Prev: String decoding
Next: A replacement for about-to-be-closed NGs
From: Helmut Meukel on 27 May 2010 12:41
"ralph" <nt_consulting64(a)yahoo.net> schrieb im Newsbeitrag
news:3jtsv5t32ncg4ocg1sue50gbn9mnfpt6a6(a)4ax.com...
> On Thu, 27 May 2010 09:19:19 +0100, MM <kylix_is(a)yahoo.co.uk> wrote:
>
>>On Sun, 23 May 2010 14:30:40 -0500, ralph <nt_consulting64(a)yahoo.net>
>>wrote:
>>
>>>Also the increase in available physical RAM always helps to improve
>>>performance.
>>
>>I have now repeated my tests with ADO, this time on the 1gb RAM PC but
>>with 98SE loaded. The findings are surprising. (Timings in seconds for
>>same Access 97 mdb (local copy) and same query.)
>>
>>1gb PC (98SE) 512mb PC (98SE)
>>pass 1 36 22
>>pass 2 37 22
>>
>
> LOL
>
> Not since the ancient days of expanded/extended memory managers and
> mixed memory modules can I remember a case where increased RAM
> degraded performance. (You can reach a point of diminishing returns,
> but *backwards*??? <g>)
>
> Enjoy the adventure.
>
> -ralph


Ralph,

I *do* remember such cases. There was an Intel chip-set which didn't
buffer all memory. Increasing memory degraded memory performance.

Helmut.

From: ralph on 27 May 2010 13:52
On Thu, 27 May 2010 16:24:37 +0100, MM <kylix_is(a)yahoo.co.uk> wrote:


>>
>>Not since the ancient days of expanded/extended memory managers and
>>mixed memory modules can I remember a case where increased RAM
>>degraded performance. (You can reach a point of diminishing returns,
>>but *backwards*??? <g>)
>>
>>Enjoy the adventure.
>>
>>-ralph
>
>Don't forget that Windows 98 doesn't 'like' 1gb of RAM! In fact, you
>HAVE to tweak system.ini else it won't boot.

I didn't know that. I haven't had a PC with less than 2gb of RAM since
my first 386. And certainly never used Win98 with less.

> ... Maybe there are other
>'things' happening behind the scenes that cause the extra memory to be
>ignored or other weird stuff.
>

Has to be something. I find it hard to believe the extra 512kb
wouldn't be advantageous.

>Mind you, I haven't given up on, f'rinstance, SQLite with its FTS3
>indexing scheme to replace Access altogether.
>

There are always alternatives.

Within a specific problem domain with limited requirements a dedicated
hand-crafted local data store will always run rings around a
commercial RDBMS product. (The trade-off coming from time to develop
and flexiblity.)

SQLite is one of the better choices.
From: MM on 27 May 2010 15:18
On Thu, 27 May 2010 10:24:18 -0500, DanS
<t.h.i.s.n.t.h.a.t(a)r.o.a.d.r.u.n.n.e.r.c.o.m> wrote:

>MM <kylix_is(a)yahoo.co.uk> wrote in
>news:atqqv5pjgfk60v9q4f1d6nme2rb80l117r(a)4ax.com:
>
>> On Wed, 26 May 2010 09:51:14 -0700, Gutless Umbrella
>> Carrying Sissy <taustinca(a)gmail.com> wrote:
>>
>>>If the problem has gone away, why bring it up again?
>>
>> Er, to inform others?
>>
>> MM
>
>You never did (,or I just didn't see it,) post your timings of
>
>ADO on Win2K vs. DAO on Win2K for the same mdb/query.
>
>For a proper scientific experiment, there can only be one
>variable and all other things must be constant.

I never tested DAO on W2K. I only used DAO on 98SE because ADO
returned such poor performance. But I want to use ADO! DAO is 'old
hat' now. Been there. Done that. I can do all kinds of wondrous things
with ADO that I can only dream about with DAO. So the minute I
discovered, quite by chance, that a simple switch to W2K improved
performance so dramatically, I chucked all the DAO stuff out. Having,
for example, to use the data control to provide a link between a DAO
recordset and the MSFlexGrid made me realise that I was resurrecting
the Ark.

Anyways, SQLite looks like the long-term aim, especially with the aid
of Olaf Schmidt's stuff. But there are only so many hours in the day
and I have to eat, sleep and take a bath occasionally...

MM
From: Henning on 27 May 2010 15:54


--
Customer Hatred Knows No Bounds at MSFT
Free usenet access at http://www.eternal-september.org
ClassicVB Users Regroup! comp.lang.basic.visual.misc
"MM" <kylix_is(a)yahoo.co.uk> skrev i meddelandet
news:j1htv5t0fmpirl0r3nko8m3q01317cghb8(a)4ax.com...
> On Thu, 27 May 2010 10:24:18 -0500, DanS
> <t.h.i.s.n.t.h.a.t(a)r.o.a.d.r.u.n.n.e.r.c.o.m> wrote:
>
>>MM <kylix_is(a)yahoo.co.uk> wrote in
>>news:atqqv5pjgfk60v9q4f1d6nme2rb80l117r(a)4ax.com:
>>
>>> On Wed, 26 May 2010 09:51:14 -0700, Gutless Umbrella
>>> Carrying Sissy <taustinca(a)gmail.com> wrote:
>>>
>>>>If the problem has gone away, why bring it up again?
>>>
>>> Er, to inform others?
>>>
>>> MM
>>
>>You never did (,or I just didn't see it,) post your timings of
>>
>>ADO on Win2K vs. DAO on Win2K for the same mdb/query.
>>
>>For a proper scientific experiment, there can only be one
>>variable and all other things must be constant.
>
> I never tested DAO on W2K. I only used DAO on 98SE because ADO
> returned such poor performance. But I want to use ADO! DAO is 'old
> hat' now. Been there. Done that. I can do all kinds of wondrous things
> with ADO that I can only dream about with DAO. So the minute I
> discovered, quite by chance, that a simple switch to W2K improved
> performance so dramatically, I chucked all the DAO stuff out. Having,
> for example, to use the data control to provide a link between a DAO
> recordset and the MSFlexGrid made me realise that I was resurrecting
> the Ark.
>
> Anyways, SQLite looks like the long-term aim, especially with the aid
> of Olaf Schmidt's stuff. But there are only so many hours in the day
> and I have to eat, sleep and take a bath occasionally...
>
> MM

Wow, what a luxury life u live ;)

/Henning


From: Mike B on 27 May 2010 18:53

"ralph" <nt_consulting64(a)yahoo.net> wrote in message
news:4natv5hoenlobqhcu4jkjn1nqbbfugu6sl(a)4ax.com...
> On Thu, 27 May 2010 16:24:37 +0100, MM <kylix_is(a)yahoo.co.uk> wrote:
>
>
>>>
>>>Not since the ancient days of expanded/extended memory managers and
>>>mixed memory modules can I remember a case where increased RAM
>>>degraded performance. (You can reach a point of diminishing returns,
>>>but *backwards*??? <g>)
>>>
>>>Enjoy the adventure.
>>>
>>>-ralph
>>
>>Don't forget that Windows 98 doesn't 'like' 1gb of RAM! In fact, you
>>HAVE to tweak system.ini else it won't boot.
>
> I didn't know that. I haven't had a PC with less than 2gb of RAM since
> my first 386. And certainly never used Win98 with less.

I would've liked to see that MoBo capable of holding 2gb in 1986. Remember,
the memory chips were press-in, not on pcboards.

I ran my manufacturing business on a PC Limited (before it was called Dell)
386 with 2mb of RAM and a PC-MOS operating system (multiple DOS memory
partitions mapped to Wyse terminals).

The TI-286 that it replaced had a 1.0mb expansion card (on top of the base
640k) that was a $2500 option, so I'm thinking 2gb of memory then would've
been a bank buster..


>> ... Maybe there are other
>>'things' happening behind the scenes that cause the extra memory to be
>>ignored or other weird stuff.
>>
>
> Has to be something. I find it hard to believe the extra 512kb
> wouldn't be advantageous.
>
>>Mind you, I haven't given up on, f'rinstance, SQLite with its FTS3
>>indexing scheme to replace Access altogether.
>>
>
> There are always alternatives.
>
> Within a specific problem domain with limited requirements a dedicated
> hand-crafted local data store will always run rings around a
> commercial RDBMS product. (The trade-off coming from time to develop
> and flexiblity.)
>
> SQLite is one of the better choices.


First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4 5
Prev: String decoding
Next: A replacement for about-to-be-closed NGs