On Monday, April 11, 2022 at 7:00:39 AM UTC-4, Whisky-dave wrote:
> On Monday, 11 April 2022 at 02:20:47 UTC+1, -hh wrote:
> > On Friday, April 8, 2022 at 8:36:47 AM UTC-4, Whisky-dave wrote:
> > > On Thursday, 7 April 2022 at 20:53:36 UTC+1, -hh wrote:
> > > > On Thursday, April 7, 2022 at 7:24:21 AM UTC-4, Whisky-dave wrote:
> > > > > On Tuesday, 5 April 2022 at 14:33:25 UTC+1, -hh wrote:
> > > > > > On Monday, April 4, 2022 at 11:11:38 AM UTC-4, Whisky-dave wrote:
> > > > > > > On Friday, 1 April 2022 at 20:11:48 UTC+1, -hh wrote:
> > > > > > > > On Friday, April 1, 2022 at 11:06:49 AM UTC-4, Whisky-dave wrote:
> > > > > > > > > [etc]
> > > > > > > > >...
> > > > > > > Well you can get 8TB SSD in a laptop today but I wouldn;t pay that much extra for one.
> > > > > > >
> > > > > > > I'll wait until they come down in price.
> > > > > >
> > > > > > Broadly speaking, $100/TB was my break point for more broad SSD adoption. Seems
> > > > > > that there's CoVid rises on prices, as it seems that they've rebounded to ~$140/TB
> > > > >
> > > > > But it doesn't aviod the problem and I'm betting you didn;t buy your current computer with
> > > > > a 4TB internal drive even though you say you need 4TB now.
> > > > Of course not, as my cheesegrater dates from a decade ago (2012) and the PCIe
> > > > based SSD was just 256GB at the time which cost IIRC $600-$700 for its OS boot
> > > > drive, and the second 'data' drive was 2*2TB HDDs in a RAID-0 for another ~$300.
> > >
> > > Seems risky to me to use RAID-0 unless you keep reqular backups.
> > Oh, of course there's regular backups.
>
> So you do use external storage ...

That too, but the primary backups are onboard (internal). Granted, this does invoke a
trade-off on convenience/reliability vs single point of failure...

> ...which is why I can''t understand why you must have everything on an internal drive.

It depends on the use case. For backups, its not a "must" as much as it is a preference,
as such additions add a tangle of wires for signal & power that have to be managed too.
Case in point, I'm overdue to 'detangle' the cheesegrater's power systems again, as an
ethernet switch snuck on there somehow ... a quick glance this AM behind the monitor
found no less than 13 plugs on the power trays.

> I know you say it's related to speed of access which might increase more with a new
> processor, graphics card, and everything else that comes with having everything on
> the same chip/board. Rather than the speed of the SSD which is likely to be faster
> in a new computer than the ten year old one you have.

Of course we want all of the specs to take a step up when we buy a new system;
question is how much is worth how much. My lessons-learned from my current
system is that the I/O bandwidth has been performance/UI pacing element, which
was why I'd invested in hardware to open up that particular bottleneck. As such,
my motivation now is that if I'm going to be laying out money to replace it, I don't
want to a step backwards, or even to just tread water; I'd like to have "more", so
long as it is reasonably attainable.
> and there are other things that make computers faster than just the boot 'drive'

But of course. Each subsystem contributes in its own way. For example, my
legacy desktop has had 24GB of RAM since it was brought up in 2012, and I've
monitored page swaps to see if that's enough or not.

FWIW, a friend has a Windows PC build for an interesting use case and he's found
a throughput bottleneck here; his trade-off is to either spend a ~manyear to rework
his code to get it to run faster, or he can just "throw money at the problem" by
bumping the hardware from 64GB to 128GB for a similar amount of gain. From a
business sense, the hardware route is the most cost-effective "make or buy" answer.

> > > So now I only 'plan' for about ~4 years ...
> >
> > I'd be pretty ticked off it a new system didn't go at least 4 years, unless my
> > use cases dramatically changed.
>
> I don't really know what my use case will be in even 2 years time.
> In 2014 I didn't know I'd spend time filming my first 2 goldfinches on a bird
> feeder and a barking squirrel in 4k
> https://youtu.be/3xjMY3h0trY
>
> and a barking squirrel
> https://youtu.be/_b1fJbdF97Q
>
> These files were about 35GB in total I don't feel the need to keep the originals
> and back them up multiple times. Kept the files I uploaded to youtube about
> 2GB on my internal drive for now until I need the space. Which might be at
> the end of may my next planned gig so will need about 40GB for that
> presently have ~70GB free.

A good illustration of how capability needs grow in generally unexpected ways,
even though in retrospect, we could have looked at the tech of 4K from back
when you bought the machine to try to guess if it was a realistic growth direction.

FWIW, this is why I'm so concerned about my own 'future proofing': I think that
its a given that I'll want to do 4K ... but the open question is if that will also include
wanting to process 8K too, which would be more demanding (and $$) on today's
hardware. This is the inherent challenge on assessing future-proofing even for
shorter periods of time (eg, <5yrs): one of my current work projects is trying to
assess if we can develop an effective ~25 year futureproofing strategy, because
of the levels of money involved for implementing any major configuration change.

> > Even sticking with what just Apple currently offers on the Studio, that's 8GB,
> > which is 2x what I have today, and thus adequate for IMO 6+ years of image
> > portfolio growth.
>
> I'd have to wonder whether I might be better off spending the money on a faster
> processor/more cores or more RAM and use the remaining money and get a
> thunderbold external drive.

Depends on one's workflows and where the performance bottleneck(s) are which
affect your UI along the way, and which portion of the UI is the one that you're
prioritizing improvements for.

> > > I'm not sure how much better and hence larger single still photos will get.
> > > I'm finding jpg file sizes of about 3-5MB plenty. But some like storing everything
> > > as RAW ~40MB. I'm not sure I'll ever need such files to be 1GB.
> >
> > Oh, my point here is merely that use cases changes are uncertainty risks: the assumptions
> > which are being used to gage variables like lifespan get tossed out the window, so one
> > can (should) do a sensitivity analysis on them...again, it can be part of risk-reduction.
>
> For me lifespan is how long I expect the computer to last, Which is more linked the warrenty
> rather than how long it will be usable for my uses.
> PCs in my lab seem to need replacing after 3-4 years. We've already had to replace the
>1TB HDDs with SSDs only about £100 each. but 3 years ago this would have been to expensive
> to buy 100 PCs with internal 1 TB SSDs otherwise we'd have bought them I assume.

Much depends on the use case(s) and what the effects are of any considered change:
that which is a good business decision to make for a lab of 100 machines for students
just isn't going to be the same as for a small cluster of 3-5 high end machines used by
specialists who get paid six digit salaries: spending an extra $1000 for the latter to be
more productive can have very short Return on Investment periods (months or weeks).

> > > > It really comes down to paying for the faster I/O where it has the most benefit.
> > >
> > > And presently that seems to be having everything on the same chip CPU, RAM, SSD,
> > > which of course can't be user upgraded.
> >
> > Where the trade-off is upfront costs & useful lifespan.
>
> Which is the difficult part, same with most tech products, ...

Yup. Its also more challenging when the use case is "hobby" instead of business, as
it is harder to objectively quantify a notional ROI ... plus there's no tax write-off (!).

> I can't see the point in upgrading my 2016 iPhone7 unless there come out
> with some serious slow-mo, like 1000fps or more.
> I spent quite a bit of time and film trying to film a ballon bursting in the mid 70s.

Sounds like a classic "Papa Flash" experiment. My work has been more in lines
with "making applesauce at MIT"; the camera I bought for the lab in 2002 was
something like just 640x640 pixels in B&W; we could have opted for color, but doing
so decreased its light sensitivity by ~2 stops, and with the era's pre-LED illumination,
we were throwing some heat on frame rates above 10K fps. With the RAM upgrade to
allow 4sec of total capture time, it was just under our $100K limit. A lot has changed
in the past 20 years.

On Monday, 11 April 2022 at 14:44:31 UTC+1, -hh wrote:
> On Monday, April 11, 2022 at 7:00:39 AM UTC-4, Whisky-dave wrote:
> > On Monday, 11 April 2022 at 02:20:47 UTC+1, -hh wrote:
> > > On Friday, April 8, 2022 at 8:36:47 AM UTC-4, Whisky-dave wrote:
> > > > On Thursday, 7 April 2022 at 20:53:36 UTC+1, -hh wrote:
> > > > > On Thursday, April 7, 2022 at 7:24:21 AM UTC-4, Whisky-dave wrote:
> > > > > > On Tuesday, 5 April 2022 at 14:33:25 UTC+1, -hh wrote:
> > > > > > > On Monday, April 4, 2022 at 11:11:38 AM UTC-4, Whisky-dave wrote:
> > > > > > > > On Friday, 1 April 2022 at 20:11:48 UTC+1, -hh wrote:
> > > > > > > > > On Friday, April 1, 2022 at 11:06:49 AM UTC-4, Whisky-dave wrote:
> > > > > > > > > > [etc]
> > > > > > > > > >...
> > > > > > > > Well you can get 8TB SSD in a laptop today but I wouldn;t pay that much extra for one.
> > > > > > > >
> > > > > > > > I'll wait until they come down in price.
> > > > > > >
> > > > > > > Broadly speaking, $100/TB was my break point for more broad SSD adoption. Seems
> > > > > > > that there's CoVid rises on prices, as it seems that they've rebounded to ~$140/TB
> > > > > >
> > > > > > But it doesn't aviod the problem and I'm betting you didn;t buy your current computer with
> > > > > > a 4TB internal drive even though you say you need 4TB now.
> > > > > Of course not, as my cheesegrater dates from a decade ago (2012) and the PCIe
> > > > > based SSD was just 256GB at the time which cost IIRC $600-$700 for its OS boot
> > > > > drive, and the second 'data' drive was 2*2TB HDDs in a RAID-0 for another ~$300.
> > > >
> > > > Seems risky to me to use RAID-0 unless you keep reqular backups.
> > > Oh, of course there's regular backups.
> >
> > So you do use external storage ...
>
> That too, but the primary backups are onboard (internal). Granted, this does invoke a
> trade-off on convenience/reliability vs single point of failure...

Personally I wouldn't use internal drive for backup other then as a temproy measure while I play with an orginal file.

>
> > ...which is why I can''t understand why you must have everything on an internal drive.
>
> It depends on the use case. For backups, its not a "must" as much as it is a preference,
> as such additions add a tangle of wires for signal & power that have to be managed too.
> Case in point, I'm overdue to 'detangle' the cheesegrater's power systems again, as an
> ethernet switch snuck on there somehow ... a quick glance this AM behind the monitor
> found no less than 13 plugs on the power trays.

There's this thing called wireless , it might not be the fastest method but it can be done 24/7.

> > I know you say it's related to speed of access which might increase more with a new
> > processor, graphics card, and everything else that comes with having everything on
> > the same chip/board. Rather than the speed of the SSD which is likely to be faster
> > in a new computer than the ten year old one you have.
> Of course we want all of the specs to take a step up when we buy a new system;
> question is how much is worth how much. My lessons-learned from my current
> system is that the I/O bandwidth has been performance/UI pacing element, which
> was why I'd invested in hardware to open up that particular bottleneck. As such,
> my motivation now is that if I'm going to be laying out money to replace it, I don't
> want to a step backwards, or even to just tread water; I'd like to have "more", so
> long as it is reasonably attainable.

I'm betting it's a difficult decision as you don't know how effecient your DAM is
with any given hardware. But I'm pretty sure that I can download and upload to the cloud
much faster than I can read or write to an intenal floppy drive or an internal HD
of an old computer.

> > and there are other things that make computers faster than just the boot 'drive'
> But of course. Each subsystem contributes in its own way. For example, my
> legacy desktop has had 24GB of RAM since it was brought up in 2012, and I've
> monitored page swaps to see if that's enough or not.
>
> FWIW, a friend has a Windows PC build for an interesting use case and he's found
> a throughput bottleneck here; his trade-off is to either spend a ~manyear to rework
> his code to get it to run faster, or he can just "throw money at the problem" by
> bumping the hardware from 64GB to 128GB for a similar amount of gain. From a
> business sense, the hardware route is the most cost-effective "make or buy" answer.

I think even professional software venders take that approach even MS.
It's far easier for them to suggest you need a new faster computer than for them to re-write their software.
On my iMac at home I have word and excel bith taking up about 2GB each on the HD.
The only reason I have them is A/ because I get them for free
B/ to be compatable with work.
I'm pretty sure I could get them both on one 800k floppy.
But most of the time I use google docs anyway, zero disc space.

> > > > So now I only 'plan' for about ~4 years ...
> > >
> > > I'd be pretty ticked off it a new system didn't go at least 4 years, unless my
> > > use cases dramatically changed.
> >
> > I don't really know what my use case will be in even 2 years time.
> > In 2014 I didn't know I'd spend time filming my first 2 goldfinches on a bird
> > feeder and a barking squirrel in 4k
> > https://youtu.be/3xjMY3h0trY
> >
> > and a barking squirrel
> > https://youtu.be/_b1fJbdF97Q
> >
> > These files were about 35GB in total I don't feel the need to keep the originals
> > and back them up multiple times. Kept the files I uploaded to youtube about
> > 2GB on my internal drive for now until I need the space. Which might be at
> > the end of may my next planned gig so will need about 40GB for that
> > presently have ~70GB free.
> A good illustration of how capability needs grow in generally unexpected ways,
> even though in retrospect, we could have looked at the tech of 4K from back
> when you bought the machine to try to guess if it was a realistic growth direction.

Don't think I could have any more than I can predict when I need a 4K 8k or 16k TV.
When I bought my current iMac you couldn't by a 4k camera. Unless you were a TV/cinema producer.

>
> FWIW, this is why I'm so concerned about my own 'future proofing':

I know I can't future proof. Well not by more than a few years, which isn't furture proofing anyway.

>I think that
> its a given that I'll want to do 4K ... but the open question is if that will also include
> wanting to process 8K too, which would be more demanding (and $$) on today's
> hardware. This is the inherent challenge on assessing future-proofing even for
> shorter periods of time (eg, <5yrs): one of my current work projects is trying to
> assess if we can develop an effective ~25 year futureproofing strategy, because
> of the levels of money involved for implementing any major configuration change.

I think that is impossible. Try using a 25 year old computer try connecting it to a 8TB disc
I very much doubt you'd have the processing power in a 20 year old CPU.
Which were only 32 bit so 4GB of RAM was the maxium until 2001 when 64 bit processors started.

> > > Even sticking with what just Apple currently offers on the Studio, that's 8GB,
> > > which is 2x what I have today, and thus adequate for IMO 6+ years of image
> > > portfolio growth.
> >
> > I'd have to wonder whether I might be better off spending the money on a faster
> > processor/more cores or more RAM and use the remaining money and get a
> > thunderbold external drive.
> Depends on one's workflows and where the performance bottleneck(s) are which
> affect your UI along the way, and which portion of the UI is the one that you're
> prioritizing improvements for.

Providing nothing else changes the software could still be the bottleneck.

> > > > I'm not sure how much better and hence larger single still photos will get.
> > > > I'm finding jpg file sizes of about 3-5MB plenty. But some like storing everything
> > > > as RAW ~40MB. I'm not sure I'll ever need such files to be 1GB.
> > >
> > > Oh, my point here is merely that use cases changes are uncertainty risks: the assumptions
> > > which are being used to gage variables like lifespan get tossed out the window, so one
> > > can (should) do a sensitivity analysis on them...again, it can be part of risk-reduction.
> >
> > For me lifespan is how long I expect the computer to last, Which is more linked the warrenty
> > rather than how long it will be usable for my uses.
> > PCs in my lab seem to need replacing after 3-4 years. We've already had to replace the
> >1TB HDDs with SSDs only about £100 each. but 3 years ago this would have been to expensive
> > to buy 100 PCs with internal 1 TB SSDs otherwise we'd have bought them I assume.
> Much depends on the use case(s) and what the effects are of any considered change:
> that which is a good business decision to make for a lab of 100 machines for students
> just isn't going to be the same as for a small cluster of 3-5 high end machines used by
> specialists who get paid six digit salaries: spending an extra $1000 for the latter to be
> more productive can have very short Return on Investment periods (months or weeks).

On Tuesday, April 12, 2022 at 9:38:28 AM UTC-4, Whisky-dave wrote:
> On Monday, 11 April 2022 at 14:44:31 UTC+1, -hh wrote:
> > On Monday, April 11, 2022 at 7:00:39 AM UTC-4, Whisky-dave wrote:
> > > On Monday, 11 April 2022 at 02:20:47 UTC+1, -hh wrote:
> > > > [etc]
> > ...
> > That too, but the primary backups are onboard (internal). Granted, this does
> > invoke a trade-off on convenience/reliability vs single point of failure...
>
> Personally I wouldn't use internal drive for backup other then as a temproy
> measure while I play with an orginal file.

I tend to organize into tiers; as each tier becomes less proximate, it should be
less vulnerable to single point of failure risks. But the trade tends to be that
they are of lower I/O bandwidth (so any recovery will be slower), with less
frequent backup rates (so newest updates may be lost) ...just more trades.

....
> > Case in point, I'm overdue to 'detangle' the cheesegrater's power systems again, as an
> > ethernet switch snuck on there somehow ... a quick glance this AM behind the monitor
> > found no less than 13 plugs on the power trays.
>
> There's this thing called wireless , it might not be the fastest method but it can be done 24/7.

Sure, but wireless doesn't eliminate power wires.

....
> >I think that
> > its a given that I'll want to do 4K ... but the open question is if that will also include
> > wanting to process 8K too, which would be more demanding (and $$) on today's
> > hardware. This is the inherent challenge on assessing future-proofing even for
> > shorter periods of time (eg, <5yrs): one of my current work projects is trying to
> > assess if we can develop an effective ~25 year futureproofing strategy, because
> > of the levels of money involved for implementing any major configuration change.
>
> I think that is impossible. Try using a 25 year old computer try connecting it to a 8TB disc

Well, the SATA interface dates from 2000, so its already 22 years old. It still
lives because it was able to be incrementally improved; my effort is similar
to planning out how SATA advanced from SATA-1, to SATA-2, SATA-3..

> Yep research machines are like that. Which is why I have an old research iMac.
> Still working fine a bit slow to start up with it''s 1TB HDD. Still have 750GB free.
> Have lots of software I don't need or use. But office 2011 can't be updated,
> and the 3D printer software I use can't be updated iether so even if someone offered me
> an 32TB SSD it'd be pretty useless because of the system software can't be updated.

Think I got one of those laying around here someplace, along with a PATA interface
for it...was kicking around the idea of a home project to get an old Cube running <g>.

> > > I can't see the point in upgrading my 2016 iPhone7 unless there come out
> > > with some serious slow-mo, like 1000fps or more. I spent quite a bit of
> > > time and film trying to film a balloon bursting in the mid 70s.
> >
> > Sounds like a classic "Papa Flash" experiment.
>
> You can get arrested for that sort of thing, well in public.

Only when skin diving, as "Doc" Edgerton did strobe development with Cousteau too...

-hh

On Tuesday, 12 April 2022 at 19:57:33 UTC+1, -hh wrote:
> On Tuesday, April 12, 2022 at 9:38:28 AM UTC-4, Whisky-dave wrote:
> > On Monday, 11 April 2022 at 14:44:31 UTC+1, -hh wrote:
> > > On Monday, April 11, 2022 at 7:00:39 AM UTC-4, Whisky-dave wrote:
> > > > On Monday, 11 April 2022 at 02:20:47 UTC+1, -hh wrote:
> > > > > [etc]
> > > ...
> > > That too, but the primary backups are onboard (internal). Granted, this does
> > > invoke a trade-off on convenience/reliability vs single point of failure...
> >
> > Personally I wouldn't use internal drive for backup other then as a temproy
> > measure while I play with an orginal file.
> I tend to organize into tiers; as each tier becomes less proximate, it should be
> less vulnerable to single point of failure risks. But the trade tends to be that
> they are of lower I/O bandwidth (so any recovery will be slower), with less
> frequent backup rates (so newest updates may be lost) ...just more trades.

Seems the opposite way to how I do it.

> > > Case in point, I'm overdue to 'detangle' the cheesegrater's power systems again, as an
> > > ethernet switch snuck on there somehow ... a quick glance this AM behind the monitor
> > > found no less than 13 plugs on the power trays.
> >
> > There's this thing called wireless , it might not be the fastest method but it can be done 24/7.
> Sure, but wireless doesn't eliminate power wires.

It does if you use rechargable batteries, suprised that isn't more common.
I recently discovered that a SSD drawing 10ma when idle, will draw around 500ma when I
was using iMovie to processes/export my 4k videos I did last week.
Note sure what an old disc would draw but they seem to use 12V too.

> ...
> > >I think that
> > > its a given that I'll want to do 4K ... but the open question is if that will also include
> > > wanting to process 8K too, which would be more demanding (and $$) on today's
> > > hardware. This is the inherent challenge on assessing future-proofing even for
> > > shorter periods of time (eg, <5yrs): one of my current work projects is trying to
> > > assess if we can develop an effective ~25 year futureproofing strategy, because
> > > of the levels of money involved for implementing any major configuration change.
> >
> > I think that is impossible. Try using a 25 year old computer try connecting it to a 8TB disc
> Well, the SATA interface dates from 2000, so its already 22 years old. It still
> lives because it was able to be incrementally improved; my effort is similar
> to planning out how SATA advanced from SATA-1, to SATA-2, SATA-3..

as long as the rest of the kit keeps up. Putting a SATA 3 drive in a 2000 computer
with NOT give you the results you want.
And thunderbold an external drive is faster than any SATA drive is currently.
So you may well get better speed using an external drive rather than an internal.

> > Yep research machines are like that. Which is why I have an old research iMac.
> > Still working fine a bit slow to start up with it''s 1TB HDD. Still have 750GB free.
> > Have lots of software I don't need or use. But office 2011 can't be updated,
> > and the 3D printer software I use can't be updated iether so even if someone offered me
> > an 32TB SSD it'd be pretty useless because of the system software can't be updated.
> Think I got one of those laying around here someplace, along with a PATA interface
> for it...was kicking around the idea of a home project to get an old Cube running <g>.

Those cudes did look good.
I was using my late 1990s G4 Dual 500MHz OS9.2 last night. Same sort of era .

> > > > I can't see the point in upgrading my 2016 iPhone7 unless there come out
> > > > with some serious slow-mo, like 1000fps or more. I spent quite a bit of
> > > > time and film trying to film a balloon bursting in the mid 70s.
> > >
> > > Sounds like a classic "Papa Flash" experiment.
> >
> > You can get arrested for that sort of thing, well in public.
> Only when skin diving, as "Doc" Edgerton did strobe development with Cousteau too...
>
> -hh