From: steve jenkin <sjenkin@canb.auug.org.au>
To: COFF <coff@tuhs.org>
Subject: [COFF] late 1990's servers vs R-Pi 5: quantifiable Performance vs Cost improvements?
Date: Sun, 1 Oct 2023 16:20:10 +1100 [thread overview]
Message-ID: <BF05A499-CC7C-4A3D-9CBA-C2BB8A1C4327@canb.auug.org.au> (raw)
An Old Farts Question, but answers unrestricted :)
In the late 1990’s I inherited a web hosting site running a number of 300Mhz SPARC SUNs.
Probably 32-bit, didn’t notice then :)
Some were multi-CPU’s + asymmetric memory [ non-uniform memory access (CC-NUMA) ]
We had RAID-5 on a few, probably a hardware controller with Fibre Channel SCSI disks.
LAN ports 100Mbps, IIRC. Don’t think we had 1Gbps switches.
Can’t recall how much RAM or the size of the RAID-5 volume.
I managed to borrow from SUN a couple of drives for 2-3 months & filled all the drive bays for ‘busy time'.
With 300MB drives, at most we had a few GB.
Don’t know the cost of the original hardware - high six or seven figures.
A single additional board with extra CPU’s & DRAM for one machine was A$250k, IIRC.
TB storage & zero ’seek & latency’ with SSD are now cheap and plentiful,
even using “All Flash” Enterprise Storage & SAN’s.
Storage system performance is now 1000x or more, even for cheap M.2 SSD.
Pre-2000, a ‘large’ RAID was GB.
Where did all this new ‘important’ data come from?
Raw CPU speed was once the Prime System Metric, based on an assumption of ‘balanced’ systems.
IO performance and Memory size needed to match the CPU throughput for a desired workload,
not be the “Rate Limiting Step”, because CPU’s were very expensive and their capacity couldn’t be ‘wasted’.
I looked at specs/ benchmarks of the latest R-Pi 5 and it might be ~10,000x cheaper than the SUN machines
while maybe 10x faster.
I never knew the webpages/ second my machines provided,
I had to focus on Application throughput & optimising that :-/
I was wondering if anyone on-list has tracked the Cost/ Performance of systems over the last 25 years.
With Unix / Linux, we really can do “Apples & Apples” comparisons now.
I haven’t done the obvious Internet searches, any comments & pointers appreciated.
============
Raspberry Pi 5 revealed, and it should satisfy your need for speed
No longer super-cheap, but boasts better graphics and swifter storage
<https://www.theregister.com/2023/09/28/raspberry_pi_5_revealed/>
~$150 + PSU & case, cooler.
Raspberry Pi 5 | Review, Performance & Benchmarks
<https://core-electronics.com.au/guides/raspberry-pi/raspberry-pi-5-review-performance-and-benchmarks/>
Benchmark Table
<https://core-electronics.com.au/media/wysiwyg/tutorials/Jaryd/pi-les-go/Benchmark_Table.png>
[ the IO performance is probably to SD-Card ]
64 bit, 4-core, 2.4Ghz,
1GB / 2GB / 4GB / 8GB DRAM
800MHz VideoCore GPU = 2x 4K displays @ 60Hz
single-lane PCI Express 2.0 [ for M.2 SSD ]
2x four-lane 1.5Gbps MIPI transceivers [ camera & display ]
2x USB 3.0 ports,
"RP1 chip reportedly allows for simultaneous 5-gigabit throughput on both the USB 3.0s now."
2x USB 2.0 ports,
1x Gigabit Ethernet,
27W USB-C Power + active cooler (fan)
============
--
Steve Jenkin, IT Systems and Design
0412 786 915 (+61 412 786 915)
PO Box 38, Kippax ACT 2615, AUSTRALIA
mailto:sjenkin@canb.auug.org.au http://members.tip.net.au/~sjenkin
next reply other threads:[~2023-10-01 5:20 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-10-01 5:20 steve jenkin [this message]
2023-10-01 19:22 ` [COFF] " Adam Thornton
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=BF05A499-CC7C-4A3D-9CBA-C2BB8A1C4327@canb.auug.org.au \
--to=sjenkin@canb.auug.org.au \
--cc=coff@tuhs.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).