On Tue, 22 Aug 2023 18:24:58 GMT
scott@slp53.sl.home (Scott Lurndal) wrote:
>Muttley@dastardlyhq.com writes:
>>>That is not an accurate statement for any processor built in
>>>the last decade or more. Most of them use a mesh, crossbar or ring
>>
>>Given millions of 8 and 16 bit processors are still designed and produced for
>
>>embedded systems that statement is nonsense and ARM based systems can use
>>whatever parts they want in the CPU.
>
>ARM based multicore systems generally use CMN-400/500/600/700 IP (from ARM)
>as the processor interconnect.
>
>https://developer.arm.com/Processors/CoreLink%20CMN-600
>
>And all those 8 and 16-bit processors are single core, so your
>statement above, which references "no matter how many cores"
>doesn't apply to them.

You said any processor built in the last decade or more. You didn't specific
only multicore. Do you wish to retract that statement?

On Tue, 22 Aug 2023 20:50:55 +0200
Bonita Montero <Bonita.Montero@gmail.com> wrote:
>Am 22.08.2023 um 20:18 schrieb Muttley@dastardlyhq.com:
>> On Tue, 22 Aug 2023 20:13:17 +0200
>> Bonita Montero <Bonita.Montero@gmail.com> wrote:
>>> Am 22.08.2023 um 19:56 schrieb Muttley@dastardlyhq.com:
>>>
>>>> If it has to be paged from disk in the first place the size of the
>>>> cache is a total irrelevance as far as time taken is concerned.
>>>
>>> No one uses software that actually pages to disk.
>>> Paging is just for that the softwar doesn't crash.
>>
>> I suggest you go and google how operating system virtual memory works.
>
>Swapping is for stability and not for performance.

For stability? Seriously?

>No one considers running applications actually swapping.
>
>> What was it intended for then given it was written in basic C?
>
>You didn't write that on a Z-80.

No, but I've got a Z80 C compiler that would happily compile and run it.

Am 23.08.2023 um 08:51 schrieb Muttley@dastardlyhq.com:
> On Tue, 22 Aug 2023 20:50:55 +0200
> Bonita Montero <Bonita.Montero@gmail.com> wrote:
>> Am 22.08.2023 um 20:18 schrieb Muttley@dastardlyhq.com:
>>> On Tue, 22 Aug 2023 20:13:17 +0200
>>> Bonita Montero <Bonita.Montero@gmail.com> wrote:
>>>> Am 22.08.2023 um 19:56 schrieb Muttley@dastardlyhq.com:
>>>>
>>>>> If it has to be paged from disk in the first place the size of the
>>>>> cache is a total irrelevance as far as time taken is concerned.
>>>>
>>>> No one uses software that actually pages to disk.
>>>> Paging is just for that the softwar doesn't crash.
>>>
>>> I suggest you go and google how operating system virtual memory works.
>>
>> Swapping is for stability and not for performance.
>
> For stability? Seriously?

Main memory has a latency from 20 - 30ns per access. My 1TB Samsung
PCIe 4.0 SSD has a synchronous latency of 32us, that's up to factor
1.800. Do you really think that a swappable application is usable ?
Swapping is just for low memory conditions so that the application
won't stop or crash.

> No, but I've got a Z80 C compiler that would happily compile and run it.

I don't believe you. And no one is using Z-80s today.

On 8/23/23 3:59 AM, Bonita Montero wrote:
>
> Main memory has a latency from 20 - 30ns per access. My 1TB Samsung
> PCIe 4.0 SSD has a synchronous latency of 32us, that's up to factor
> 1.800. Do you really think that a swappable application is usable ?
> Swapping is just for low memory conditions so that the application
> won't stop or crash.

Your forgetting that in thst 32us you get a whole sector of data, not
just one word, so the slow down in much less of a factor.

Also, and application being "slow" is better than the application just
can't be run because it uses too much resource (even having the machine
all to itself).

Yes, with todays 64-bit address space, we tend to not need to design in
"swapping" as might have been needed with 16-bit machines, but many
applications are still designed based on the knowledge that they are
likely using more memory that is likely physically present, so they try
to optimize access to take that into account.

Good practice looks at the whole hierarchy, after all, data in Level 1
cache is MUCH faster to access then data in main memory, which is faster
than data on local storage, is much faster than data out on the network.

Forget that and you task suddenly hits a wall and just crawls.

Am 23.08.2023 um 13:52 schrieb Richard Damon:

> Your forgetting that in thst 32us you get a whole sector of data,
> not just one word, so the slow down in much less of a factor.

I'm not forgetting that. I dropped this because the relationship is
that hughe anyway. The sector is the size of a cacheline and when you
switch on your computer the SDRAM's burst size is configured according
to the cacheline size.

> Also, and application being "slow" is better than the application just
> can't be run because it uses too much resource (even having the machine
> all to itself).

That's what I told: swapping is just a stability feature. Applications
are normally not useable when they swap or even the whole machine isn't
usable because essential pages of other processes were swapped out.

> Yes, with todays 64-bit address space, we tend to not need to design
> in "swapping" as might have been needed with 16-bit machines, ...

Swapspace makes sense even if the swap actually isn't touched. When you
commit some memory with VirtualAlloc() under Windows and memory hasn't
been touched yet the kernel has subtracted an according amount of space
from swap so that if swapping would be necessary when the pages are
touched the first time there's enough swap space.
Solaris works the same way. But Linux by default does overcommit so that
an application might crash if there's not enough swapspace if a page is
touched the first time. But Linux can be configured like Windows and
Solaris so that there's enough swap reserved when you allocate memory.
The problem with that under Linux is that if you have a massively for-
king application where the child processes modify only a small portion
of the shared memory there might be a large amount of memory subtracted
from swap.

> ... but many applications are still designed based on the knowledge
> that they are likely using more memory that is likely physically present,
> ...

That's not really an optimizatiob but a common assumption.

On Wed, 23 Aug 2023 09:59:31 +0200
Bonita Montero <Bonita.Montero@gmail.com> wrote:
>Am 23.08.2023 um 08:51 schrieb Muttley@dastardlyhq.com:
>> For stability? Seriously?
>
>Main memory has a latency from 20 - 30ns per access. My 1TB Samsung
>PCIe 4.0 SSD has a synchronous latency of 32us, that's up to factor
>1.800. Do you really think that a swappable application is usable ?
>Swapping is just for low memory conditions so that the application
>won't stop or crash.

And low memory conditions can easily occur on machines running large databases
if large query(s) are happening even with gigs of memory so swap space is still
a thing.

And thats not what I would call "stability" but I'm not going to get into
an argument over word meanings.

>> No, but I've got a Z80 C compiler that would happily compile and run it.
>
>I don't believe you. And no one is using Z-80s today.

Believe what you like. Some TI calculators use Z80s FYI. Besides which
a lot of 8 bit microcontrollers are LESS powerful than a Z80 and they still
get programmed in C as well as assembler.

On Wed, 23 Aug 2023 14:40:36 +0200
Bonita Montero <Bonita.Montero@gmail.com> wrote:
>Am 23.08.2023 um 13:52 schrieb Richard Damon:
>> Yes, with todays 64-bit address space, we tend to not need to design
>> in "swapping" as might have been needed with 16-bit machines, ...
>
>Swapspace makes sense even if the swap actually isn't touched. When you
>commit some memory with VirtualAlloc() under Windows and memory hasn't

Does windows still use a filesystem file for swap making it horrendously
slow as swap data has to go in and out via the high level filesystem layer
or has MS discovered the concept of swap partitions yet?

>The problem with that under Linux is that if you have a massively for-
>king application where the child processes modify only a small portion
>of the shared memory there might be a large amount of memory subtracted
>from swap.
>
>> ... but many applications are still designed based on the knowledge
>> that they are likely using more memory that is likely physically present,
>> ...
>
>That's not really an optimizatiob but a common assumption.

Sadly not in the Java world where "grab as much memory as you can on
startup" seems to be the JVM motto.

Am 23.08.2023 um 16:24 schrieb Muttley@dastardlyhq.com:

> Does windows still use a filesystem file for swap making it horrendously
> slow as swap data has to go in and out via the high level filesystem layer
> or has MS discovered the concept of swap partitions yet?

That's not slow if you simply memorize the cluster-addresses of the
blocks holding the pagefile in non-pageable memory. That's that simple
that I think it's the way how ms does that.

> Sadly not in the Java world where "grab as much memory as you can on
> startup" seems to be the JVM motto.

That's not true. Java-applications would be even less responsive
since a garbage collection involves copying the whole heap to a
new heap.

Am 23.08.2023 um 16:20 schrieb Muttley@dastardlyhq.com:

> And low memory conditions can easily occur on machines running large
> databases ...

That's not true. If you have a database you chose the size of the
buffer pool and the heaps in a way that the machine doesn't swap.
Swapping is simply extremely uncommon.

> Believe what you like. Some TI calculators use Z80s FYI.

Prove that.

> Besides which a lot of 8 bit microcontrollers are LESS powerful
> than a Z80 and they still get programmed in C as well as assembler.

Show me some of them.

On Wed, 23 Aug 2023 16:30:01 +0200
Bonita Montero <Bonita.Montero@gmail.com> wrote:
>Am 23.08.2023 um 16:24 schrieb Muttley@dastardlyhq.com:
>> Sadly not in the Java world where "grab as much memory as you can on
>> startup" seems to be the JVM motto.
>
>That's not true. Java-applications would be even less responsive
>since a garbage collection involves copying the whole heap to a
>new heap.

Certainly seemed to be the case back when I was involved in Java. One would
hope its evolved since then.

Am 23.08.2023 um 16:36 schrieb Muttley@dastardlyhq.com:
> On Wed, 23 Aug 2023 16:30:01 +0200
> Bonita Montero <Bonita.Montero@gmail.com> wrote:
>> Am 23.08.2023 um 16:24 schrieb Muttley@dastardlyhq.com:
>>> Sadly not in the Java world where "grab as much memory as you can on
>>> startup" seems to be the JVM motto.
>>
>> That's not true. Java-applications would be even less responsive
>> since a garbage collection involves copying the whole heap to a
>> new heap.
>
> Certainly seemed to be the case back when I was involved in Java. One would
> hope its evolved since then.
>

Others just have the composure to devote
themselves to complex problems, you don't.

On Wed, 23 Aug 2023 16:32:19 +0200
Bonita Montero <Bonita.Montero@gmail.com> wrote:
>Am 23.08.2023 um 16:20 schrieb Muttley@dastardlyhq.com:
>
>> And low memory conditions can easily occur on machines running large
>> databases ...
>
>That's not true. If you have a database you chose the size of the
>buffer pool and the heaps in a way that the machine doesn't swap.
>Swapping is simply extremely uncommon.

Depends on the DB. If like Oracle its tablespaces are on bespoke disk
partitions then it will do the equivalent of swapping if it needs to.

>> Believe what you like. Some TI calculators use Z80s FYI.
>
>Prove that.

You ever heard of google? Try "ti calculator z80". You never know what
you might find.

>> Besides which a lot of 8 bit microcontrollers are LESS powerful
>> than a Z80 and they still get programmed in C as well as assembler.
>
>Show me some of them.

FFS man, asre you a baby that needs to be spoon fed? Grow up!
Google "8 bit PIC".

We already had a "discussion" with you a while where you demonstrated your
complete and utter pig ignorance of the embedded world. Perhaps don't make
a fool of yourself again.

On Wed, 23 Aug 2023 16:38:51 +0200
Bonita Montero <Bonita.Montero@gmail.com> wrote:
>Am 23.08.2023 um 16:36 schrieb Muttley@dastardlyhq.com:
>> On Wed, 23 Aug 2023 16:30:01 +0200
>> Bonita Montero <Bonita.Montero@gmail.com> wrote:
>>> Am 23.08.2023 um 16:24 schrieb Muttley@dastardlyhq.com:
>>>> Sadly not in the Java world where "grab as much memory as you can on
>>>> startup" seems to be the JVM motto.
>>>
>>> That's not true. Java-applications would be even less responsive
>>> since a garbage collection involves copying the whole heap to a
>>> new heap.
>>
>> Certainly seemed to be the case back when I was involved in Java. One would
>> hope its evolved since then.
>>
>
>Others just have the composure to devote
>themselves to complex problems, you don't.

It wasn't my problem other than the JVM was killing our servers performance
so go swivel on it.

Am 23.08.2023 um 16:43 schrieb Muttley@dastardlyhq.com:

> On Wed, 23 Aug 2023 16:38:51 +0200
> Bonita Montero <Bonita.Montero@gmail.com> wrote:

>> Others just have the composure to devote
>> themselves to complex problems, you don't.

> It wasn't my problem other than the JVM was killing our
> servers performance so go swivel on it.

Java Frameworks are really complex and you're at the level of
simple details. So I think you never were involved in such projects.

Am 23.08.2023 um 16:41 schrieb Muttley@dastardlyhq.com:
> On Wed, 23 Aug 2023 16:32:19 +0200
> Bonita Montero <Bonita.Montero@gmail.com> wrote:
>> Am 23.08.2023 um 16:20 schrieb Muttley@dastardlyhq.com:
>>
>>> And low memory conditions can easily occur on machines running large
>>> databases ...
>>
>> That's not true. If you have a database you chose the size of the
>> buffer pool and the heaps in a way that the machine doesn't swap.
>> Swapping is simply extremely uncommon.
>
> Depends on the DB. If like Oracle its tablespaces are on bespoke disk
> partitions then it will do the equivalent of swapping if it needs to.

The Oracle Data Warehousing Guide says that the performance
advantage of raw partitions is between three and five percent.
And you lose much comfort features like shnapshots. So no one
actually uses raw partitions.
And this feature has nothing to do with how you dimension your
buffer pool or your heaps.

>
>>> Believe what you like. Some TI calculators use Z80s FYI.
>>
>> Prove that.
>
> You ever heard of google? Try "ti calculator z80". You never know what
> you might find.

LOL

>>> Besides which a lot of 8 bit microcontrollers are LESS powerful
>>> than a Z80 and they still get programmed in C as well as assembler.
>>
>> Show me some of them.
>
> FFS man, asre you a baby that needs to be spoon fed? Grow up!
> Google "8 bit PIC".

LOL

>
> We already had a "discussion" with you a while where you demonstrated your
> complete and utter pig ignorance of the embedded world. Perhaps don't make
> a fool of yourself again.
>
>

On Wednesday, August 23, 2023 at 10:20:59 AM UTC-4, Mut...@dastardlyhq.com wrote:

> And low memory conditions can easily occur on machines running large databases
> if large query(s) are happening even with gigs of memory so swap space is still
> a thing.
>
When I started programming in the mid eighties, memory was a scarce resource, and
one of my earliest lessons was that there were only three numbers that counted:
zero, one, and as many as you like. Many years later, when memory was abundant,
I found it astonishing that so many applications were nonperforming when
processing large amounts of data. Few were designed to process "as many as you like",
a great number, from desktop photo imaging software to multimillion dollar capital
markets applications, were designed to load everything into memory, as if it were a
zero cost resource. But large database queries need not translate to unbounded
memory requirements, RDBMs have cursors, caching on the server, and other
features to avoid it. Applications that swapped too much were generally poorly
designed.
Daniel

On Wed, 23 Aug 2023 16:45:06 +0200
Bonita Montero <Bonita.Montero@gmail.com> wrote:
>Am 23.08.2023 um 16:43 schrieb Muttley@dastardlyhq.com:
>
>> On Wed, 23 Aug 2023 16:38:51 +0200
>> Bonita Montero <Bonita.Montero@gmail.com> wrote:
>
>>> Others just have the composure to devote
>>> themselves to complex problems, you don't.
>
>> It wasn't my problem other than the JVM was killing our
>> servers performance so go swivel on it.
>
>Java Frameworks are really complex and you're at the level of
>simple details. So I think you never were involved in such projects.

You must be a very insecure person constantly putting others down to make
yourself feel better.

What job do you do? Do you even have one?

On Wed, 23 Aug 2023 16:47:46 +0200
Bonita Montero <Bonita.Montero@gmail.com> wrote:
>Am 23.08.2023 um 16:41 schrieb Muttley@dastardlyhq.com:
>> Depends on the DB. If like Oracle its tablespaces are on bespoke disk
>> partitions then it will do the equivalent of swapping if it needs to.
>
>The Oracle Data Warehousing Guide says that the performance
>advantage of raw partitions is between three and five percent.
>And you lose much comfort features like shnapshots. So no one
>actually uses raw partitions.

Oh right. You're an Oracle expert too now? :)

>And this feature has nothing to do with how you dimension your
>buffer pool or your heaps.

Its a type of paging.

>> FFS man, asre you a baby that needs to be spoon fed? Grow up!
>> Google "8 bit PIC".
>
>LOL

Get a grip aspie.

Am 23.08.2023 um 19:49 schrieb Muttley@dastardlyhq.com:

> You must be a very insecure person constantly
> putting others down to make yourself feel better.

No, I just feel uncomfortable if people produce such images.
You're simply lying.

Am 23.08.2023 um 19:51 schrieb Muttley@dastardlyhq.com:
> On Wed, 23 Aug 2023 16:47:46 +0200
> Bonita Montero <Bonita.Montero@gmail.com> wrote:
>> Am 23.08.2023 um 16:41 schrieb Muttley@dastardlyhq.com:
>>> Depends on the DB. If like Oracle its tablespaces are on bespoke disk
>>> partitions then it will do the equivalent of swapping if it needs to.
>>
>> The Oracle Data Warehousing Guide says that the performance
>> advantage of raw partitions is between three and five percent.
>> And you lose much comfort features like shnapshots. So no one
>> actually uses raw partitions.
>
> Oh right. You're an Oracle expert too now? :)

I have a OCP-DBA for Oracle 8i and I read that 20 years ago.
I checked that in the current DWH-guide and they reverted
that statement to five to ten percent.

>> And this feature has nothing to do with how you dimension your
>> buffer pool or your heaps.

> Its a type of paging.

Paging is synchronous, i.e. the thread which pages is stopped until
the page is fetched. All modern database engines use asynchronous
I/O, thereby issuing multiple requests and answering the responses
often in a totally different order.
And it's also not paging because this is explicitly issued by the
database server, whereas real paging is a transparent memory access
which isn't noticed by the application.

On Wednesday, August 23, 2023 at 1:49:23 PM UTC-4, Mut...@dastardlyhq.com wrote:

> You must be a very insecure person constantly putting others down to make
> yourself feel better.
>
Regrettably, many participants on this group seem to like to put others down,
like Old Betsie in the village: giving them a piece of their mind. Öö Tiib does it,
Scott Lurndal does it, James, Keith, David, you're doing it now. The heavyweights
that used to post here, they never did that, they were invariably civil. But they're gone.

Best regards,
Daniel

Am 23.08.2023 um 20:13 schrieb Daniel:

> Regrettably, many participants on this group seem to like to put others down,
> like Old Betsie in the village: giving them a piece of their mind. Öö Tiib
> does it, Scott Lurndal does it, James, Keith, David, you're doing it now.
> The heavyweights that used to post here, they never did that, they were
> invariably civil. But they're gone.

Muttley uses the scheme I mentioned very often,
at some point you have to say something.

On 8/23/2023 11:17 AM, Bonita Montero wrote:
> Am 23.08.2023 um 20:13 schrieb Daniel:
>
>> Regrettably, many participants on this group seem to like to put
>> others down,
>> like Old Betsie in the village: giving them a piece of their mind. Öö
>> Tiib
>> does it, Scott Lurndal does it, James, Keith, David, you're doing it now.
>> The heavyweights that used to post here, they never did that, they were
>> invariably civil. But they're gone.
>
> Muttley uses the scheme I mentioned very often,
> at some point you have to say something.
>

Pot Kettle? Bonita, you have a lot of experience wrt verbal assaults, right?

Am 23.08.2023 um 20:20 schrieb Chris M. Thomasson:
> On 8/23/2023 11:17 AM, Bonita Montero wrote:
>> Am 23.08.2023 um 20:13 schrieb Daniel:
>>
>>> Regrettably, many participants on this group seem to like to put
>>> others down,
>>> like Old Betsie in the village: giving them a piece of their mind. Öö
>>> Tiib
>>> does it, Scott Lurndal does it, James, Keith, David, you're doing it
>>> now.
>>> The heavyweights that used to post here, they never did that, they were
>>> invariably civil. But they're gone.
>>
>> Muttley uses the scheme I mentioned very often,
>> at some point you have to say something.
>>
>
> Pot Kettle? Bonita, you have a lot of experience wrt verbal assaults,
> right?

Muttley is just annoying. If I show a solution to a problem and think
it is technically elegant, he somehow wants to explain it away because
my approach is always too complicated for him. He never understands the
advantages of my approach and when I then explain them, in the end he
doesn't say anything more, except again that it's all too complicated.
It's been like this for months now. The guy is disturbed.
And because he then compares himself to me at that point, he produces
such images as saying that he was involved in some Java project,
although based on other statements about Java you can say that
he only has clichéd views of it.

On Wednesday, August 23, 2023 at 2:21:18 PM UTC-4, Chris M. Thomasson wrote:
> >
> Bonita, you have a lot of experience wrt verbal assaults, right?

How would you know? You don't seem to have _any_ experience of it
yourself :-) Alf neither.

Daniel