Hardware (Read Only)
cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

Ram and Performance

17 REPLIES 17
Reply
Message 1 of 18
Anonymous
274 Views, 17 Replies

Ram and Performance

I'm told that if you buy too much ram, it would bog the performance of your
computer down. I currently work with 2GB of ram (lots of large
images/aerials) and wondered if there was an optimum memory setting withing
CAD or the operating system. Any suggestions?

Thanks,
David
17 REPLIES 17
Message 2 of 18
Anonymous
in reply to: Anonymous

what OS?

--
remove .ns to email
David wrote in message
news:F8D7D90206A5B3279E7DE33728444653@in.WebX.maYIadrTaRb...
> I'm told that if you buy too much ram, it would bog the performance of
your
> computer down. I currently work with 2GB of ram (lots of large
> images/aerials) and wondered if there was an optimum memory setting
withing
> CAD or the operating system. Any suggestions?
>
> Thanks,
> David
>
>
Message 3 of 18
Anonymous
in reply to: Anonymous

RAM will not bog down the computer. Fill it up and max it out. If you are
still using Win98 then just change the RAM setting in the .ini file.
Message 4 of 18
Anonymous
in reply to: Anonymous

You are correct, adding lots and lots of RAM CAN have a negative impact on
system performance. A system using dense RAM sticks will perform benchmark
tests a little more slowly than a system using smaller RAM modules. This is
one reason why benchmarks are often run on systems with 256 or 512 MB of
RAM. In addition, unless the system supports dual channel memory, having 1
memory module is faster than 2. If it does support dual ddr, or whatever
they're calling it these days, an even number of memory modules will give
you slightly better memory performance than an odd number.

However, if you are really using 1-2 GB of memory in your day to day work,
having that much physical RAM will make your system perform much better than
a system with less memory.

Jon




"David" wrote in message
news:F8D7D90206A5B3279E7DE33728444653@in.WebX.maYIadrTaRb...
> I'm told that if you buy too much ram, it would bog the performance of
your
> computer down. I currently work with 2GB of ram (lots of large
> images/aerials) and wondered if there was an optimum memory setting
withing
> CAD or the operating system. Any suggestions?
>
> Thanks,
> David
>
>
Message 5 of 18
Anonymous
in reply to: Anonymous

Windows NT 4.0 sp6 at work with PC800@400MHZ, ECC, 8X256, 2X4 RIMM Risers on
a dual chip 866MHZ PIIIXeon Dell workstation

Four Rambus Ram 512k chips on a 1.8G processsor with XP Pro at home.

Is there a certain page swap file size I should be using?

"Zach" wrote in message
news:2C7269D93EB0985B9F42D6F34047F96B@in.WebX.maYIadrTaRb...
> what OS?
>
> --
> remove .ns to email
> David wrote in message
> news:F8D7D90206A5B3279E7DE33728444653@in.WebX.maYIadrTaRb...
> > I'm told that if you buy too much ram, it would bog the performance of
> your
> > computer down. I currently work with 2GB of ram (lots of large
> > images/aerials) and wondered if there was an optimum memory setting
> withing
> > CAD or the operating system. Any suggestions?
> >
> > Thanks,
> > David
> >
> >
>
>
Message 6 of 18
Anonymous
in reply to: Anonymous

John,

On Wed, 16 Jul 2003 06:42:50 -0700, "Jon Rizzo"
com> wrote:

>...adding lots and lots of RAM CAN have a negative impact on
>system performance.

Only if you are running some brain-dead OS like Windows 98 or ME. A
modern OS like Win2K, Linux or XP will run better with more RAM. After
512MB it might not make much of a difference until you work with very
large files.

>A system using dense RAM sticks will perform benchmark
>tests a little more slowly than a system using smaller RAM modules.

Would you have any data for evidence of this? I've never heard of such
a thing.

Matt
mstachoni@comcast.net
mstachoni@bhhtait.com



On Wed, 16 Jul 2003 06:42:50 -0700, "Jon Rizzo"
com> wrote:

>You are correct, adding lots and lots of RAM CAN have a negative impact on
>system performance. A system using dense RAM sticks will perform benchmark
>tests a little more slowly than a system using smaller RAM modules. This is
>one reason why benchmarks are often run on systems with 256 or 512 MB of
>RAM. In addition, unless the system supports dual channel memory, having 1
>memory module is faster than 2. If it does support dual ddr, or whatever
>they're calling it these days, an even number of memory modules will give
>you slightly better memory performance than an odd number.
>
>However, if you are really using 1-2 GB of memory in your day to day work,
>having that much physical RAM will make your system perform much better than
>a system with less memory.
Message 7 of 18
Anonymous
in reply to: Anonymous

> Would you have any data for evidence of this? I've never heard of such
> a thing.

It is a well established fact - more chips/module and/or more modules =
slower timings/command rate. If you are not overclocking, the default
settings for these modules are necessarily slower to ensure stability in a
wide range of systems. If you are overclocking and/or tweaking your own
system for max. performance, you will be able to squeeze more performance
out of a system that has a single low density memory module vs. one with
multiple high density modules.

As I said, however, the difference shows up IN BENCHMARKS. I have no idea
if casual users would notice any difference. Obviously, a system that is
paging to disk cannot compare to a system that is not, so if you USE more
than 1 GB of RAM, then by all means it is faster to have more memory.

A quick search on tom's hw yielded this article. It refers to this fact,
but doesn't really explain all there is to it. There are more in depth
articles out there, but this is just one mention that I found with about a
minute of searching.
http://www6.tomshardware.com/howto/20030701/memory_tuning-07.html

Jon
Message 8 of 18
Anonymous
in reply to: Anonymous

John,

Call me skeptical. All I asked for was some references or data - even
if it is just a benchmark session. But I'm certainly not going to
believe it just because you (or Tom's) say it is so without some
metric to back it up. Sorry.

RAM timings are a function of the RAM module quality, not how many
modules you have in the system. You can have 8 RAM sticks installed
and STILL get 2-2-2 CAS timings - you just have to buy the right ones
and tweak the BIOS as required (which isn't the same as overclocking).

I've read that Tom's article, and if you believe it (hey, Tom's been
wrong before), you first response STILL doesn't make sense to me.

You said:
> A system using dense RAM sticks will perform benchmark
>tests a little more slowly than a system using smaller RAM modules

Which infers that using 4-512MB modules to get to 2 GB rather than 2
1GB modules makes your machine faster.

But Tom's article states:
> If you've filled all your memory banks, you'll generally have
>to increase the rate from one to two clock cycles to keep your
>system stable. Unfortunately, that will also impair performance by up to three percent

..so, color me confused on your position.

Matt
mstachoni@comcast.net
mstachoni@bhhtait.com


On Wed, 16 Jul 2003 14:10:31 -0700, "Jon Rizzo"
com> wrote:

>> Would you have any data for evidence of this? I've never heard of such
>> a thing.
>
>It is a well established fact - more chips/module and/or more modules =
>slower timings/command rate. If you are not overclocking, the default
>settings for these modules are necessarily slower to ensure stability in a
>wide range of systems. If you are overclocking and/or tweaking your own
>system for max. performance, you will be able to squeeze more performance
>out of a system that has a single low density memory module vs. one with
>multiple high density modules.
>
>As I said, however, the difference shows up IN BENCHMARKS. I have no idea
>if casual users would notice any difference. Obviously, a system that is
>paging to disk cannot compare to a system that is not, so if you USE more
>than 1 GB of RAM, then by all means it is faster to have more memory.
>
>A quick search on tom's hw yielded this article. It refers to this fact,
>but doesn't really explain all there is to it. There are more in depth
>articles out there, but this is just one mention that I found with about a
>minute of searching.
>http://www6.tomshardware.com/howto/20030701/memory_tuning-07.html
>
>Jon
>
Message 9 of 18
Anonymous
in reply to: Anonymous

I don't understand why you are confused - the quote you posted only
reinforces what I said. Perhaps if you understood what "command rate" is,
my point would be a little more clear. Increasing the command rate makes
your memory run more slowly, not faster.

As I said, this was the first article that I stumbled across doing some
quick research for you. I have read similar articles on other hardware
sites, but I have better things to do than hunt them down so that you will
"believe me". I don't really care if you believe me or not. This is a
message board, not a debate society. Do your own research, and come to your
own conclusion.



"Matt Stachoni" wrote in message
news:as5chvsatfos941v31ktgdv5dpumgmjrvs@4ax.com...
> John,
>
>
> RAM timings are a function of the RAM module quality, not how many
> modules you have in the system. You can have 8 RAM sticks installed
> and STILL get 2-2-2 CAS timings - you just have to buy the right ones
> and tweak the BIOS as required (which isn't the same as overclocking).
>
> I've read that Tom's article, and if you believe it (hey, Tom's been
> wrong before), you first response STILL doesn't make sense to me.
>
> You said:
> > A system using dense RAM sticks will perform benchmark
> >tests a little more slowly than a system using smaller RAM modules
>
>
> But Tom's article states:
> > If you've filled all your memory banks, you'll generally have
> >to increase the rate from one to two clock cycles to keep your
> >system stable. Unfortunately, that will also impair performance by up to
three percent
>
> ..so, color me confused on your position.
>
> Matt
> mstachoni@comcast.net
> mstachoni@bhhtait.com
>
>
> On Wed, 16 Jul 2003 14:10:31 -0700, "Jon Rizzo"
> com> wrote:
>
> >> Would you have any data for evidence of this? I've never heard of such
> >> a thing.
> >
> >It is a well established fact - more chips/module and/or more modules =
> >slower timings/command rate. If you are not overclocking, the default
> >settings for these modules are necessarily slower to ensure stability in
a
> >wide range of systems. If you are overclocking and/or tweaking your own
> >system for max. performance, you will be able to squeeze more performance
> >out of a system that has a single low density memory module vs. one with
> >multiple high density modules.
> >
> >As I said, however, the difference shows up IN BENCHMARKS. I have no
idea
> >if casual users would notice any difference. Obviously, a system that is
> >paging to disk cannot compare to a system that is not, so if you USE more
> >than 1 GB of RAM, then by all means it is faster to have more memory.
> >
> >A quick search on tom's hw yielded this article. It refers to this fact,
> >but doesn't really explain all there is to it. There are more in depth
> >articles out there, but this is just one mention that I found with about
a
> >minute of searching.
> >http://www6.tomshardware.com/howto/20030701/memory_tuning-07.html
> >
> >Jon
> >
>
Message 10 of 18
Anonymous
in reply to: Anonymous

Very well said. Thank you.
Message 11 of 18
Anonymous
in reply to: Anonymous

John,

Thanks, but I went through Kindergarten just fine, so I know all about
"Command rate" as it applies to memory timings.

However, you are truly splitting hairs here. Setting the command rate
to 1T in the BIOS is generally only possible if you have (a) truly
fantastic memory, such as Crucial's XMS series and (b) a motherboard
that support a user setting this in the first place. Yes, 1T will
perform somewhat faster than 2T (which is more the norm in most
systems by default), and having 4 DRAM modules will definitely make 1T
very hard, if not impossible to attain with any stability.

But I would be hard pressed to find a system as shipped by Dell or
anyone else that had such aggessive memory timings set by default
ANYWAY. Why? Because it's probably not too stable. Setting 1T can also
contribute to the infamous "infinite loop" error with NVidia graphics
drivers.

In the end analysis I think your previous statement that "adding lots
and lots of RAM CAN have a negative impact on system performance" due
to this 1T-2T issue is a bit unfocused, especially in Windows XP and
using memory intensive applications, where adding more memory lowers
HD paging operations and command rate is simply not an issue.

Matt
mstachoni@comcast.net
mstachoni@bhhtait.com
Message 12 of 18
Anonymous
in reply to: Anonymous

Thank you for that insightful critique of my writing style. I feel truly
sorry for your coworkers. Focused or not, nothing that I said is untrue
regardless of whether you "believe me".





"Matt Stachoni" wrote in message
news:uhlehvk5vqg6at79k7cfsfrafgpbode57u@4ax.com...
> John,
>
> Thanks, but I went through Kindergarten just fine, so I know all about
> "Command rate" as it applies to memory timings.
>
> However, you are truly splitting hairs here. Setting the command rate
> to 1T in the BIOS is generally only possible if you have (a) truly
> fantastic memory, such as Crucial's XMS series and (b) a motherboard
> that support a user setting this in the first place. Yes, 1T will
> perform somewhat faster than 2T (which is more the norm in most
> systems by default), and having 4 DRAM modules will definitely make 1T
> very hard, if not impossible to attain with any stability.
>
> But I would be hard pressed to find a system as shipped by Dell or
> anyone else that had such aggessive memory timings set by default
> ANYWAY. Why? Because it's probably not too stable. Setting 1T can also
> contribute to the infamous "infinite loop" error with NVidia graphics
> drivers.
>
> In the end analysis I think your previous statement that "adding lots
> and lots of RAM CAN have a negative impact on system performance" due
> to this 1T-2T issue is a bit unfocused, especially in Windows XP and
> using memory intensive applications, where adding more memory lowers
> HD paging operations and command rate is simply not an issue.
>
> Matt
> mstachoni@comcast.net
> mstachoni@bhhtait.com
>
>
>
Message 13 of 18
Anonymous
in reply to: Anonymous

If you are using a lot of large Aerial images set the swap file size up to
4GB. Set your initial and max to be equal.

Allen

"David" wrote in message
news:A83F56C1EFF50250533900569630EE8B@in.WebX.maYIadrTaRb...
> Windows NT 4.0 sp6 at work with PC800@400MHZ, ECC, 8X256, 2X4 RIMM Risers
on
> a dual chip 866MHZ PIIIXeon Dell workstation
>
> Four Rambus Ram 512k chips on a 1.8G processsor with XP Pro at home.
>
> Is there a certain page swap file size I should be using?
>
Message 14 of 18
Anonymous
in reply to: Anonymous

This is a little wasteful, since windows can only address 3.5 GB of memory
(total). Since he already has 2 GB, this doesn't need to be larger than 1.5
GB.


"Allen Jessup" wrote in message
news:7CED1256618B54F77B9CB44CFB5A1D50@in.WebX.maYIadrTaRb...
> If you are using a lot of large Aerial images set the swap file size up to
> 4GB. Set your initial and max to be equal.
>
> Allen
>
> "David" wrote in message
> news:A83F56C1EFF50250533900569630EE8B@in.WebX.maYIadrTaRb...
> > Windows NT 4.0 sp6 at work with PC800@400MHZ, ECC, 8X256, 2X4 RIMM
Risers
> on
> > a dual chip 866MHZ PIIIXeon Dell workstation
> >
> > Four Rambus Ram 512k chips on a 1.8G processsor with XP Pro at home.
> >
> > Is there a certain page swap file size I should be using?
> >
>
>
Message 15 of 18
Anonymous
in reply to: Anonymous

Something I didn't know. However from personal experience I had someone give
me some 60+MB DXF files and I had to kick the virtual memory up that high to
open them. Since I only have 512MB RAM this is probably why I noticed the
difference. I'll pull mine down to 3GB and see what happens.

Thanks
Allen

"Jon Rizzo" wrote in message
news:2ECBA8ADCD720C9EAA1573102E5D1DB6@in.WebX.maYIadrTaRb...
> This is a little wasteful, since windows can only address 3.5 GB of memory
> (total). Since he already has 2 GB, this doesn't need to be larger than
1.5
> GB.
Message 16 of 18
Anonymous
in reply to: Anonymous

This is not a limitation of Windows; 3.5GB is the maximum amount of
system memory on all x86 systems. Anything more is used to control PCI
circuitry.

Matt
mstachoni@comcast.net
mstachoni@bhhtait.com


On Tue, 22 Jul 2003 06:07:20 -0700, "Jon Rizzo"
com> wrote:

>This is a little wasteful, since windows can only address 3.5 GB of memory
>(total). Since he already has 2 GB, this doesn't need to be larger than 1.5
>GB.
>
>
>"Allen Jessup" wrote in message
>news:7CED1256618B54F77B9CB44CFB5A1D50@in.WebX.maYIadrTaRb...
>> If you are using a lot of large Aerial images set the swap file size up to
>> 4GB. Set your initial and max to be equal.
>>
>> Allen
>>
>> "David" wrote in message
>> news:A83F56C1EFF50250533900569630EE8B@in.WebX.maYIadrTaRb...
>> > Windows NT 4.0 sp6 at work with PC800@400MHZ, ECC, 8X256, 2X4 RIMM
>Risers
>> on
>> > a dual chip 866MHZ PIIIXeon Dell workstation
>> >
>> > Four Rambus Ram 512k chips on a 1.8G processsor with XP Pro at home.
>> >
>> > Is there a certain page swap file size I should be using?
>> >
>>
>>
>
Message 17 of 18
Anonymous
in reply to: Anonymous

Thaks, last word Matt. What would the world do without you?


"Matt Stachoni" wrote in message
news:ishrhvonh3pee8old07mqr3crd58uknntk@4ax.com...
> This is not a limitation of Windows; 3.5GB is the maximum amount of
> system memory on all x86 systems. Anything more is used to control PCI
> circuitry.
>
> Matt
> mstachoni@comcast.net
> mstachoni@bhhtait.com
>
>
> On Tue, 22 Jul 2003 06:07:20 -0700, "Jon Rizzo"
> com> wrote:
>
> >This is a little wasteful, since windows can only address 3.5 GB of
memory
> >(total). Since he already has 2 GB, this doesn't need to be larger than
1.5
> >GB.
> >
> >
> >"Allen Jessup" wrote in message
> >news:7CED1256618B54F77B9CB44CFB5A1D50@in.WebX.maYIadrTaRb...
> >> If you are using a lot of large Aerial images set the swap file size up
to
> >> 4GB. Set your initial and max to be equal.
> >>
> >> Allen
> >>
> >> "David" wrote in message
> >> news:A83F56C1EFF50250533900569630EE8B@in.WebX.maYIadrTaRb...
> >> > Windows NT 4.0 sp6 at work with PC800@400MHZ, ECC, 8X256, 2X4 RIMM
> >Risers
> >> on
> >> > a dual chip 866MHZ PIIIXeon Dell workstation
> >> >
> >> > Four Rambus Ram 512k chips on a 1.8G processsor with XP Pro at home.
> >> >
> >> > Is there a certain page swap file size I should be using?
> >> >
> >>
> >>
> >
>
Message 18 of 18
Anonymous
in reply to: Anonymous

On Wed, 23 Jul 2003 05:00:04 -0700, "Jon Rizzo"
com> wrote:

>Thaks, last word Matt.

Hey, I was just clarifying your post, which implied a Windows-specific
problem in terms of addressible memory. I've heard TONS of misleading
or downright wrong computer related statements in this and many other
newsgroups. Especially when related to Windows memory management.

Actually, I just re-read your post and it IS wrong.

You said, in regards to swap file size, that "windows can only address
3.5 GB of memory (total)."

This is incorrect. While there is a hard limit of 3.5GB usable
addressable system RAM, Windows XP and 2000 can "address" as much
_virtual_ memory as you have hard disk space (well, up to 16
Terabytes, the upper limit on a Windows file size). It simply swaps
out pages of system memory to the hard disk. Hence the name, "Swap
file."

So, if you want to make your Swap file 20 GB you can certainly do so,
and Windows will use it. Of course, this becomes less and less
efficient because Windows has to keep track of the addresses in system
RAM, so it's definitely a point of diminishing returns.

>What would the world do without you?

So, you can either read my posts or ignore them. Doesn't matter to me
or anyone else. But I do speak with a certain degree of expertise.

Matt
mstachoni@comcast.net
mstachoni@bhhtait.com




>
>
>"Matt Stachoni" wrote in message
>news:ishrhvonh3pee8old07mqr3crd58uknntk@4ax.com...
>> This is not a limitation of Windows; 3.5GB is the maximum amount of
>> system memory on all x86 systems. Anything more is used to control PCI
>> circuitry.
>>
>> Matt
>> mstachoni@comcast.net
>> mstachoni@bhhtait.com
>>
>>
>> On Tue, 22 Jul 2003 06:07:20 -0700, "Jon Rizzo"
>> com> wrote:
>>
>> >This is a little wasteful, since windows can only address 3.5 GB of
>memory
>> >(total). Since he already has 2 GB, this doesn't need to be larger than
>1.5
>> >GB.
>> >
>> >
>> >"Allen Jessup" wrote in message
>> >news:7CED1256618B54F77B9CB44CFB5A1D50@in.WebX.maYIadrTaRb...
>> >> If you are using a lot of large Aerial images set the swap file size up
>to
>> >> 4GB. Set your initial and max to be equal.
>> >>
>> >> Allen
>> >>
>> >> "David" wrote in message
>> >> news:A83F56C1EFF50250533900569630EE8B@in.WebX.maYIadrTaRb...
>> >> > Windows NT 4.0 sp6 at work with PC800@400MHZ, ECC, 8X256, 2X4 RIMM
>> >Risers
>> >> on
>> >> > a dual chip 866MHZ PIIIXeon Dell workstation
>> >> >
>> >> > Four Rambus Ram 512k chips on a 1.8G processsor with XP Pro at home.
>> >> >
>> >> > Is there a certain page swap file size I should be using?
>> >> >
>> >>
>> >>
>> >
>>
>

Can't find what you're looking for? Ask the community or share your knowledge.

Post to forums  

Autodesk Design & Make Report