1 |
On Wed, Aug 30, 2017 at 9:49 AM, Daniel Frey <djqfrey@×××××.com> wrote: |
2 |
> On 08/29/2017 08:09 PM, R0b0t1 wrote: |
3 |
>> On Tue, Aug 29, 2017 at 12:07 AM, Dale <rdalek1967@×××××.com> wrote: |
4 |
>>> Rich Freeman wrote: |
5 |
>>> |
6 |
>>> Isn't it amazing how efficient and fast newer computers are? It's |
7 |
>>> almost worth the energy saving to upgrade. If a person runs their |
8 |
>>> system 24/7, that is even more reason. |
9 |
>>> |
10 |
>> |
11 |
>> It's something I used to bring up in #gentoo when people would come in |
12 |
>> and ask or complain about compilation times. The amortization period |
13 |
>> for computers is generally 3 years. If you use them longer, you are |
14 |
>> (in theory) losing money relative to your competitors. |
15 |
>> |
16 |
>> As a home user, your time and energy budgets might not be so tight, |
17 |
>> but the lack of stress is worth a nice desktop for compiling your |
18 |
>> software. |
19 |
>> |
20 |
>> |
21 |
> |
22 |
> I'm sort of on the fence for now. I bought my computer in 2008 (but an |
23 |
> expensive processor, QX9650, with 8GB of RAM, very $$$ in 2008) and it |
24 |
> still is running fine today. It still compiles reasonably quickly, as I |
25 |
> use distcc with one other computer) and am still on spinning rust, but |
26 |
> in RAID. My mythtv frontends have SSDs and every one has had the SSD |
27 |
> replaced at least once. |
28 |
> |
29 |
|
30 |
A decade was about how long I expect my i7-4770k based system to last, |
31 |
assuming I have absolutely no money to upgrade before then. |
32 |
|
33 |
> My main frontend in the living room has shown problems (hanging while |
34 |
> playing back recordings, and I mean a hard-lock... no ssh, no response |
35 |
> at all, and the backend had in its logs "where'd it go? closing |
36 |
> connection") and so I think the hardware may finally be at the end of |
37 |
> its life. That was also built in 2008. |
38 |
> |
39 |
> I'm not going to throw a working computer out (well, recycle it) if I |
40 |
> don't have to. Power is cheap here. |
41 |
> |
42 |
> I have concerns about the backend (Q6600) and frontend (E7500), as they |
43 |
> were both built in 2008, but the thing is, they're just as fast as I |
44 |
> remember them when built new. |
45 |
> |
46 |
> For the frontend replacement I think I'm going to jump to one of the |
47 |
> Ryzen products. I don't need ThreadRipper there, but one of their other |
48 |
> processors will work. The backend will get a faster processor but it's |
49 |
> part of my distcc cluster. |
50 |
> |
51 |
> I think Dale posted a libreoffice build speed, mine isn't so bad either, |
52 |
> but I'm using RAID and distcc: |
53 |
> |
54 |
> $ genlop -t libreoffice |
55 |
> * app-office/libreoffice |
56 |
> |
57 |
> Tue Aug 1 08:45:28 2017 >>> app-office/libreoffice-5.2.7.2 |
58 |
> merge time: 1 hour, 22 minutes and 53 seconds. |
59 |
> |
60 |
> I *really* hate the climate nowadays of toss it out when it acts up/gets |
61 |
> slow. |
62 |
> |
63 |
|
64 |
Like I mention in another thread (and like Rich touches on) power |
65 |
savings can be an incentive to upgrade, besides the increase in speed. |
66 |
Power efficiency and speed generally increase in multiples greater |
67 |
than one, so you are reducing the cost and time of compilations or |
68 |
general use by a lot in the end. |
69 |
|
70 |
Look at Alan's quoted build times for an example. |
71 |
|
72 |
R0b0t1. |