View previous topic :: View next topic |
RAM on your main Gentoo box? |
Less than 512MiB |
|
2% |
[ 2 ] |
512MiB to less than 1GiB |
|
0% |
[ 0 ] |
1GiB to less than 2GiB |
|
1% |
[ 1 ] |
2GiB to less than 4GiB |
|
2% |
[ 2 ] |
4GiB to less than 8GiB |
|
12% |
[ 11 ] |
8GiB to less than 16GiB |
|
23% |
[ 21 ] |
16GiB to less than 64GiB |
|
52% |
[ 47 ] |
64GiB or more |
|
6% |
[ 6 ] |
|
Total Votes : 90 |
|
Author |
Message |
Tony0945 Watchman
Joined: 25 Jul 2006 Posts: 5127 Location: Illinois, USA
|
Posted: Sat Oct 28, 2017 5:34 pm Post subject: |
|
|
Goverp wrote: | FWIW my friend's first mainframe had 4K. Not sure how much disk, possibly none, as this was in the days of TOS. |
It might not have even been semiconductor RAM! It might have been magnetic core. I remember (what a fossil!) when 256 bytes of semiconductor RAM first appeared. This was when the Univac 1108 was God. |
|
Back to top |
|
|
John R. Graham Administrator
Joined: 08 Mar 2005 Posts: 10587 Location: Somewhere over Atlanta, Georgia
|
Posted: Sat Oct 28, 2017 6:01 pm Post subject: |
|
|
The IBM 370/158 I programmed on in college ca. 1978 was a two node cluster (with shared peripherals): one with 8 MiB of magnetic core and the other with 6 MiB of solid state RAM.
- John _________________ I can confirm that I have received between 0 and 499 National Security Letters. |
|
Back to top |
|
|
Tony0945 Watchman
Joined: 25 Jul 2006 Posts: 5127 Location: Illinois, USA
|
Posted: Sat Oct 28, 2017 10:48 pm Post subject: |
|
|
John R. Graham wrote: | The IBM 370/158 I programmed on in college ca. 1978 was a two node cluster (with shared peripherals): one with 8 MiB of magnetic core and the other with 6 MiB of solid state RAM.
- John |
370 didn't even exist when I went to college twelve years before you. Told you I was a fossil. I do remember punching cards on a Model 33 for the IBM 360/30 |
|
Back to top |
|
|
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9601 Location: almost Mile High in the USA
|
Posted: Sun Oct 29, 2017 7:21 am Post subject: |
|
|
Speaking of oddities, pfSense actually drops to mb/sec when not much traffic is going through.
So on the tables, it actually is reporting to me:
mb/sec (yes, millibits; I would hope that blocked/firewalled packets is rated in the millibits/second; also I do not have a native ipv6 uplink so I do not expect ipv6 packets to be passed upwards except as encapsulated 6 over 4 which count as ipv4 packets)
b/sec
kb/sec
Mb/sec
Fortunately it also has a hover-over that describes b/sec as bits/sec so there's no ambiguity; but it does not clarify mbit versus Mbit except that mbit is always in the "minimum" column and can be inferred to be millibits/second this way. _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
Back to top |
|
|
Goverp Veteran
Joined: 07 Mar 2007 Posts: 1966
|
Posted: Sun Oct 29, 2017 10:34 am Post subject: |
|
|
Tony0945 wrote: | [...
It might not have even been semiconductor RAM! It might have been magnetic core.
... |
Indeed, 'twas magnetic core.
OK, lets see if there are any antedeluvians here:
Anyone used a machine with delay line memory?
(Not me, certainly.) _________________ Greybeard |
|
Back to top |
|
|
Tony0945 Watchman
Joined: 25 Jul 2006 Posts: 5127 Location: Illinois, USA
|
Posted: Sun Oct 29, 2017 8:57 pm Post subject: |
|
|
Goverp wrote: |
OK, lets see if there are any antedeluvians here:
Anyone used a machine with delay line memory?
(Not me, certainly.) | Got me there! |
|
Back to top |
|
|
krinn Watchman
Joined: 02 May 2003 Posts: 7470
|
Posted: Mon Oct 30, 2017 10:56 am Post subject: |
|
|
eccerr0r wrote: | mb/sec (yes, millibits; I would hope that blocked/firewalled packets is rated in the millibits/second; also I do not have a native ipv6 uplink so I do not expect ipv6 packets to be passed upwards except as encapsulated 6 over 4 which count as ipv4 packets)
|
Silly option, except to get bigger scale graph to see a bit coming in, but it's kinda too much, just settings the graph scale to bit would display a minimal (even hard to see) bit coming in.
And like i think, it must in fact only handle packet and not really bit going thru the network (nobody really send a raw 1 bit thru its network), it mean a minimal packet size of 64 bytes for ethernetII frame, which make minimal 512 bits, and 512bits display in a graph that have its scale base to a bit is clearly visible.
Using millibits scale, it mean the graph would hit 1000 for a bit, a byte at 8000 and a frame at 512000 (lol).
Even if not use for graph scale, it looks really wanker's option to display 8000millibits/s like it kick ass, when you have just get 1 byte. |
|
Back to top |
|
|
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9601 Location: almost Mile High in the USA
|
Posted: Mon Oct 30, 2017 2:17 pm Post subject: |
|
|
Nah, it's because these are averages over time. pfSense wanted to make sure that you knew something got passed over that period of time. It actually marks categories that had nothing transmitted as 0.000 bytes/sec, but if one ping packet got sent over the course of a minute, that would still round down to 0; but instead it goes to millibytes/sec (instead of bytes/min or bytes/hour) to indicate something went through. _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
Back to top |
|
|
1clue Advocate
Joined: 05 Feb 2006 Posts: 2569
|
Posted: Mon Oct 30, 2017 2:34 pm Post subject: |
|
|
eccerr0r wrote: | Nah, it's because these are averages over time. pfSense wanted to make sure that you knew something got passed over that period of time. It actually marks categories that had nothing transmitted as 0.000 bytes/sec, but if one ping packet got sent over the course of a minute, that would still round down to 0; but instead it goes to millibytes/sec (instead of bytes/min or bytes/hour) to indicate something went through. |
Seems kinda silly. They could put =0 for absolutely no traffic passed, and ~0 if a very small amount of data passed. |
|
Back to top |
|
|
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9601 Location: almost Mile High in the USA
|
Posted: Mon Oct 30, 2017 2:43 pm Post subject: |
|
|
However ~0 does not distinguish between 1 or 2 or even 10 unresponded SYN packets over an hour range; millibits will however, as these significant digits will get shown without changing the time base (so everything on the table is still shown as per second). I think it's reasonable - it's clear and concise versus 0.00001 bits/second. _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
Back to top |
|
|
1clue Advocate
Joined: 05 Feb 2006 Posts: 2569
|
Posted: Mon Oct 30, 2017 2:53 pm Post subject: |
|
|
eccerr0r wrote: | However ~0 does not distinguish between 1 or 2 or even 10 unresponded SYN packets over an hour range; millibits will however, as these significant digits will get shown without changing the time base (so everything on the table is still shown as per second). I think it's reasonable - it's clear and concise versus 0.00001 bits/second. |
Given enough time span, neither would millibits. The point is, no data passed or some very small amount of data passed. |
|
Back to top |
|
|
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9601 Location: almost Mile High in the USA
|
Posted: Mon Oct 30, 2017 3:06 pm Post subject: |
|
|
Zero is significant, but the difference between 1 and 10 can be significant. Though both may round to 0, it's still one order of magnitude apart separating 1 and 10. Sure packet of 64 bytes and one packet of 65 bytes doesn't matter over a large window, but one packet versus 10 packets over a timebox can mean something. _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
Back to top |
|
|
stephan-t Tux's lil' helper
Joined: 12 May 2014 Posts: 122
|
Posted: Wed Nov 01, 2017 5:05 pm Post subject: |
|
|
I'm only have 4 GiB of RAM, but I think for compile (like bigger package as Chromium, libreoffice, Qt5) need such more ram.
Everyday usage enough this little memory size |
|
Back to top |
|
|
eccerr0r Watchman
Joined: 01 Jul 2004 Posts: 9601 Location: almost Mile High in the USA
|
Posted: Wed Nov 01, 2017 6:45 pm Post subject: |
|
|
What part has issues with 4GiB RAM? I really don't think 4GiB is a small amount of RAM; maybe compared to the average in this poll it is, but it's a very usable amount of RAM.
As long as you're not using tmpfs as PORTAGE_TMPDIR I would think they should still build fine? I've seen g++ RSS exceed 300M but even on an 8 core machine, it should fit in 4GB RAM. _________________ Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching? |
|
Back to top |
|
|
John R. Graham Administrator
Joined: 08 Mar 2005 Posts: 10587 Location: Somewhere over Atlanta, Georgia
|
Posted: Wed Nov 01, 2017 8:22 pm Post subject: |
|
|
Agreed. The Chromium main app link phase barely exceeds 2 GiB of allocated RAM, which may reduce the size of the VFS cache but doesn't cause the system to thrash.
However, it isn't as if more isn't useful. For reference, on my 32 GiB, 24-core workstation (with MAKEOPTS=-j25), in the middle of a chromium emerge (at some arbitrary point):- 12.5 GiB used.
- 13.9 GiB buffers / cache.
- 6.4 GiB free.
- 228 KiB swap used.
- John _________________ I can confirm that I have received between 0 and 499 National Security Letters. |
|
Back to top |
|
|
Proinsias Tux's lil' helper
Joined: 06 Oct 2014 Posts: 133 Location: Scotland
|
Posted: Wed Nov 01, 2017 9:39 pm Post subject: |
|
|
8GiB RAM here.
I was using a single 4GiB stick but had to dedicate some of that to the intel graphics. With gentoo and a dual monitor set up I tried various splits but the intel graphics really needed the full 1024MiB to run smoothly when busy, and I found running gentoo with <3GiB available a little tight. With a second stick it reports 6.75GiB and all is well. |
|
Back to top |
|
|
ChalkboardHero n00b
Joined: 11 Nov 2017 Posts: 5
|
Posted: Sat Nov 11, 2017 6:14 pm Post subject: |
|
|
I've got 16GB in my gentoo workstation, but I use it for general computing, NodeJS, Clojure (Euler Problems). I don't know how long I can resist the urge to have bigger numbers though. |
|
Back to top |
|
|
JackHunt n00b
Joined: 21 Aug 2016 Posts: 47 Location: Oxfordshire, England
|
Posted: Mon Nov 13, 2017 11:18 am Post subject: |
|
|
Home machine - 16GB
Lab machine - 64GB |
|
Back to top |
|
|
tuggbuss Apprentice
Joined: 20 Mar 2017 Posts: 222
|
Posted: Wed Nov 22, 2017 8:36 pm Post subject: |
|
|
Laptop Lenovo 430 - 4 GiB
Intel NUC 7i7BNH - 32 GiB
Desktop home build - 64 GiB |
|
Back to top |
|
|
rosomak n00b
Joined: 13 Sep 2006 Posts: 15
|
Posted: Thu Nov 23, 2017 8:44 am Post subject: |
|
|
Jaglover wrote: | I've always wondered why people buy excess RAM. |
1. Usually they don't. The devil is in the use of the word: "excess". Haven't you meant: "I've always wondered why people buy RAM which in my opinion they do not need"?
2. The mechanism works like this:
A) Buying SSD
B) Bye bye swap
C) Hmmm ... what about /tmp (and sometimes /var/tmp) ? Let's see what wiki/gentoo says.
D) Oh, problem of compiling because of lack of space ...
E) How much RAM do I really need? How much money will it cost? Optimal amount - from my point of view - is ...
3. I had 4 or 8 on my HDD laptops. Probably still would. I have 24 on my SSD laptop. I don't think I'm not typical with my view.
PS to D: Seriously. It so happened, that I had to do "zero option" recently. CD from september, stage3 from October, and gcc had to be upgraded to 6. 12GiB was not enough. Obviously I don't have 16GiB (tempfs) or more on day-to-day basis, but I can recall four or five times, when I saw that lack_of_space_error during update. Gentoo: ~amd64 / KDE5.
And, You know, I really love Gentoo, I've had it since 2005 or 2006, after an affair with Debian and a really long relationship with Slack, but it is mature marriage. These days of long, hot nights are, unfortunately, over. |
|
Back to top |
|
|
|