2018-08-24 02:05:39
@Fireduck don't stop with FrostyTrader --- where there's innovation, there's a way, don't get scared off by liability. We should talk through and find a way to address it but it probably is in some way.

PAPACABEZA
2018-08-24 02:06:59
I'm not concerned about civil liability.

Fireduck
2018-08-24 02:07:54
It sounds like qtrade is about a month out so makes sense to wait for them

Fireduck
2018-08-24 03:12:52
integrated with qtrade?

alistar
2018-08-24 03:37:18
http://qtrade.io is an exchange that is working on listing snowblossom

Fireduck
2018-08-24 03:38:01
@Fireduck the slowdown if the whole field is not available… you had a neat formula… something containing (1/2)^6?

penfold
2018-08-24 03:38:11
ran out of scrollback in Slack I think :slightly_smiling_face:

penfold
2018-08-24 03:38:38
well 1/2 assuming 50% available I guess

penfold
2018-08-24 03:40:22
yeah, just basic probabilities

Fireduck
2018-08-24 03:42:33
so is that right, if you had 50% of the field the effective hashrate would be reduced to ~1.56% ?

penfold
2018-08-24 03:44:13
I think I’m getting about 78% efficiency with 128GB RAM, around 122-123 is used for caching the field, so ~95% of the field… ouch :slightly_smiling_face:

penfold
2018-08-24 03:46:32
I guess I need to try Arktika or w/ever it is to fill out that last 5%

penfold
2018-08-24 03:52:43
if you have some percent of the field p, then it is 1/(p^6) is the prob of being able to check a complete pow

Fireduck
2018-08-24 03:53:46
nevermind the 1 over, losing my mind

Fireduck
2018-08-24 03:54:22
anyways, the problem with PoolMiner is that it doesn't discard when it gets a cache miss, it goes to disk and gets it

Fireduck
2018-08-24 03:54:28
which doesn't help you much

Fireduck
2018-08-24 03:54:46
Arktika can discard it (or more correctly, put it on a layer with no or few assigned threads)

Fireduck
2018-08-24 03:58:11
Also, with Arktika you could host those last 6 GB on a separate machine and have them use each other

Fireduck
2018-08-24 06:31:46
I guess the idle GPU with 8GB+ could work too

penfold
2018-08-24 06:32:12
anyone tried GPU yet?

penfold
2018-08-24 06:32:30
6x skein is going to be slow though, I presume the slowdown is too much given the limited RAM

penfold
2018-08-24 10:13:11
You´d need to connect enough GPU´s to get at least close to 128GB, otherwise (p^6) will kill your hashrate

mjay
2018-08-24 10:14:36
I did a few experiments on this, not actual GPU, but just loading the needed data from ram without any hashing, just like it would have to be done when offloading the hashes to gpu. This was only slightly faster than regular CPU hashing

mjay
2018-08-24 10:15:56
I guess its still the memory bandwidth thats limiting here. Had 100% cpu load on my machine, but it actually was the cpu waiting for the memory to respond

mjay
2018-08-24 10:17:09
Copying all data to the GPU would need even more I/O. I don´t think this is faster.

mjay
2018-08-24 10:17:32
Unless someone can get their hands on a 128GB+ GPU. Is there even such a thing?

mjay
2018-08-24 12:42:55
interesting stuff, and no I don’t think as of now there are GPUs with >32GB

penfold
2018-08-24 12:45:14
Even 32GB are way too expensive, like the V100

mjay
2018-08-24 12:46:49
plenty of bandwidth if you batch it like arktika does for network access

Fireduck
2018-08-24 12:47:32
https://github.com/snowblossomcoin/snowblossom/wiki/Mining-Tuning#why-is-remote-faster-than-ssd

Fireduck
2018-08-24 12:47:54
but that would be some fancy gpu programming

Fireduck
2018-08-24 12:51:14
given PCI-e x16 is ~16GB/s it’d just be very expensive RAM methinks :slightly_smiling_face:

penfold
2018-08-24 12:56:56
can do a pretty insane hash rate with that

Fireduck
2018-08-24 12:57:27
oh? isn’t RAM faster? I thought I was seeing ~30GB/s on DDR4

penfold
2018-08-24 12:58:03
yeah, but you are absolutely cpu bound way before that

Fireduck
2018-08-24 12:58:21
ohh right, hmm, that’d be a nice stat to output in the miner

penfold
2018-08-24 12:58:37
what is it called? memory bandwidth? or more genreally, I/O bandwidth?

penfold
2018-08-24 12:58:59
there is a fake mode in arktika to see what the cpu limit is

Fireduck
2018-08-24 12:59:10
I guess I really need to download arktika :slightly_smiling_face:

penfold
2018-08-24 12:59:28
not sure where I got 30GB/s from I only get 7-8GB/s

penfold
2018-08-24 13:00:12
I think I get 24 GB/s on a gce instance

Fireduck
2018-08-24 13:00:13
well, that’s using `mbw`, whatever that measures :smile:

penfold
2018-08-24 13:00:19
but not sure

Fireduck
2018-08-24 13:00:21
oh maybe that’s what it was, I did test on GCE

penfold
2018-08-24 13:08:49
do you also get 24GB/s with 16-byte-random-access?

mjay
2018-08-24 13:08:57
for me these numbers are a lot lower

mjay
2018-08-24 13:10:42
no but I’d like to know what those numbers are… any tools that can do it? or had to roll your own?

penfold
2018-08-24 13:11:29
I think the peak was large block memcpy using `mbw`

penfold
2018-08-24 13:11:56
I just modified the java miner, about 500MB/s on a 40 core 4 socket server

mjay
2018-08-24 13:12:08
this is just counting the 16 bytes each

mjay
2018-08-24 13:12:22
18GB/s with memcpy otherwise

mjay
2018-08-24 13:12:34
wow, quite stark

penfold
2018-08-24 13:12:47
would that have two (or more) memory channels per socket?

penfold
2018-08-24 13:13:30
I wonder, is it possible to calculate effective I/O bandwidth if you know the hashrate?

penfold
2018-08-24 13:13:53
I suppose it’s just 16*6*hashrate

penfold
2018-08-24 13:14:43
meaning I’m getting about 50MB/s but <80% efficiency (not quite enough RAM) so maybe ~64MB/s is possible for my hardware

penfold
2018-08-24 13:15:43
so maxing out PCI-e x16 would be a 250x improvement

penfold
2018-08-24 13:16:46
to get enough memory though I’d need to drop some to x8 and/or x4

penfold
2018-08-24 13:16:58
or, limit to 2-3 GPUs per motherboard I suppose

penfold
2018-08-24 13:17:03
gets expensive quickly

penfold
2018-08-24 13:20:10
This server has 2 memory channels per CPU, thats 8 total. DDR4

mjay
2018-08-24 13:20:24
nice

penfold
2018-08-24 13:21:21
With server hardware you can easily connect 8 GPUs, each 16x

mjay
2018-08-24 13:21:39
this gets expensive, yes :wink:

mjay
2018-08-24 13:21:51
laugh, yep, plus I’m not keen on having a jet engine in the same room as me :slightly_smiling_face:

penfold
2018-08-24 13:22:17
You dont sleep next to these things

mjay
2018-08-24 13:22:23
or: you cant.

mjay
2018-08-24 13:23:06
haha, yes… I don’t intend to but the thing is, using a couple of appliances + a rig in one part of the house will trip the breaker… so I run it in part of the house that while in a separate room, is still audible at night with the door open… thin walls X)

penfold
2018-08-24 13:23:42
glad I have a cellar to do this

mjay
2018-08-24 13:23:53
big walls, several doors, always cool

mjay
2018-08-24 13:24:11
also very humid, but hey :smile:

mjay
2018-08-24 13:24:33
maybe the hardware will be obsolete before it corrodes too badly :slightly_smiling_face:

penfold
2018-08-24 14:16:08
keep it hot and the relative humidity is lower

Fireduck
2018-08-24 17:29:18
@mjay one would have to do something silly like writing an in-gpu nvme direct driver, but that’d still not be significantly quick

Rotonen
2018-08-24 17:30:07
@Rotonen Imagine the GPU working on a bunch of PoW units and filling a buffer with the offsets of the words it needs from the snowfield.

Fireduck
2018-08-24 17:30:32
Then the CPU side reads that buffer in chunks and fills in those words in another buffer and sends the results back.

Fireduck
2018-08-24 17:30:53
Basically can offload the hash processing to the GPU and make use of the CPUs memory bus bandwidth.

Fireduck
2018-08-24 17:31:11
i have imagined this

Rotonen
2018-08-24 17:31:19
it has not turned me on

Rotonen
2018-08-24 17:31:32
haha, then what do you hang around here for? :wink:

Fireduck
2018-08-24 17:31:53
8TB fields

Rotonen
2018-08-24 17:32:19
ok, good

Fireduck
2018-08-24 17:32:24
I have the 4tb field generated

Fireduck
2018-08-24 17:33:53
@mjay if you squeeze more than 300kH/s per memory channel from RAM, do tell

Rotonen
2018-08-24 17:34:23
high end NVMe goes up to close to 200kH/s per disk

Rotonen
2018-08-24 17:34:43
-> for now AMD EPYC is the thing

Rotonen
2018-08-24 17:48:54
@Fireduck I'm prepared to seed a lot more

Clueless
2018-08-24 17:49:36
make sure you have the things there, especially chunked for 7,8,9

Fireduck
2018-08-24 17:49:55
you can also grab https://snowblossom.org/snowfields/snowblossom.12.torrent if you are into it

Fireduck
2018-08-24 17:54:23
@Fireduck
I haven't setup any of the chucked ones yet, nor monitors for them. I may have to buy some storage if we count those.
Which is fine. I've been looking for an excuse to buy more drives. It'll triple my storage.

Clueless
2018-08-24 18:17:58
I don't think your monitors exist

Fireduck
2018-08-24 18:19:27
gradle gradle gradle, I made you out of clay

Fireduck
2018-08-24 18:19:35
Why are there so many terrible build systems

Fireduck
2018-08-24 19:43:08
Would you expect the following to make a shortcut that will open node.sh?

THX 1138 4EB
2018-08-24 19:43:12
[Desktop Entry]
Encoding=UTF-8
Type=Application
Name=Snowblossom node
Name[en_GB]=Snowblossom node
Icon=/home/thx11384eb/Downloads/snowflake_logo.png
Exec=/home/thx11384eb/snowblossom-1.3.0/node.sh
Comment[en_GB]=
Terminal=true
StartupNotify=true
X-KeepTerminal=true

THX 1138 4EB
2018-08-24 19:44:08
When I click on the shortcut, the cursor gets a sand timer but no terminal window opens

THX 1138 4EB
2018-08-24 20:29:13
I have no idea what that configuration notation is. Some sort of window manager specific thing? It looks reasonable. The only thing suspect is that node.sh assumes you are in working directory that node.sh is in. Maybe specify a working directory or put a 'cd' into the start of node.sh ?

Fireduck
2018-08-24 21:21:36
that's windows `.lnk`

Rotonen
2018-08-24 21:21:41
i guess

Rotonen
2018-08-24 21:23:26
yeah, but it is a linux path

Fireduck
2018-08-24 21:25:26
gnome?

Rotonen
2018-08-24 21:28:08
`https://<redacted>/video/search?search=gnome`

Rotonen
2018-08-24 21:34:12
The movie I linked is not worth searching for, watching or buying

Fireduck
2018-08-24 22:30:02
@THX 1138 4EB
It should work, but you'd have to install snowblossom as your own user.

Clueless
2018-08-24 23:20:39
@Fireduck @THX 1138 4EB
So typically the scripts I provided defaulted/leaned towards creating a dedicated system user/service. This favors servers and such... but..
I should probably improve upon the two avenues of install. Headless, and headed. So gui/windows/ubuntu users can point and click, and more easily run as a GUI application.

Clueless
2018-08-24 23:21:08
I think I'll do that built into my Python GUI run.

Clueless
2018-08-24 23:28:58
@Clueless systemd has user services

Rotonen
2018-08-24 23:59:55
I usually run new suspicious crypto coins in a terminal under screen

Fireduck