So we shall.
Phoronix now has an overview post about the recent OpenPOWER Summit NA event; this is where the previously mentioned A2O core announced. Videos are not yet available, but slides currently are.
Larabel, Michael (Phoronix) ā 2020-09-20
OpenPOWER Summit 2020 Was This Week With Many Interesting Hardware/Software Talks
Videos have now been uploaded to YouTube, here is the OpenPOWER Summit NA 2020 playlist:
Hangover, a Wine project to get Win86 binaries running on Arm, has merged in Power support.
Hack cough excuse me what
Linux dropping old 32-bit PowerPC support
Phoronix ā 2020-10-09
PowerPC 601 Support Being Retired In Linux 5.10 - The First 32-bit PowerPC CPU
The PowerPC 601 as the first-generation processor supporting the 32-bit PowerPC RISC instruction set in the early 90ās is being retired with the upcoming Linux 5.10 kernel.
OpenBSD adding modern 64-bit Power BE support
OpenBSD now lists POWER9, specifically on the Raptor hardware, as the supported hardware for its powerpc64. The upcoming OpenBSD 6.8 release, which looks like it is due to come out this month, will be the first release officially supporting powerpc64.
I remember someone from OpenBSD mentioning that it has been very helpful to have bi-endian code to uncover some unexpected bugs; so I wonder how long until we also see a powerpc64le portā¦
Huh, I wonder if we will see a port to Big Endian as well; is not most ARM code written for Big Endian?
Nope. While I was typing this up I read some more, and realised I had it completely backwards:
While both ARM and Power are technically Bi-endian, Power is primarily Big Endian, while ARM is primarily Little Endian.
To encourage porting from x86, Raptor seems to be focusing more on Little Endian on POWER9 (falling under the ppc64le name in Linux-land), but in this case I guess it might also make the ARMāPower porting easier.
I do still hope we might see a Big Endian port as wellā¦
Firstly, I seemingly never posted this link from a few weeks ago;
Phoronix ā 2020-11-11
AMD + IBM Team Up To Tackle Confidential Computing
AMD and IBM are this morning announcing a multi-year, joint development agreement focused on "building upon open-source software, open standards, and open system architectures to drive Confidential Computing in the cloudā¦
I am assuming this will result in some form of standardisation of their respective secure virtual machine systems; IBM calls this the Protected Execution Facility (PEF), while AMD uses the name Secure Encrypted Virtualization (SEV) ā both of these (if I understand correctly) rely partly on encrypted memory to hide what a VM is doing from its hypervisor. On POWER9 silicon that supports it, this is done using the code running in Ultravisor State (akin to a Ring level in x86), while AMD appears to use its AMD Secure Processor component (sometimes compared to the Intel ME).
More recently, there was a spectre-ish-sounding vulnerability disclosed in POWER9 (CVE-2020-4788) and Phoronix has done some testing showing the performance impact:
Phoronix ā 2020-11-25
The Performance Impact To POWER9ās Eager L1d Cache Flushing Fix
Last week a new vulnerability was made public for IBM POWER9 processors resulting in a mitigation of the processorās L1 data cache needing to be flushed between privilege boundaries. ā¦
Raptor had also retweeted someone elseās preliminary testing as well, if you want more data.
I couldnāt find a RISC-V General Discussion / News Thread, so Iām posting this here:
Micro Magic adviser Andy Huang claimed the CPU could produce 13,000 CoreMarks (more on that later) at 5GHz and 1.1V while also putting out 11,000 CoreMarks at 4.25GHzāthe latter all while consuming only 200mW. Huang demonstrated the CPUārunning on an Odroid boardāto EE Times at 4.327GHz/0.8V and 5.19GHz/1.1V.
Thenā¦ make one. I used to make these all the time. This was my fave one, just keep in mind if youāre going to upkeep something like this you have to be FUCKING ON TOP OF IT.
example
Has anyone in here played with Qemu on a powermac G4/5 yet?
Nope. I tried running it on a PS3 for teh LuLz and then immediately regretted it.
That is the wrong architecture.
If reading question right. Using Qemu, off a Powermac G4/G5 CPUā¦ for say Linux?
actually for emulating an openfirmware environment and live editing it
I, for one, am very confused; are you talking about emulating a G4 or G5 system with QEMU, or about running QEMU on a G4 or G5 system?
For ISA reference, the RCS Wiki has a partial table,
https://wiki.raptorcs.com/wiki/Power_ISA
which shows G5 and PS3 to be fairly close in terms of ISA versioning:
PowerPC v2.01 - PPC970 (G5 in Apple branding)
PowerPC v2.02 - Cell PPE (PS3 primary CPU)
Iām talking about running a QEMU virtual environment on a G4 chip or a G5 chip to run openfirmware as I edit it to then later flash it and use it.
Ah, so you want to use QEMU as a testbed for IEEE 1275-1994 (Open Firmware). Neat. I wonder, for something that low-level, can you use virtualisation, or do you really need to just straight emulate to simulate low-level hardware interaction of the firmware?
We were briefly talking about IEEE 1275 firmware last year, in case there is something useful for you there.
Minecraft on POWER9
On an unrelated note, EMue on the RCS Forums has patched LWJGL to be able to run Minecraft (via MultiMC on OpenJDK) on POWER9 machines. It is apparently working in both 4k and 64k kernel page modes, but only in Little Endian mode (ppc64le). He intends to upstream the patches to LWJGL proper, but, ādue to lack of time and motivation (right now) I [EMue] donāt have a fixed time planā.
Trung LĆŖ has a video on YouTube:
Oh, no you can boot it straight off a CD if you wanted to. Itād just be like āIf I do this does it change a color, yes? Cool, now how do I make a menuā
That does not sound very low-level to meā¦
Are you modifying the IEEE 1275 implementation (compiling OpenBIOS/OpenFirmware/OpenBOOT/SLOF/SmartFirmware yourself),
or are you just writing things on top of it in the Forth programming language?
As of RN Iād be editing the image Iād download, I think. Idk yet really. Havenāt had time to just dump into it.