00:13 |
dllud |
asciilifeform, what's the specific processor on your desktop? Those numbers are tantalizing. |
00:14 |
asciilifeform |
2393SE |
| |
~ 26 minutes ~ |
00:41 |
dllud |
asciilifeform, do you have two of them? |
| |
↖ |
00:41 |
dllud |
I was looking at some synthetic benchmarks and the gap is bigger than 1/2 the speed. |
| |
↖ |
00:41 |
dllud |
https://browser.geekbench.com/v4/cpu/14186745 |
00:41 |
dllud |
https://browser.geekbench.com/v4/cpu/2353732 |
00:41 |
dllud |
They point to 1/3 the speed for single threaded performance. |
00:41 |
dllud |
But the next SoC from RockChip is much closer: https://browser.geekbench.com/v4/cpu/14208552 |
00:41 |
dllud |
If, as it appears, that these benchmarks exacerbate the difference when compared to the real task of compiling stuff, then this RK3399 should be good enough for a laptop. |
| |
↖ |
00:41 |
dllud |
It seems to have the same level of support in mainline Linux as the RK3328: http://opensource.rock-chips.com/wiki_Status_Matrix |
| |
↖ |
00:41 |
dllud |
Hmm… interesting, I was quite outdated about this. |
00:44 |
asciilifeform |
http://logs.nosuchlabs.com/log/asciilifeform/2019-09-21#1000448 << aha |
00:44 |
snsabot |
Logged on 2019-09-21 00:41:33 dllud: asciilifeform, do you have two of them? |
00:45 |
asciilifeform |
http://logs.nosuchlabs.com/log/asciilifeform/2019-09-21#1000449 << usually i find actual programs that i work with every day, to be more interesting measures than e.g. 'dhrystone' |
00:45 |
snsabot |
Logged on 2019-09-21 00:41:34 dllud: I was looking at some synthetic benchmarks and the gap is bigger than 1/2 the speed. |
00:46 |
asciilifeform |
http://logs.nosuchlabs.com/log/asciilifeform/2019-09-21#1000455 << >> http://www.loper-os.org/?p=2295 |
00:46 |
snsabot |
Logged on 2019-09-21 00:41:40 dllud: It seems to have the same level of support in mainline Linux as the RK3328: http://opensource.rock-chips.com/wiki_Status_Matrix |
00:49 |
asciilifeform |
http://logs.nosuchlabs.com/log/asciilifeform/2019-09-21#1000454 << rk3399 is slightly faster but considerably more expensive ( and largely on account of the -- few -- boards it is available on, having questionable bells and whistles, e.g. soldered-on wifi ) |
00:49 |
snsabot |
Logged on 2019-09-21 00:41:39 dllud: If, as it appears, that these benchmarks exacerbate the difference when compared to the real task of compiling stuff, then this RK3399 should be good enough for a laptop. |
00:49 |
asciilifeform |
the c101pa fwiw had the rk3399. |
00:51 |
asciilifeform |
afaik to this day no one offers a rk (either type) board w/ socketed ram or pcie -- which would make it potentially 'workstationable' |
00:56 |
dllud |
Bummer, that was gonna be my next question. And it seems they all have at most 4 GB RAM chips. |
00:56 |
asciilifeform |
afaik no one sells an arm of any description with >4 |
00:57 |
asciilifeform |
with possible exception of one chinese outfit, which offered one-offs at ludicrous golden toilet 4fig prices, that i won't even bother to try and dig up link to again |
00:58 |
dllud |
yeah, nevermind |
00:59 |
asciilifeform |
my main gripe re the rk3328 board, wasn't even the 4g ram cap, but lack of sata ( or any means, aside from the cpu-taxing usb3 , to connect same ) |
01:00 |
asciilifeform |
i use'em as servers ( currently building yet another cluster ) and this is main practical limit on performance. |
01:01 |
dllud |
ahahah exactly! The existence of a SATA port is the reason why I keep using SBCs with the AllWinner A20. |
01:01 |
dllud |
That's why, a few minutes ago this article got me interested: http://linuxgizmos.com/rk3399-based-odroid-n1-sbc-doubles-up-on-sata-iii/ |
01:01 |
asciilifeform |
from my pov, those may as well be intels, 100% blob boot |
01:01 |
asciilifeform |
( the 'allwinner' that is ) |
01:02 |
dllud |
But it seems that Hardkernel discontinued this product. |
01:02 |
asciilifeform |
right, they seem to be out of print |
01:03 |
dllud |
The replacement (ODROID-N2) doesn't have SATA ports. |
01:05 |
dllud |
asciilifeform, do those RKs boot without any proprietary blob? |
01:05 |
dllud |
As for the AllWinners, are you referring to this: https://github.com/qemu/u-boot/blob/master/board/sunxi/README.sunxi64#L155 ? |
01:06 |
asciilifeform |
dllud: 33xx aha |
01:08 |
asciilifeform |
what i haven't even tried, is the video output in 3328. last i knew, wasn't fully functional without shitware. |
01:08 |
asciilifeform |
fortunately my application doesnt use display. |
| |
~ 23 minutes ~ |
01:32 |
dllud |
> Hardkernel announced it was shelving the board due to sourcing problems |
01:32 |
dllud |
http://linuxgizmos.com/odroid-n2-sbc-has-hexa-core-amlogic-s922x-and-63-to-79-price/ |
01:32 |
dllud |
Hm… All RK3399 getting used on Chromebooks? Does Rockchip still produce them? |
01:32 |
dllud |
Any idea of Amlogic blobiness? https://libre.computer/ sells a few boards with those SoCs. |
01:38 |
asciilifeform |
ugh, no usb3 or sata or gb nic |
01:38 |
asciilifeform |
wtf is the use of these |
01:39 |
asciilifeform |
re amlogic -- no idea. last i knew rk was the only reasonably hygienic arm64; it is possible that this has changed. |
01:40 |
asciilifeform |
i find board where impossible to attach fast storage, to be 100% useless tho. |
| |
~ 40 minutes ~ |
02:20 |
dllud |
Yup, same for me. The only interesting board with an Amlogic SoC seems to be the ODROID-N2 (Amlogic SOC). |
02:20 |
dllud |
I am wondering if it can boot without blobs. At least it has plenty of USB3 ports. |
02:23 |
dllud |
But I do hate USB-SATA converters. On my experience they always underperform and disconnect for no reason and such. |
02:23 |
dllud |
Nothing like native SATA. |
02:30 |
dllud |
asciilifeform, it seems that after all you can boot AllWinner without blobs: https://redmine.replicant.us/projects/replicant/wiki/Pinephone#fn7 |
02:30 |
dllud |
I am trying to track down the code. |
02:33 |
dllud |
Source of that info was MoeIcenowy. |
| |
~ 7 hours 13 minutes ~ |
09:47 |
shinohai |
Did you read about this faux phuctorism yesterday asciilifeform? http://archive.is/Qgxck |
09:47 |
shinohai |
Pretty lulzy. |
| |
~ 1 hours 31 minutes ~ |
11:18 |
asciilifeform |
looks pretty tedious |
| |
~ 1 hours 8 minutes ~ |
12:26 |
BingoBoingo |
Breaking undersized keys is not news |
12:35 |
asciilifeform |
BingoBoingo: apparently there's a 'craig wright' type in that story, complete with selling magick elixirs etc. still pretty snoar. |
12:38 |
BingoBoingo |
Kinda why the things hasn't been determined news. Herp derp gonna break keys of a size demonstrated breakable more than a decade ago pls to buy my Quantum Copywriting |