From george at ceetonetechnology.com  Tue May  2 11:57:19 2023
From: george at ceetonetechnology.com (George Rosamond)
Date: Tue, 2 May 2023 11:57:19 -0400
Subject: [talk] NYC*BUG tomorrow/Wednesday: Ori on GEFS
Message-ID: <8c8df4d6-75fa-9eff-f7e4-b954f4ed7c3f@ceetonetechnology.com>

GEFS, A Good Enough File System, Ori Bernstein

2023-05-03 @ 18:45 - Five Mile Stone at 1640 2nd Ave (northeast Corner
of 2nd Ave and 85th St, 2nd floor). Please note the stairs to the second
floor are on the north wall as you enter from 2nd Ave.

GEFS is an experimental file system built for Plan 9. It aims to be a
crash-safe, corruption-detecting, simple, and fast, snapshotting file
system, in that order. It's built on top of a relatively new data
structure known as a B? tree. Ori will be talking about how it works
internally, and his ambitions to port it to OpenBSD.

Speaker Biography

Ori is 100% human, or triple your money back.



From george at ceetonetechnology.com  Wed May  3 12:37:16 2023
From: george at ceetonetechnology.com (George Rosamond)
Date: Wed, 3 May 2023 12:37:16 -0400
Subject: [talk] NYC*BUG TONIGHT: Ori on GEFS
Message-ID: <c570fb69-0519-8a23-a81a-2472f5a5ef5b@ceetonetechnology.com>

GEFS, A Good Enough File System, Ori Bernstein

2023-05-03 @ 18:45 - Five Mile Stone at 1640 2nd Ave (northeast Corner
of 2nd Ave and 85th St, 2nd floor). Please note the stairs to the second
floor are on the north wall as you enter from 2nd Ave.

GEFS is an experimental file system built for Plan 9. It aims to be a
crash-safe, corruption-detecting, simple, and fast, snapshotting file
system, in that order. It's built on top of a relatively new data
structure known as a B? tree. Ori will be talking about how it works
internally, and his ambitions to port it to OpenBSD.

Speaker Biography

Ori is 100% human, or triple your money back.



From bonsaime at gmail.com  Wed May  3 14:39:17 2023
From: bonsaime at gmail.com (Jesse Callaway)
Date: Wed, 3 May 2023 11:39:17 -0700
Subject: [talk] NYC*BUG TONIGHT: Ori on GEFS
In-Reply-To: <c570fb69-0519-8a23-a81a-2472f5a5ef5b@ceetonetechnology.com>
References: <c570fb69-0519-8a23-a81a-2472f5a5ef5b@ceetonetechnology.com>
Message-ID: <CAJU5ZBNZhG-3c6VoODS37oDsL=yQBwQGsBynkyRJ5qqJ--5EgA@mail.gmail.com>

On Wed, May 3, 2023 at 9:37?AM George Rosamond <george at ceetonetechnology.com>
wrote:

> GEFS, A Good Enough File System, Ori Bernstein
>
> 2023-05-03 @ 18:45 - Five Mile Stone at 1640 2nd Ave (northeast Corner
> of 2nd Ave and 85th St, 2nd floor). Please note the stairs to the second
> floor are on the north wall as you enter from 2nd Ave.
>
> GEFS is an experimental file system built for Plan 9. It aims to be a
> crash-safe, corruption-detecting, simple, and fast, snapshotting file
> system, in that order. It's built on top of a relatively new data
> structure known as a B? tree. Ori will be talking about how it works
> internally, and his ambitions to port it to OpenBSD.
>
> Speaker Biography
>
> Ori is 100% human, or triple your money back.
>
> _______________________________________________
> talk mailing list
> talk at lists.nycbug.org
> https://lists.nycbug.org:8443/mailman/listinfo/talk
>

Great looking talk! I did a quick search of B-epsilon and found this
paper... sitting down with my lunch about to read it.

An Introduction to B?-trees and Write-Optimization -  Michael A. Bender,
Martin Farach-Colton, William Jannen, Rob Johnson, Bradley C. Kuszmaul,
Donald E. Porter, Jun Yuan, and Yang Zhan
<http://supertech.csail.mit.edu/papers/BenderFaJa15.pdf>
-- 
-jesse
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.nycbug.org:8443/pipermail/talk/attachments/20230503/4f553365/attachment.htm>

From mcevoy.pat at gmail.com  Thu May  4 09:09:15 2023
From: mcevoy.pat at gmail.com (Pat McEvoy)
Date: Thu, 4 May 2023 09:09:15 -0400
Subject: [talk] Video for last night posted
Message-ID: <207401AD-27AF-4478-821E-76AC7634865C@gmail.com>

I posted the video to both Peertube and YouTube. Links below. 

https://toobnix.org/w/17JKjqWTPfiSksA5QUkaeN

https://youtu.be/juFndFy72gI

Links will be posted to NYC*Bug talk page shortly. 



From george at ceetonetechnology.com  Thu May  4 12:41:38 2023
From: george at ceetonetechnology.com (George Rosamond)
Date: Thu, 4 May 2023 12:41:38 -0400
Subject: [talk] Cloud exit strategy
Message-ID: <c324cd20-7965-9e61-5788-501724ebc7b3@ceetonetechnology.com>

 Why a "Cloud Exit Strategy" is essential to enable the future

https://kpmg.com/uk/en/blogs/home/posts/2023/04/why-a--cloud-exit-strategy--is-essential-to-enable-the-future.html

Nice phrase "Cloud repatriation".

Seems to be increasing in mainstream channels, not just among loopy
cults like NYC*BUG and #metabug.

g



From pete at nomadlogic.org  Thu May  4 13:02:00 2023
From: pete at nomadlogic.org (Pete Wright)
Date: Thu, 4 May 2023 17:02:00 +0000
Subject: [talk] Cloud exit strategy
In-Reply-To: <c324cd20-7965-9e61-5788-501724ebc7b3@ceetonetechnology.com>
References: <c324cd20-7965-9e61-5788-501724ebc7b3@ceetonetechnology.com>
Message-ID: <jdevm6hj3k6tmkglgttxfzejj3xasbs7h52siujlrqj6lhqwey@vhpuekk63h2f>

On Thu, May 04, 2023 at 12:41:38PM -0400, George Rosamond wrote:
>  Why a "Cloud Exit Strategy" is essential to enable the future
> 
> https://kpmg.com/uk/en/blogs/home/posts/2023/04/why-a--cloud-exit-strategy--is-essential-to-enable-the-future.html
> 
> Nice phrase "Cloud repatriation".
> 
> Seems to be increasing in mainstream channels, not just among loopy
> cults like NYC*BUG and #metabug.
now if only there was some global consultancy that could help me realize
this, then we'd be on to something :p

-p

-- 
Pete Wright
pete at nomadlogic.org



From george at ceetonetechnology.com  Thu May  4 18:06:34 2023
From: george at ceetonetechnology.com (George Rosamond)
Date: Thu, 4 May 2023 18:06:34 -0400
Subject: [talk] Ori meeting last night
Message-ID: <e16f013b-2e31-475c-9f04-125b003cb0f2@ceetonetechnology.com>

It went very well.. prob hit around 17 people with a lot of new faces.

It was a significant step in the right direction. We need to continue to
map out future meetings and keep the momentum going. The future meeting
section on the www site needs to be populated a few months into the future.

I know there has been chatter about another NYCBSDCon. It's way too
early in the post-pandemic period (post?..) to plan anything like that.
We need to again have regular vibrant meetings with a lot of hands ready
to be involved. We are not there yet.

There is clearly a massive vacuum in user group activity in NYC, which
makes it all hard to read. Are we the last ones standing and the only
non-corporate events around? Is the past era of NYC user groups done? I
tend to think it's more the former, but it doesn't mean we jump into
grand plans. Let's build some basic consistency and see where it leads.

The one persistent issue to me is that space. We are in the second floor
of a bar and there is a lot of noise coming from the other side of the
floor, although Ori had no issues and it seems to have worked last night.

Ideally space in downtown Manhattan would work if possible, but maybe
because that's more convenient to those in Brooklyn.

If anyone has ideas on space, ping admin@ offline.

Huge credit (again) to Patrick who got things moving.. and found that
space. Plus getting the video online.. everyone says they are going to
do things like that, but in practice, Patrick always delivers.

Also if anyone has more proposals for meetings, ping admin@ or post your
idea to talk@, better yet.  We have some discussions and plans but we
always need more. If you haven't presented before, consider NYC*BUG your
launch pad. A number of people have done their first meetings at
NYC*BUG, then became fixtures at a number of other events.

It's almost always a relaxed environment for speakers, well, outside of
the Puppet meeting so many years ago...

g



From ori at eigenstate.org  Thu May  4 23:46:14 2023
From: ori at eigenstate.org (Ori Bernstein)
Date: Thu, 4 May 2023 23:46:14 -0400
Subject: [talk] NYC*BUG TONIGHT: Ori on GEFS
In-Reply-To: <CAJU5ZBNZhG-3c6VoODS37oDsL=yQBwQGsBynkyRJ5qqJ--5EgA@mail.gmail.com>
References: <c570fb69-0519-8a23-a81a-2472f5a5ef5b@ceetonetechnology.com>
 <CAJU5ZBNZhG-3c6VoODS37oDsL=yQBwQGsBynkyRJ5qqJ--5EgA@mail.gmail.com>
Message-ID: <20230504234614.3ef14bcc678ed47d37402b3c@eigenstate.org>

On Wed, 3 May 2023 11:39:17 -0700, Jesse Callaway <bonsaime at gmail.com> wrote:

> Great looking talk! I did a quick search of B-epsilon and found this
> paper... sitting down with my lunch about to read it.
> 
> An Introduction to B?-trees and Write-Optimization -  Michael A. Bender,
> Martin Farach-Colton, William Jannen, Rob Johnson, Bradley C. Kuszmaul,
> Donald E. Porter, Jun Yuan, and Yang Zhan
> <http://supertech.csail.mit.edu/papers/BenderFaJa15.pdf>


That's the one.

Another thing to look at is https://www.betrfs.org/; it's got
a lot of interesting ideas, and influenced gefs quite a bit.

-- 
    Ori Bernstein



From nonesuch at longcount.org  Mon May  8 07:31:57 2023
From: nonesuch at longcount.org (Mark Saad)
Date: Mon, 8 May 2023 07:31:57 -0400
Subject: [talk] Uptick in spam
Message-ID: <FBCAF3BE-4FDC-40B4-800B-31DAC43A795A@longcount.org>

Hi everybody,
   I wanted to see if anyone is seeing this . A number of accounts of mine and family have seen an odd uptick in spam all with the same format . 

Example subject : Y?o H?av?e a B?i?g?g C?ha?n?ce To ?re?ce?iv?e A f?r?e?e C?o?ff?e?e M?ak?er?==The=20May=20Diamabrush=20Download=20=F0=9F=92=8E

Then some odd image that may match the subject and then some other garbage .


What?s piqued my interest was a the bit that looks like a MAC address at the end of the subject . Any ideas ? 

---
Mark Saad | nonesuch at longcount.org


From ahpook at mitnal.com  Wed May 10 00:47:36 2023
From: ahpook at mitnal.com (Ah Pook)
Date: Wed, 10 May 2023 00:47:36 -0400
Subject: [talk] Uptick in spam
In-Reply-To: <FBCAF3BE-4FDC-40B4-800B-31DAC43A795A@longcount.org>
References: <FBCAF3BE-4FDC-40B4-800B-31DAC43A795A@longcount.org>
Message-ID: <6abc4c17-34c2-738e-fd64-9d675292b314@strawandfrost.com>

Unicode diamond?  ??
F0 9F 92 8E
U+1F48E

On 5/8/23 07:31, Mark Saad wrote:
> Hi everybody,
>     I wanted to see if anyone is seeing this . A number of accounts of mine and family have seen an odd uptick in spam all with the same format .
> 
> Example subject : Y?o H?av?e a B?i?g?g C?ha?n?ce To ?re?ce?iv?e A f?r?e?e C?o?ff?e?e M?ak?er?==The=20May=20Diamabrush=20Download=20=F0=9F=92=8E
> 
> Then some odd image that may match the subject and then some other garbage .
> 
> 
> What?s piqued my interest was a the bit that looks like a MAC address at the end of the subject . Any ideas ?
> 
> ---
> Mark Saad | nonesuch at longcount.org



From nonesuch at longcount.org  Wed May 10 06:54:48 2023
From: nonesuch at longcount.org (Mark Saad)
Date: Wed, 10 May 2023 06:54:48 -0400
Subject: [talk] Uptick in spam
In-Reply-To: <6abc4c17-34c2-738e-fd64-9d675292b314@strawandfrost.com>
References: <6abc4c17-34c2-738e-fd64-9d675292b314@strawandfrost.com>
Message-ID: <9030765C-D790-4C72-AC2F-9FF00884AE07@longcount.org>

Ah
   Thanks for that It never occurred to me that it was some mangled Unicode . I am getting hammered by it from all over the place .  

---
Mark Saad | nonesuch at longcount.org

> On May 10, 2023, at 12:47 AM, Ah Pook <ahpook at mitnal.com> wrote:
> 
> ?Unicode diamond?  ??
> F0 9F 92 8E
> U+1F48E
> 
>> On 5/8/23 07:31, Mark Saad wrote:
>> Hi everybody,
>>    I wanted to see if anyone is seeing this . A number of accounts of mine and family have seen an odd uptick in spam all with the same format .
>> Example subject : Y?o H?av?e a B?i?g?g C?ha?n?ce To ?re?ce?iv?e A f?r?e?e C?o?ff?e?e M?ak?er?==The=20May=20Diamabrush=20Download=20=F0=9F=92=8E
>> Then some odd image that may match the subject and then some other garbage .
>> What?s piqued my interest was a the bit that looks like a MAC address at the end of the subject . Any ideas ?
>> ---
>> Mark Saad | nonesuch at longcount.org
> 



From george at ceetonetechnology.com  Fri May 12 14:09:45 2023
From: george at ceetonetechnology.com (George Rosamond)
Date: Fri, 12 May 2023 14:09:45 -0400
Subject: [talk] APU2 post-mortem options
Message-ID: <70c52c53-e0e4-3bc8-d4fb-112254240993@ceetonetechnology.com>

There's been some chatter on IRC and other places about ideal hardware
options after APU2/PCEngines go away.

amd64 remains the best-supported architecture, and aarch64 or RISC-V
especially are not there yet. Oh, the dreams we all had years ago about
aarch64...

On that note, what similiar small fanless hardware are others looking at?

There's the Qotom stuff, which I think ProtectCLI uses. And there's a
variety of very small desktops (ThinkCentre M600, eg) that are nice and
cheap, but one NIC and have useless things like HDMI, bluetooth and wifi
which aren't ideal for servers. Plus idk about db9/serial...

>1u in height is very common with those little desktops.

So I guess I have two questions:

Is it still amd64 or bust for the best stability and support?

And then, what comes after APU2 for the many who use it?

g



From spork at bway.net  Fri May 12 14:22:38 2023
From: spork at bway.net (Charles Sprickman)
Date: Fri, 12 May 2023 14:22:38 -0400
Subject: [talk] APU2 post-mortem options
In-Reply-To: <70c52c53-e0e4-3bc8-d4fb-112254240993@ceetonetechnology.com>
References: <70c52c53-e0e4-3bc8-d4fb-112254240993@ceetonetechnology.com>
Message-ID: <926BE240-1C63-4DA8-BE6D-7FD29D38AA3A@bway.net>


> On May 12, 2023, at 2:09 PM, George Rosamond <george at ceetonetechnology.com> wrote:
> 
> There's been some chatter on IRC and other places about ideal hardware
> options after APU2/PCEngines go away.
> 
> amd64 remains the best-supported architecture, and aarch64 or RISC-V
> especially are not there yet. Oh, the dreams we all had years ago about
> aarch64...
> 
> On that note, what similiar small fanless hardware are others looking at?
> 
> There's the Qotom stuff, which I think ProtectCLI uses.

Of note, the Protectli stuff also seems to have a custom BIOS which supports
console redirection (including getting into the BIOS) which can be handy.

Some of their 4-port units have an odd internal USB to serial setup that
only has support for Linux and Windows (more info on their site about this
here: https://protectli.com/kb/com-port-tutorial/). These use a "Fintek"
chip: https://www.fintek.com.tw/index.php/supports/i-o-controllerdrvuti

Other units use either a real serial port or an FTDI USB/serial that should
be supported under any OS.


> And there's a
> variety of very small desktops (ThinkCentre M600, eg) that are nice and
> cheap, but one NIC and have useless things like HDMI, bluetooth and wifi
> which aren't ideal for servers. Plus idk about db9/serial...
> 
>> 1u in height is very common with those little desktops.

Finding a good "I sell off-lease desktops" seller on Bay is good for this stuff!

> 
> So I guess I have two questions:
> 
> Is it still amd64 or bust for the best stability and support?

I recall in IRC there was some talk that ARM, while in theory is "open", is
chock-full of proprietary stuff when you get down to trying to buy anything
that looks like a commodity mini-PC, right?

Charles

> 
> And then, what comes after APU2 for the many who use it?
> 
> g
> 
> _______________________________________________
> talk mailing list
> talk at lists.nycbug.org
> https://lists.nycbug.org:8443/mailman/listinfo/talk



From crossd at gmail.com  Fri May 12 15:57:48 2023
From: crossd at gmail.com (Dan Cross)
Date: Fri, 12 May 2023 15:57:48 -0400
Subject: [talk] APU2 post-mortem options
In-Reply-To: <70c52c53-e0e4-3bc8-d4fb-112254240993@ceetonetechnology.com>
References: <70c52c53-e0e4-3bc8-d4fb-112254240993@ceetonetechnology.com>
Message-ID: <CAEoi9W5t+jOOufom=HDbVXznZ9V5evFEASOo-JbkziDw=krStA@mail.gmail.com>

On Fri, May 12, 2023 at 2:10?PM George Rosamond
<george at ceetonetechnology.com> wrote:
> There's been some chatter on IRC and other places about ideal hardware
> options after APU2/PCEngines go away.
>
> amd64 remains the best-supported architecture, and aarch64 or RISC-V
> especially are not there yet. Oh, the dreams we all had years ago about
> aarch64...
>
> On that note, what similiar small fanless hardware are others looking at?

OnLogic makes some very nice kit that I'm running at home, but the
firmware story is sort of pedestrian there: UEFI+BIOS. I haven't
looked into seeing if I can get coreboot running or something like
that. But for fanless industrial computers, they have some very nice
options, though at a pricepoint rather higher than PCEngines did.

        - Dan C.


>
> There's the Qotom stuff, which I think ProtectCLI uses. And there's a
> variety of very small desktops (ThinkCentre M600, eg) that are nice and
> cheap, but one NIC and have useless things like HDMI, bluetooth and wifi
> which aren't ideal for servers. Plus idk about db9/serial...
>
> >1u in height is very common with those little desktops.
>
> So I guess I have two questions:
>
> Is it still amd64 or bust for the best stability and support?
>
> And then, what comes after APU2 for the many who use it?
>
> g
>
> _______________________________________________
> talk mailing list
> talk at lists.nycbug.org
> https://lists.nycbug.org:8443/mailman/listinfo/talk



From ike at blackskyresearch.net  Fri May 12 15:59:02 2023
From: ike at blackskyresearch.net (Isaac (.ike) Levy)
Date: Fri, 12 May 2023 15:59:02 -0400
Subject: [talk] APU2 post-mortem options
In-Reply-To: <70c52c53-e0e4-3bc8-d4fb-112254240993@ceetonetechnology.com>
References: <70c52c53-e0e4-3bc8-d4fb-112254240993@ceetonetechnology.com>
Message-ID: <AED23A28-64BE-4DA3-A6C2-6DA050B5EA7A@blackskyresearch.net>

I have one option in hand,

> On May 12, 2023, at 2:09 PM, George Rosamond <george at ceetonetechnology.com> wrote:
> 
> ideal hardware options after APU2/PCEngines go away

I've been hoarding PCEngines boards for years, please ping me off list and have bank routing numbers in hand, and don't get sassy when I state my prices.

I won't begin selling my APU2's until I depelete my supply of APU/i386 boards.

I may have a Soekris or two left as well, if you're lucky.  And if you'd like, I am able to paint the APU machines pea green, as an add on.

If you've read this far and don't yet understand I'm joking, please know I write this holding back tears that my favorite compute platform of all time is gone...  And Pascal is the best.

Best,
.ike




From george at ceetonetechnology.com  Fri May 12 16:39:22 2023
From: george at ceetonetechnology.com (George Rosamond)
Date: Fri, 12 May 2023 16:39:22 -0400
Subject: [talk] APU2 post-mortem options
In-Reply-To: <AED23A28-64BE-4DA3-A6C2-6DA050B5EA7A@blackskyresearch.net>
References: <70c52c53-e0e4-3bc8-d4fb-112254240993@ceetonetechnology.com>
 <AED23A28-64BE-4DA3-A6C2-6DA050B5EA7A@blackskyresearch.net>
Message-ID: <9650437e-486f-4a6d-6157-38cf96be20e2@ceetonetechnology.com>

On 5/12/23 15:59, Isaac (.ike) Levy wrote:
> I have one option in hand,
> 
>> On May 12, 2023, at 2:09 PM, George Rosamond <george at ceetonetechnology.com> wrote:
>>
>> ideal hardware options after APU2/PCEngines go away
> 
> I've been hoarding PCEngines boards for years, please ping me off list and have bank routing numbers in hand, and don't get sassy when I state my prices.
> 
> I won't begin selling my APU2's until I depelete my supply of APU/i386 boards.
> 
> I may have a Soekris or two left as well, if you're lucky.  And if you'd like, I am able to paint the APU machines pea green, as an add on.
> 
> If you've read this far and don't yet understand I'm joking, please know I write this holding back tears that my favorite compute platform of all time is gone...  And Pascal is the best.

Ooops. You were banned after I read the first line.  Hope this email
finds you well!

Also in seriousness.. I do remember the last analog-based trading phone
systems slowly dying many years ago. A few people in NYC had decent
supplies to keep them going. Everyone went digital obviously, but at
that point there was a reliability in copper pairs over anything
internet based. Dropping packets was not an option for traders.

I didn't stop using Soekris for at least console servers until a few
years ago... and I assume my APU2s will keep chugging away for many
years. But just like the Soekris' specs were a drag, the APU2s were
showing it for a while already.

Getting those mSATAs was rough, and the underpowered CPUs were killing
me. Plus the heat dissipation was problematic. If anyone hasn't heard
the story, one production APU2 I had hit 113C.. Don't tell me "you could
fry an egg on it".. since it would burn too quickly.

I have a hard time opting for a replacement system meant for the
desktop. I don't want your wifi/bluetooth/HDMI or even VGA. I don't want
cabinet space wasted with those useless extras. And give me db9 or give
me death.

I think I speak for many others when I say it became natural and
comfortable to just order a bunch of APU2s when in need, and they always
seemed to come in cheaper than their quoted estimate, and they arrived
faster than expected.

And I had interacted with Pascal a number of times going way back, and
really appreciated what one engineer could do to satisfy the needs for
so many of our types. And PCEngines was a sponsor at NYCBSDCon 2008*

g

* http://www.nycbsdcon.org/2008/



From jim at netgate.com  Fri May 12 17:33:31 2023
From: jim at netgate.com (Jim Thompson)
Date: Fri, 12 May 2023 15:33:31 -0600
Subject: [talk] APU2 post-mortem options
In-Reply-To: <9650437e-486f-4a6d-6157-38cf96be20e2@ceetonetechnology.com>
References: <70c52c53-e0e4-3bc8-d4fb-112254240993@ceetonetechnology.com>
 <AED23A28-64BE-4DA3-A6C2-6DA050B5EA7A@blackskyresearch.net>
 <9650437e-486f-4a6d-6157-38cf96be20e2@ceetonetechnology.com>
Message-ID: <6A5AEAED-A58A-4CF0-84C4-A8A9925FAE7E@netgate.com>


Last I checked, Protectli was just rebranded Yanbling/Minisys hardware.  You can find them on AliExpress.
Qotom is a different Chinese ODM who you can also find on AliExpress.

Patrick at STH has been doing a good job detailing what?s available.   I?d check STH if you?re looking for an inexpensive platform from China.

If you?re looking just a bit into the future, there is a new Atom coming: 
https://www.intel.com/content/www/us/en/products/sku/229206/intel-atom-processor-c1110-6m-cache-2-10-ghz/specifications.html

While I can?t say much about it, Silicom has leaked that this is based on Gracemont cores (this is the ?little? core in Alder Lake).
https://www.silicom-usa.com/pr/4g-5g-products/4g-5g-appliances/valencia-network-appliance-series/

This is not your father?s Atom.  It?s not what you?re used to for performance from an Atom.  Four cores of Gracemont deliver 80% more performance than two Skylake cores.

https://chipsandcheese.com/2021/12/21/gracemont-revenge-of-the-atom-cores/
https://www.anandtech.com/show/16881/a-deep-dive-into-intels-alder-lake-microarchitectures/4
[?]

Four cores of Gracemont and four i226 2.5gbps Ethernet devices will run fanless at over 40C ambient.

In addition to a much more performant front-end and scheduler, the C1110 also supports AVX/AVX2 and even VAES/VPCMULQDQ if you encode the instructions using VEX.  This makes AES-GCM, AES-CBC/SHA-256 and ChaCha20/Poly1305 go real fast.

As an example: Using VPP Intel will show it doing IPsec (AES-128-GCM) at ? well 14.99gbps on a single core.  But you don?t care about VPP, you want to hear about FreeBSD performance, right?

I?m not going to quote benchmarking on the C1110 (yet), but as an illustration of why you might want to consider having AVX/AVX2 in your next platform:

On a Ryzen 5 5600, using 4 FreeBSD VMs, each with 4 guest-CPUs, 4GB RAM, 2 SR-IOV passthrough ConnectX virtual function NICs
connected via a Mellanox ConnectX-5EN 25Gbps PCIe-3 which features an embedded switch capable of delivering 50Gbps traffic between the virtual functions and a single TCP stream via iperf:

?

OCF below is what you can get with FreeBSD
IIMB is the BSD-licensed Intel IPsec-MB library, ported to the OCF framework.

IPsec-ESP (AES-128-GCM)
- OCF: sync: 4.88Gbps async: 4.46Gbps
- IIMB: sync: 5.48Gbps async: 7.68Gbps

IPsec-ESP (AES-256-CBC,HMAC-SHA256)
- OCF async: 4.2Gbps
- IIMB async: 5.3Gbps

ChaCha20/Poly1305 also gets a big lift from AVX/AVX2.  
WireGuard
- OCF: 4.4Gbps
- IIMB: 6.0Gbps

Notes:
- WireGuard consumes all CPUs with low idle time (<15%), whereas IPsec and OpenVPN uses 3 CPUs for network and encryption workloads
(20% idle) and are otherwise 60% idle during this test.

- Linux native WireGuard attained 7.5Gbps on the same virtual setup

As a bit more illustration of what is possible with some software work:

OpenVPN DCO (AES-256-GCM):
- OCF: sync: 4.1Gbps async: 6.0Gbps
- IIMB: sync: 4.9Gbps async: 10Gbps (*)

OpenVPN DCO (ChaCha20-Poly1305):
- IIMB: sync: 2.9Gbps async: 5.1Gbps

(*) peak throughput for OpenVPN DCO was observed to be 12Gbps 

This is a big reason we did the work to bring IPsec-MB to pfSense Plus. https://redmine.pfsense.org/issues/14291
There was a talk at AsiaBSDCon about this, by Leon Dang, presented by Kristof Provost
https://2023.asiabsdcon.org/program.html.en

Some of the results above are from that paper.

I?m not saying you want an Alder Lake for your server, you don?t, in part because the FreeBSD scheduler doesn?t do well on a big/little architecture, and there isn?t anyone in the project to work on that.  But if you cut the big (Arizona Beach) *or* little cores (Sapphire Rapids) off, then FreeBSD can run very well on the result.

Responding to Charles Sprickman:
> I recall in IRC there was some talk that ARM, while in theory is "open", is
> chock-full of proprietary stuff when you get down to trying to buy anything
> that looks like a commodity mini-PC, right?


They?re not PCs, if that?s what you mean, but it?s straight-forward to get FreeBSD running on them.  Then the work of writing drivers for that vendor?s devices begins.  They tend to ship drivers for linux, but not for anything else.

Jim


> On May 12, 2023, at 2:39 PM, George Rosamond <george at ceetonetechnology.com> wrote:
> 
> On 5/12/23 15:59, Isaac (.ike) Levy wrote:
>> I have one option in hand,
>> 
>>> On May 12, 2023, at 2:09 PM, George Rosamond <george at ceetonetechnology.com> wrote:
>>> 
>>> ideal hardware options after APU2/PCEngines go away
>> 
>> I've been hoarding PCEngines boards for years, please ping me off list and have bank routing numbers in hand, and don't get sassy when I state my prices.
>> 
>> I won't begin selling my APU2's until I depelete my supply of APU/i386 boards.
>> 
>> I may have a Soekris or two left as well, if you're lucky.  And if you'd like, I am able to paint the APU machines pea green, as an add on.
>> 
>> If you've read this far and don't yet understand I'm joking, please know I write this holding back tears that my favorite compute platform of all time is gone...  And Pascal is the best.
> 
> Ooops. You were banned after I read the first line.  Hope this email
> finds you well!
> 
> Also in seriousness.. I do remember the last analog-based trading phone
> systems slowly dying many years ago. A few people in NYC had decent
> supplies to keep them going. Everyone went digital obviously, but at
> that point there was a reliability in copper pairs over anything
> internet based. Dropping packets was not an option for traders.
> 
> I didn't stop using Soekris for at least console servers until a few
> years ago... and I assume my APU2s will keep chugging away for many
> years. But just like the Soekris' specs were a drag, the APU2s were
> showing it for a while already.
> 
> Getting those mSATAs was rough, and the underpowered CPUs were killing
> me. Plus the heat dissipation was problematic. If anyone hasn't heard
> the story, one production APU2 I had hit 113C.. Don't tell me "you could
> fry an egg on it".. since it would burn too quickly.
> 
> I have a hard time opting for a replacement system meant for the
> desktop. I don't want your wifi/bluetooth/HDMI or even VGA. I don't want
> cabinet space wasted with those useless extras. And give me db9 or give
> me death.
> 
> I think I speak for many others when I say it became natural and
> comfortable to just order a bunch of APU2s when in need, and they always
> seemed to come in cheaper than their quoted estimate, and they arrived
> faster than expected.
> 
> And I had interacted with Pascal a number of times going way back, and
> really appreciated what one engineer could do to satisfy the needs for
> so many of our types. And PCEngines was a sponsor at NYCBSDCon 2008*
> 
> g
> 
> * https://www.google.com/url?q=http://www.nycbsdcon.org/2008/&source=gmail-imap&ust=1684528879000000&usg=AOvVaw21__LKxrafaNPaaMbpwgus
> 
> _______________________________________________
> talk mailing list
> talk at lists.nycbug.org <mailto:talk at lists.nycbug.org>
> https://www.google.com/url?q=https://lists.nycbug.org:8443/mailman/listinfo/talk&source=gmail-imap&ust=1684528879000000&usg=AOvVaw3GLllRw5MRdhQkahcaQOr_

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.nycbug.org:8443/pipermail/talk/attachments/20230512/06507b31/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: PastedGraphic-1.png
Type: image/png
Size: 43856 bytes
Desc: not available
URL: <https://lists.nycbug.org:8443/pipermail/talk/attachments/20230512/06507b31/attachment.png>

From mcevoy.pat at gmail.com  Tue May 23 19:18:20 2023
From: mcevoy.pat at gmail.com (Pat McEvoy)
Date: Tue, 23 May 2023 19:18:20 -0400
Subject: [talk] Next NYC*BUG: 2 Lightning Talks, 2023-06-14 @ 18:45
Message-ID: <B5E32ADB-2B17-4BE7-B70A-35F976E12388@gmail.com>

2 Lightning Talks, _
2023-06-14 @ 18:45

We are excited about this one as one talk is by Josh, a new NYC*Bug member and the other by Ra?l, a member who has been with us much longer and has generously helped in the running our user group.


https://www.nycbug.org/index?action=view&id=10689






From jkeenan at pobox.com  Tue May 23 21:11:40 2023
From: jkeenan at pobox.com (James E Keenan)
Date: Tue, 23 May 2023 21:11:40 -0400
Subject: [talk] Next NYC*BUG: 2 Lightning Talks, 2023-06-14 @ 18:45
In-Reply-To: <B5E32ADB-2B17-4BE7-B70A-35F976E12388@gmail.com>
References: <B5E32ADB-2B17-4BE7-B70A-35F976E12388@gmail.com>
Message-ID: <3e2c6f98-4ec4-e39c-69d7-80d17e38ff6b@pobox.com>

On 5/23/23 19:18, Pat McEvoy wrote:
> 2 Lightning Talks, _
> 2023-06-14 @ 18:45
> 
> We are excited about this one as one talk is by Josh, a new NYC*Bug member and the other by Ra?l, a member who has been with us much longer and has generously helped in the running our user group.
> 
> 
> https://www.nycbug.org/index?action=view&id=10689
> 

Just to clarify ... this is a change from our usual first Wednesday of 
the month schedule -- correct?

jimk



From mcevoy.pat at gmail.com  Tue May 23 21:31:39 2023
From: mcevoy.pat at gmail.com (Pat McEvoy)
Date: Tue, 23 May 2023 21:31:39 -0400
Subject: [talk] Next NYC*BUG: 2 Lightning Talks, 2023-06-14 @ 18:45
In-Reply-To: <3e2c6f98-4ec4-e39c-69d7-80d17e38ff6b@pobox.com>
References: <3e2c6f98-4ec4-e39c-69d7-80d17e38ff6b@pobox.com>
Message-ID: <A7CF8780-BAB6-4F2C-98E7-CAE515288D3A@gmail.com>

Yes. A bunch of teachers snagged our day. And while we are on the subject, we also have July 12 as July 5th would be a failure of previous commitments for everyone. 

Patrick McEvoy


> On May 23, 2023, at 21:28, James E Keenan <jkeenan at pobox.com> wrote:
> 
> ?On 5/23/23 19:18, Pat McEvoy wrote:
>> 2 Lightning Talks, _
>> 2023-06-14 @ 18:45
>> We are excited about this one as one talk is by Josh, a new NYC*Bug member and the other by Ra?l, a member who has been with us much longer and has generously helped in the running our user group.
>> https://www.nycbug.org/index?action=view&id=10689
> 
> Just to clarify ... this is a change from our usual first Wednesday of the month schedule -- correct?
> 
> jimk
> 
> _______________________________________________
> talk mailing list
> talk at lists.nycbug.org
> https://lists.nycbug.org:8443/mailman/listinfo/talk



From george at ceetonetechnology.com  Tue May 23 22:12:22 2023
From: george at ceetonetechnology.com (George Rosamond)
Date: Tue, 23 May 2023 22:12:22 -0400
Subject: [talk] BSDCan 2024
Message-ID: <0058cc4c-0a0c-8fe7-2370-2dbb21c11569@ceetonetechnology.com>

Dan is stepping back, and a federally overseer has been mandated.

https://mwl.io/archives/22799

I am disturbed by the mask policy. COVID-mitigating masks are fine with
me. But apparently this BSDCan overseer character is a heavy in the
horror/scifi writers guild, and I suspect they don't mean COVID masks.

In all seriousness, it was a nice run Dan, and we were glad to be a
small part from the beginning.  Running a con is a lot of work. It's not
just having some checklist that you move through quickly. It's like
having an intense part-time job for months and the boss is particularly
demanding. And the stakes are high.

Like MWL wrote,

One thing that the BSD community has historically excelled at is passing
leadership from one generation to the next.

I hope others on talk@ can take the time to attend and contribute to the
new era of BSDCans.

g



From assaf at eml.cc  Wed May 24 14:41:08 2023
From: assaf at eml.cc (assaf rutenberg)
Date: Wed, 24 May 2023 14:41:08 -0400
Subject: [talk] Request for assistance
Message-ID: <d9d1e288-1cba-4951-808e-b4f6c1fc2dc1@app.fastmail.com>

Hey all,
In a moment of recent recklessness and euphoria I bought an ixsystems truenas  mini +. It has an 8 core cpu, 32gb of ram, and dual 10gb enet ports. To that I added (5) 12TB ironwolf drives I pulled from my openmediavault external enclosure and (2) 1tb ssd's to use as read and write cache drives. The supplied os is truenas core.
To put it mildly, I am in over my head as far as basic setup and would like to pay one of you fine and talented folks to help me set it up. I just need to add a user and configure the shares but I am oddly confounded by what should be a simple process.
All of this is compounded by the startling fact that despite the relative expense of the unit, ixsystems does not offer and support to home users.
So, if anyone has 30 minutes or so to assist me in the basic setup of this unit, I would be very grateful. 
Thank you, and I look forward to seeing you at the June meet-up.

Sheepishly yours,

Assaf Rutenberg 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.nycbug.org:8443/pipermail/talk/attachments/20230524/50b3f008/attachment.htm>

From pete at nomadlogic.org  Wed May 24 19:06:00 2023
From: pete at nomadlogic.org (Pete Wright)
Date: Wed, 24 May 2023 16:06:00 -0700
Subject: [talk] BSDCan 2024
In-Reply-To: <0058cc4c-0a0c-8fe7-2370-2dbb21c11569@ceetonetechnology.com>
References: <0058cc4c-0a0c-8fe7-2370-2dbb21c11569@ceetonetechnology.com>
Message-ID: <ydkng54upx4kiatxbbur54peziqxetxdez6hw5kv2dwgky7vmn@kjev2mzl3ufl>

On Tue, May 23, 2023 at 10:12:22PM -0400, George Rosamond wrote:
> 
> Dan is stepping back, and a federally overseer has been mandated.
> 
> https://mwl.io/archives/22799
> 
> I am disturbed by the mask policy. COVID-mitigating masks are fine with
> me. But apparently this BSDCan overseer character is a heavy in the
> horror/scifi writers guild, and I suspect they don't mean COVID masks.
> 
> In all seriousness, it was a nice run Dan, and we were glad to be a
> small part from the beginning.  Running a con is a lot of work. It's not
> just having some checklist that you move through quickly. It's like
> having an intense part-time job for months and the boss is particularly
> demanding. And the stakes are high.
> 
> Like MWL wrote,
> 
> One thing that the BSD community has historically excelled at is passing
> leadership from one generation to the next.
> 
> I hope others on talk@ can take the time to attend and contribute to the
> new era of BSDCans.
> 

hey this is great news - so thankful for all the work Dan has done, i can't
imagine the months of prep that need to get done to make it happen every
year.  sounds like its getting put in good hands too - hopefully this'll
be happening for years to come!

-pete


-- 
Pete Wright
pete at nomadlogic.org