Thread 107163655

AndJeff Geerling made a Career falling for it again and again.

risc-v but corporate proprietary slop without a bios so youre at the whims of manufacturers upstreaming hardware profiles

>>107163841
(Which they never do)

>>107163841
riscv isn't any better

>>107163655 (OP)
It depends how much you value efficiency. ARM shackles you with soldered-everything but it gives you great performance/watt. For example most phones seem to be able to encode preset 8 AV1 full 1080p video at 20-40 FPS. Slow compared to 100W desktop CPUs but when you factor in that it's only using a few watts of electricity to achieve this it's pretty cool.

Most people probably wouldn't care and the fact that there isn't a big phone video encoding scene pretty much supports this. I mean running a 100W CPU all day long will barely make a dent in the electricity bill.

>>107163940
are you actually using the cpu there or is it accelerating with the gpu?

>>107163995
Nope, SVT-AV1 through ffmpeg. It's honestly absurd how much they keep improving this software AV1 encoder. Like it doesn't NEED to be so fast people can now rip up blu-rays on their phone buy it's now a legitimately possible thing you can do now lol.

>>107164026
wew

The shitty version of Windows on ARM is getting in the way of Nvidia making better ARM CPUs.

Become A better programmer on the pentium you can do anything

>>107163655 (OP)
>>107163841
I'm confused. Are you dissing arm, risc-v, or just hate risc architecture whatsoever?

>>107164615
It's more addressing what you have with x86 that you don't have with other architectures.

X86 is designed to be software agnostic, the hardware must work with software that already exists.
Literally every other architecture doesn't have this expectation leading to bullshit like custom kernels and device trees for every device, even for devices that share the same exact CPU but are made by different companies. 99% of the time if you want to install an OS you need a specific OS image for that hardware.
If the hardware manufacturer drops support and there is no one specifically developing an OS image for your device, you're kinda screwed.

Whatever merits a RISC architecture may have, it's severely dampened by the market reality. Arm devices already suffer from this. RISC-V devices are going to be just as bad or worse.

>>107163655 (OP)
>Jeff Geerling
The guy that makes yet another worthless rpi project in a 10 minute long yt video with 2 minutes of ads and 3 minutes of unfunny jokes.
Truly the peak of content.

>>107164701
>. Arm devices already suffer from this. RISC-V devices are going to be just as bad or worse.
If ARM is already suffering from this, couldn't RISC-V learn from their mistakes and try to device a way so that doesn't happen to RISC?

>>107165615
RISC-Vs licensing wouldn't allow that

>>107163655 (OP)
Works for Applel, and would have worked for Snapdragon if they had their own OS.
M-series macs are their first worthwhile products since the first iphone

>>107163655 (OP)
This thing is 500 fucking dollars.
The Apple M4 Mini is less expensive and will destroy it in any available benchmark.
For $150 you can go on ebay and get a mini computer in the same form factor, x86, 4 -5 years old that will match it for performance and not be hindered by ARM architecture.
WHO THE FUCK is this for?

Intel and AMD solutions for embedded devices are hot garbage. MIPS and PowerPC were valid, but SGI died and IBM is only interested in big iron government contracts. ARM is now mature, and they have a disruptive business model of licensing the tech. That last bit is what's going to eventually kill Intel.

>>107163655 (OP)
Another Cix P1 product. How long has the orion been shipping and major features still aren't working in mainline. Fuck ARM manufacturers who dump barely working shit onto the market and basically abandon it. No, I'm not running your ancient 4.x or 5.x kernel. Get your fucking drivers in mainline or fuck off.

>>107163655 (OP)
>ARM is such a meme

my m1 max macbook will stomp your gayming laptop

>>107165615
suffering? as far as SoC vendors are concerned this is all working out perfectly

>>107166603
if that is what kills intel then intel will also start licensing designs like arm.
also, intel has gotten much more efficient since lunar lake.

not saying arm is bad but intel still has tricks up their sleeves.
also, they have a bunch of experience.
until recently, they never really had to compete for power efficiency the same way.
things were simpler when their main competitor was AMD and they both worked on the heels of each other.

now with arm, intel and amd both have to adopt more energy-efficient designs which is exactly what they are doing.

>>107164701
There are exceptions, it's just not the default. Look at ARM servers.
https://developer.arm.com/documentation/102677/latest/UEFI-requirements
They have UEFI and "sockets" just like x86 but someone will just need to port it down to the consumers.
>>107165878
RISC-V's licensing is even more lax than ARM, excuse me? Fuck you for not citing your shit.
https://github.com/riscv-non-isa/riscv-uefi
And people are already playing around with quasi socketability.
https://www.cnx-software.com/2022/12/14/sipeed-lm4a-t-head-th1520-risc-v-module-to-power-raspberry-pi-4-competitor-and-cluster-board/

>>107166771
This anon is right! Don't buy this kind of chinked out hardware. Only consider Mainline Linux Support.

>>107163940
This.
>331.7 million units in the fourth quarter of 2024
quick maffs gives us close to a billion phones shipped in 2024 with what OP calls "a meme."

>>107165212
Kek I like the guy but this is valid.
>What if I LE TOOK an affordable board pc and made a GIGACLUSTER out of it xDDD?!

OP being a moron for not linking the video
https://www.youtube.com/watch?v=WXfd0rOOtkg

>>107163841
There is nothing about arm preventing having a bios...

>>107163655 (OP)
i bet that thing costs an arm

>>107167052
They had a chance to do this almost 50 years ago. Instead they sued everyone making x86 compatibles, AMD included. If it weren't for the US judicial system, most of the world would be running on Itanium CPUs right now.

>>107167248
It's not just UEFI, although in this instance Arm does seem to dictate ACPI for servers atleast.
What does bug me however is the differing standards.
SBBR is mandated for servers and EBBR is supposed to be for embedded but edge is also mentioned.
SBBR is very much ripped straight from x86 systems. EBBR still gives the option for u-boot and device trees. I guess the one saving grace is the device tree is meant to be supplied by firmware rather than software but DT is a mess apparently.

>>107163655 (OP)
there's 1 (one) ARM vendor that isn't shit and they have zero (0) drivers because they don't sell their shit and its all proprietary

>>107167916
and who would this be?

>>107163655 (OP)
I don't know what ARM is and at this point I'm too embarassed to ask.

>>107167948
basically
ARM = Romanian
RISCV = Italian
x86 = Tok Pisin
makes sense now im sure

>>107163655 (OP)
He like Linus are the Leo Laporte from TechTV Shill Echo Chambers who should never be trusted, oooh you and pops did a collab video with a Hotdog and an AM Radio Tower.. Fuck off.

Anyone who advocates for true tech independence gets shut down, never trust a leaf.

>>107167948
its an arms race.

>>107168000
>Anyone who advocates for true tech independence gets shut down, never trust a leaf.

Gerling isn't a leaf, he's a pure-blood midwestern american.

He's against abortion, he thinks women are property, and he used to be a minister. He had to quit because he was caught sucking dick (unconfirmed) so now he shills ARM on youtube instead.

>>107163655 (OP)
>Minisforum MS-R1
>The World's First ARM Mini Workstation with UEFI Boot
Okay, this is actually interesting.
>CPU: CIX CP8180 12C/12T 2.6GHz 28W TDP
Wtf is a CP8180? Some Chinesium SoC apparently.
Well, what kind of cores does it have? Is it any good?
>With 12 cores / 12 threads at 2.6GHz, 28W TDP, and 45 TOPS compute including 28.8 TOPS NPU
>MS-R1 comes with the ARM-based CP8180 CPU and true UEFI BIOS support
They don't say. Probably 12 Cortex-A53s or something equally terrible.

>>107168033
Yea was talking about Linus and Leo, but didn't know Jeff was a cocksucker (allegedly) lol yea he and his dad prolly have a lotta holdings, their bs on ARM is like cancer. lol @ sucking dick i'll really never look at him the same now

>>107168111
>first mini workstation with uefi boot
the "mini workstation" bit is doing the heavy lifting there to exclude rpi, rk3588, and ampere.
>>107168111
this isn't the first board with the chip. look at the radxa orion.

>>107168183
rpi does not have UEFI, neither does rk3588 except for random github repos that are probably mostly broken. Ampere does but those are $$$.
>radxa orion
Looks like a CIX P1 with eight A720s and four A520s, so meh. I don't see anything to imply the P1 is the same as the CP8180 though.

>>107168223
>rpi
https://github.com/pftf/RPi4
you may not like the source, but the rpi and rk3588 both have functional uefi implementations. the cp8180 is the p1 chip. same shit.

>>107163655 (OP)
ARM actually rules.

>>107168279
Doesn't exactly look functional, and still no support for the Pi 5 despite it being out for two years.

>>107168223
>>107168279
https://forum.armbian.com/topic/51960-orion-o6-cix-p1/
check the dmesg to see how the chip is IDing itself
https://github.com/radxa-pkg/edk2-cix/releases/tag/1.0.0-2
here's the UEFI bios that minisforum is probably shipping--totally still the first mini workstation though.
>>107168309
https://github.com/worproject/rpi5-uefi
I don't know why you're expecting polished software. This is ARM.

>>107168347
>https://github.com/worproject/rpi5-uefi
>If you are still interested in running UEFI on these boards, please contact Raspberry Pi for further support.
>You may also want to consider using better supported hardware instead
kek

>>107168372
...and that's for rpi, which is one of the best supported arm boards in linux. I like arm stuff, but the hardware support from vendors is extremely bad and I wish it would change.

>>107167948
Phone CPUs, ARM = phone CPU and x86 = computer CPU.

Both have pros/cons but many feel we've gone technologically backwards with phone CPUs since they limit what you can do.