08:44 <Jmabsd> mps,Psydroid,*: Guys, Radxa are saying they will offer their Rock5 as 10gbps Ethernet and 24GB RAM.

08:45 <Jmabsd> Also their USBC PD input will autonegotiate the highest voltage it can, meaning it will hit the 18-20V mode which can go up to 100 watts input.

08:45 <Jmabsd> In the 10x7.2x~5 CM size category this is the most powerful computer available, especially open computer.

08:51 <mps> Jmabsd: what is your intended development? AI?

08:52 <Jmabsd> mps: home server, desktop type

08:52 <Jmabsd> 1000mbps ethernet is too slow. 10gbps gives you nice latency in a network etc.

08:52 <Jmabsd> you can barely sync vacation photos over 1000mbps

08:53 <Jmabsd> 16GB RAM can be enough for a desktop but more is preferable.

08:53 <Jmabsd> i will be really curious to see how fast the CPU is. the benchmarks say 50% faster than RK3399. But, it is a new CPU core and the lithography is 6nm.

08:55 <mps> hm, my old arm32 exynos chromebook with 4GB ram is quite fine as desktop for me

08:55 <Jmabsd> I understand it will not be as fast as an Apple M1, however still quite fast. And the Apple M1 is problematic other ways

08:55 <Jmabsd> great.

08:55 <Jmabsd> RAM also helps as disk cache. you always want more RAM.

08:56 <mps> there is olimex teres-i notebook which is very open, though firefox is somewhat slow on it

08:56 <Jmabsd> mps: one thing I'm saying is that ARM is moving into the desktop/laptop/server category now. which is great. i really dislike the Intel hegemony.

08:56 <Jmabsd> ok

08:56 <Jmabsd> Intel can't be trusted to be safe.

08:57 <mps> yes, I stopped to by intels when they added SME

08:57 <mps> and then switched everything to arm

08:58 <mps> but I'm not sure arm is safe, also

08:59 <mps> safe things in the world are things which doesn't exists

09:07 <Jmabsd> there can be CPU bugs and backdoors, and also firmware and especially service processors can both have 0days and backdoors

09:07 <Jmabsd> indeed CPU:s do not go through formal validation for all functions.

09:08 <Jmabsd> Intel is on the extreme side though, with huge service processor system, huge firmware, huge microcode, all absolutely closed

09:08 <Jmabsd> anyhow in a normal ethics argument, RISCV is more ethical than ARM, and ARM is more ethical than Intel.

09:12 <mps> well, when we come to philosophical field talking about ethic I must say that any technology is unethical

09:14 <mps> all technology however simple or complicated cripple us

09:37 <psydroid> Jmabsd, that is good news, 10gbps should become standard these days. 24 GB of RAM sounds like overkill for that chip, but it can't hurt to have it available when needed

09:38 <psydroid> I only have these Intel laptops, which I hope to relegate to legacy tasks as (more) laptops with ARM and RISC-V chips become available.

09:55 <Jmabsd> psydroid: RAM access is faster than SSD access. and sometimes you want to precalculate stuff in RAM etc. without exposing the SSD to write wear.

09:56 <Jmabsd> psydroid: a highest-end 2017 Intel laptop had 2 cores with 4 threads, and uh was it uh 16GB RAM??

09:56 <Jmabsd> The RK3588 must be as fast as a highest-end 2015 or 2014 Intel laptop really I think. Afterall the RK3588 has 4 quite fast cores, is it 3MB L3 cache etc, and the lithography is so small. Anyhow looking forward to get personal experience soon.

10:01 <Jmabsd> the extra 4 slow cores, how slow are they, how many % expected of the fast cores?

10:02 <Jmabsd> a lot of work is cache/RAM access really. i guess for instance opening/reading files and such other very IO dense operations very well may end up being done quite effectively on the low power cores

10:07 <Jmabsd> psydroid,mps,*: btw also very interesting fact, Radxa guy says the biggest RAM chip on the market is 12GB. for this reason with their SBC layout, the biggest they can go is 24GB. he asked if that is OK. i say sure.

10:14 <mps> huh, I remember that I had computer with 32KB ram :)

10:37 <psydroid> Jmabsd: my most powerful Intel laptops have 4 cores and 24 and 32 GB respectively, but they emit a lot of heat, so I don't usually have them in my lap. I am writing this on an Intel laptop with 2 low-end Celeron ("Atom") cores that are comparable to those from a Raspberry Pi 4.

10:37 <psydroid> Rock 5 is twice as fast per core and there are 4 cores instead of 2. So a laptop with that chip would be faster than everything but these two laptops.

10:38 <psydroid> 32 kB of RAM :)

10:38 <psydroid> I had 128 kB even in my MSX2

10:39 <Jmabsd> psydroid: Rock 5 core is twice as fast as what?

10:40 <Jmabsd> 32KB what?

10:44 <psydroid> Jmabsd, twice as fast per core as Intel Celeron 3060 and Raspberry Pi 4

10:44 <psydroid> N3060*

10:44 <psydroid> 32 kB is what mps mentioned, probably in a home computer

10:45 <mps> psydroid: yes, first home computer ZX

10:46 <Jmabsd> Ah. So you are saying the RK3588 should be around 4-5x faster than a 2016 Celeron. OK. https://www.intel.com/content/www/us/en/products/sku/91832/intel-celeron-processor-n3060-2m-cache-up-to-2-48-ghz/specifications.html

10:46 <psydroid> yes

10:47 <psydroid> but I got this laptop used 2 years ago and only paid €75 for it and have only replaced the default 128 GB with the old 256 GB SSD from my main laptop

10:48 <Jmabsd> ;) OK.

10:48 <Jmabsd> psydroid,mps: i asked Radxa now what PCIe lane they will give to the PCIe ethernet. perhaps they just give it one PCIe v2 lane, that would mean 500MB/sec (5gbps)

10:48 <Jmabsd> that would be OK still

10:48 <psydroid> RK3588 shouldn't be too far from the Skylake is my HP Zbook Studio G3

10:49 <Jmabsd> latency will be low etc. any wired PCIe ethernet is superior to USB

10:49 <psydroid> in*

10:49 <psydroid> yes, that's why I'm interested in Rockchip's roadmap

10:49 <psydroid> RK3588 is already going to be great, but what will be even greater?

10:51 <Jmabsd> yea

10:51 <Jmabsd> anyhow the RK3588 appears to represent a breaking point

10:51 <Jmabsd> when ARM became cheap and kind of reasonable performance interval

10:51 <Jmabsd> (The LX2 has already accomplished this but it's a bit costly and demanding like get your own Radeon etc.)

10:52 <Jmabsd> the RK3588 will feed 8K@60Hz. really wholesome.

11:00 <psydroid> we're talking about a chip that was supposed to become available 2 years ago, so things haven't been moving as fast as expected

11:00 <psydroid> it's really nice that powerful chips come with phones, but in general phones aren't as useful to me as laptops and SBCs

13:54 <Jmabsd> psydroid: I guess to be a useful computer, you need a good screen, good keyboard, good mouse. HDMI and USB can solve that

13:55 <Jmabsd> Are there some small fans that can help cool an RK3588 phone? Ha ha

13:55 <Jmabsd> By the way did anyone 3D print smartphones yet ha ha

13:56 <Jmabsd> psydroid: A shame re. taking much time yea. Though also Covid caused a lot of delays.

14:05 <psydroid> Jmabsd, phones can be useful as computers when they run operating systems that serve their owners, but even that development seems marginal with mobile devices such as the Pinephone

14:14 <Jmabsd> jakllsch: a pm

14:14 <jakllsch> i saw that

14:14 <Tenkawa> my odroid-m1 will be here Wed... we have already started building a replacement os image to not use petitboot

14:15 <Tenkawa> I wish the 3588's would get released so mine would get shipped

14:15 <Tenkawa> heheh

16:39 <Jmabsd> Tenkawa: Radxa ship in July.

16:47 <Tenkawa> Cool

16:47 <Tenkawa> BBl..

16:47 <Tenkawa> Powerline down in backyard :(

21:12 <robmur01> macc24: just in case you didn't figure out the fan thing in the meantime, see rk3399-nanopc-t4.dts for an example

21:13 <macc24> robmur01: stuck on the dsi display ,_,

21:13 <macc24> thanks

21:13 <robmur01> ah, right