reverse-engineered effort by Alyssa Rosenzweig to support Arm Mali Bifrost and Midgard hardware. This driver had a slow start but Rosenzweig has been employed by Collabora for a while now and they’ve been making steady progress on supporting newer Mali hardware and advancing the supported OpenGL / GLES capabilities of the driver.
This is a major departure from previous policy for ARM, since the company always shied away from open source efforts around its Mali GPUs.
Reminds me of 3dfx open-sourcing Glide just before they were bought by Nvidia.
I didn’t know about that. Was it later base for noveau?
No, no, something completely different. Back in the 1990s, before Direct3D was released by Microsoft, every manufacturer of 3D accelerators had their own 3D API. Most of those 3D APIs went nowhere and only got utilized by a handful of games, and even those games removed support in subsequent patches when they added Direct3D support (https://www.youtube.com/playlist?list=PLOeoPVvEK8ft-yumHjp4w4rQnr_MIab17). But 3dfx Glide API was successful enough (due to the prevalence of 3dfx Voodoo cards) that some important games released before the era of Direct3D supported Glide exclusively. For example Need For Speed II SE, Screamer II and Tomb Raider I come to mind.
As a result, there was a desire to run those Glide-exclusive games on non-Voodoo hardware using Glide wrappers, but 3dfx was aggressive to stop any such projects, as they wanted Glide to be a Voodoo exclusive in the hopes this exclusivity would drive Voodoo sales at a time, since by the time those wrappers were developed Voodoo sales were taking a huge beating by Nvidia’s GeForce.
But just before 3dfx was acquired by Nvidia, 3dfx management decided to gift the gaming community the Voodoo Linux drivers (complete with Glide support) as open source, and the Glide specification, so Nvidia couldn’t have it as an exclusive. A rare case of the Honest Corporate Executive: https://tvtropes.org/pmwiki/pmwiki.php/Main/HonestCorporateExecutive
I suspect ARM is doing the same thing: Giving us as much open source BiFrost and Midgard specification they can before Nvidia gobbles them up.
Question – is it coincidence that both companies were bought by Nvidia?
Yes. However, ARM’s reaction is for the same reasons 3DFX open sourced their software. nVidia is known as a bit of a IP hoarder, and enjoy making money of patent licenses as much as they enjoy making graphics cards.
Notably, nVidia already have quite an investment in ARM, having several chipsets and S0C’s based around the ARM architecture. Owning the rights to ATM also means that they would essentially need to pay ARM licenses to themselves, which ultimately negates that cost. By not having any licencing expenditure, nVidia can essentially undercut the market, pushing out other competitors due to low prices. And in a market like SoCs, where margins are razor thin, negating ARM license fees could make nVidia SoCs a great deal (25%, 50%) cheaper than competitors
@The123king
Nobody knows why Nvidia has bought ARM yet, we are all in wait-and-see mode. I have heard all kinds of theories being thrown around, none of them is anything more than a hypothesis. Here are some:
1. Nvidia wants to do something to Qualcomm, despite Qualcomm not licensing any IP cores and only licensing the instruction set architecture patents, which means that whatever ARM does to the new IP cores is irrelevant.
2. The ARM Mali IP cores get replaced by Nvidia GPUs for phone SoCs. But considering anything but Qualcomm is low-end at this point due to the fact only Qualcomm has the modems required for hi-end phones and even mid-range phones, is there any money there?
3. What you said. Nvidia undercuts everyone in embedded. Considering the impending move to RISC-V after the announcement, and considering the fact there isn’t much money there, is it worth the massive expense?
4. GPGPU SoCs for supercomputers. Basically tiny ARM cores attached to a monster GPGPU. But Nvidia is moving away from that and into RTX. And then there is the question of how niche is a supercomputer with tiny CPU cores anyway, and the fact Nvidia could have done this anyway with their Denver core.
And you know what? It could be nothing of all that. It can as well be a case of that Nvidia CEO wanting to “park” the massive piles of money he made from that conspicuous consumption craze known as the PC Mustard Race. A craze which is slowly drawing to a close, because those PC Mustard Racers went so elite and demanded for so high system requirements from publishers that nobody buys PC games anymore. Even multi-platform publishers like Rockstar treat the PC as an afterthought and take anywhere from 6 months to a year to release the PC port.
I thought OpenGL already existed before Direct3D/DirectX ?
No early 3D card for Windows PCs supported OpenGL. They all had their own APIs initially, then they all supported Direct3D practically the moment it came out, then they added support for “MiniGL” (a subset of the OpenGL API, essentially just the OpenGL commands required to run Quake) and then they gradually started to support all of OpenGL.
That’s interesting information! I like such news about developing, I wish such information were in college, but instead we have only unnecessary writing assignments. Thanks to writing services, I don’t need to write them by myself. I found a few reliable services on https://pickthewriter.com/ it’s a good site, that prepared true reviews about different writing companies, perhaps it could be helpful for someone else.