Got into fixing the then freshly dropped VIA Xfree86 driver for my Epia M10000 back in the summer of 2003. This turned into me putting a few things together, on the then only graphics hw with enough info to become totally open, to realize that display drivers, when structured properly, do not need to depend on the Int10 BIOS. This structured display driver was labelled "modesetting", and is now the driving principle of anything display on open source software. I spent years driving this topic, not only gathering info, but also convincing the community of the way forwards for display drivers. And when people were convinced, some parts of the X.org community who previously opposed this work, suddenly forgot where all this came from...
I was hired into SuSE in 2007, and as luck had it AMD was looking to get a handle on the canadian mess that they had just bought. So i ended up writing the bulk of the display code for the open source driver for the ATI R5x and Rx6xx. ATI and redhat were not amused and torpoed this by any means possible. AMD lost the internal struggle, especially with their post-K8 CPUs flagging, internal support for the radeonhd project dried up, and then AMD ran out of credit in the credit crunch of 2008. And with it NDA free datasheets died as well. But at least i had created some facts that made it impossible for ATI to completely go back to their old way of working, and some sembleance of an open source driver still exists today.
SuSE laid off 25 developers in just nuernberg in early 2009, and i was among them. I then finished my earlier modesetting work and wrote the first fully native display driver, from scratch, for coreboot. And also proved just how trivial VIA unichrome mpeg-2 decoding is if you have control over the rest of the related hw.
Then i spent some time at Nokia, working on exposing the PowerVR SGX debugging infrastructure to userspace in a meaningful fashion, so we could take control of our hw a bit more... But this was then cut short by the elopcalypse.
The tail end of the elopcalypse gave me some time to work on something else interesting, and i ended up reverse engineering the ARM Mali. This opened the floodgates for everything GPU on ARM. I even had to correct the RPI people on their poor attempt to fool us all with only a tiny portion of the code needed to provide a driver. In the wake of this i ended up being one of the founders of linux-sunxi.org.
ARM was having none of it though, and this made me untouchable once again. So now i am self-employed, selling my time to mostly automotive companies. The person who was blocking everything open source left ARM in 2017, and of course no-one bothered to give me a ring.
In the last few years i REed the flashing of ATI bioses, and added support to flashrom.
When the pandemic hit i was working on the FOSDEM Video-box, an allwinner A20 based hdmi and audio capture device which outputs to a hdmi, and which encodes a stream to be sent over the network. Running into all sorts of infrastructure shortcomings on both the V4L and KMS side, and finalizing the RE work on the allwinner h.264 encoder...
During 2021 and early 2022, i was REing the PowerVR PCX2, on linux, with win98 in qemu, from a laptop 2 rooms away... But this was kind of made moot by the img code release in march 2022. Still have not finished porting the build system for the demos into the img tree.
I am mostly into less popular graphics cards. Lots of VIA/S3, lots of matrox, a whole set of ati (for flashrom, and for r5xx/r6xx nostalgia), quite some tseng stuff (i reworked the Xfree86 driver in 2005/2006). I own almost no 3dfx or nvidia hw. I have about 40 different VIA unichrome based boxes standing around, with all sorts of interesting display additions. A similar amount of sunxi devices is also present, but those take almost no space in comparison.
I was hired into SuSE in 2007, and as luck had it AMD was looking to get a handle on the canadian mess that they had just bought. So i ended up writing the bulk of the display code for the open source driver for the ATI R5x and Rx6xx. ATI and redhat were not amused and torpoed this by any means possible. AMD lost the internal struggle, especially with their post-K8 CPUs flagging, internal support for the radeonhd project dried up, and then AMD ran out of credit in the credit crunch of 2008. And with it NDA free datasheets died as well. But at least i had created some facts that made it impossible for ATI to completely go back to their old way of working, and some sembleance of an open source driver still exists today.
SuSE laid off 25 developers in just nuernberg in early 2009, and i was among them. I then finished my earlier modesetting work and wrote the first fully native display driver, from scratch, for coreboot. And also proved just how trivial VIA unichrome mpeg-2 decoding is if you have control over the rest of the related hw.
Then i spent some time at Nokia, working on exposing the PowerVR SGX debugging infrastructure to userspace in a meaningful fashion, so we could take control of our hw a bit more... But this was then cut short by the elopcalypse.
The tail end of the elopcalypse gave me some time to work on something else interesting, and i ended up reverse engineering the ARM Mali. This opened the floodgates for everything GPU on ARM. I even had to correct the RPI people on their poor attempt to fool us all with only a tiny portion of the code needed to provide a driver. In the wake of this i ended up being one of the founders of linux-sunxi.org.
ARM was having none of it though, and this made me untouchable once again. So now i am self-employed, selling my time to mostly automotive companies. The person who was blocking everything open source left ARM in 2017, and of course no-one bothered to give me a ring.
In the last few years i REed the flashing of ATI bioses, and added support to flashrom.
When the pandemic hit i was working on the FOSDEM Video-box, an allwinner A20 based hdmi and audio capture device which outputs to a hdmi, and which encodes a stream to be sent over the network. Running into all sorts of infrastructure shortcomings on both the V4L and KMS side, and finalizing the RE work on the allwinner h.264 encoder...
During 2021 and early 2022, i was REing the PowerVR PCX2, on linux, with win98 in qemu, from a laptop 2 rooms away... But this was kind of made moot by the img code release in march 2022. Still have not finished porting the build system for the demos into the img tree.
I am mostly into less popular graphics cards. Lots of VIA/S3, lots of matrox, a whole set of ati (for flashrom, and for r5xx/r6xx nostalgia), quite some tseng stuff (i reworked the Xfree86 driver in 2005/2006). I own almost no 3dfx or nvidia hw. I have about 40 different VIA unichrome based boxes standing around, with all sorts of interesting display additions. A similar amount of sunxi devices is also present, but those take almost no space in comparison.
