PDA

View Full Version : CAS latency really that important?


Spaceman
March 13th 06, 06:21 AM
is CAS latency really that important? i see very expensive modules out there
with a cas 3 latency, and less expensive ones with a cas 2 latency. so i'm a
bit confused.

Ed Light
March 13th 06, 08:37 AM
It's only worth a few % in performance.


--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at

Thanks, robots.

Bring the Troops Home:
http://bringthemhomenow.org

Fight Spam:
http://bluesecurity.com

Ed Light
March 13th 06, 09:31 AM
Here's the article.

http://www.tomshardware.com/2004/01/19/ups_and_downs/


--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at

Thanks, robots.

Bring the Troops Home:
http://bringthemhomenow.org

Fight Spam:
http://bluesecurity.com

Ed Light
March 13th 06, 10:39 AM
Yipes, I found another article that shows more differences.

http://www.anandtech.com/memory/showdoc.aspx?i=2392&p=1


--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at

Thanks, robots.

Bring the Troops Home:
http://bringthemhomenow.org

Fight Spam:
http://bluesecurity.com

Conor
March 13th 06, 06:45 PM
In article >, Spaceman
says...
> is CAS latency really that important?

Absolutely.

> i see very expensive modules out there
> with a cas 3 latency, and less expensive ones with a cas 2 latency. so i'm a
> bit confused.
>
CAS3 is slow, CAS2 is faster. You want as lower figure as possible.

--
Conor,

Same ****, different day.

rex roland
March 14th 06, 03:52 AM
Conor wrote:
> In article >, Spaceman
> says...
>> is CAS latency really that important?
>
> Absolutely.
>
>> i see very expensive modules out there
>> with a cas 3 latency, and less expensive ones with a cas 2 latency. so i'm a
>> bit confused.
>>
> CAS3 is slow, CAS2 is faster. You want as lower figure as possible.
>

****ing idiot

John
March 14th 06, 04:28 AM
"rex roland" > wrote in message . ..
> Conor wrote:
> > In article >, Spaceman
> > says...
> >> is CAS latency really that important?
> >
> > Absolutely.
> >
> >> i see very expensive modules out there
> >> with a cas 3 latency, and less expensive ones with a cas 2 latency. so i'm a
> >> bit confused.
> >>
> > CAS3 is slow, CAS2 is faster. You want as lower figure as possible.
> >
>
> ****ing idiot

Bingo.

Phil Weldon
March 14th 06, 05:40 AM
'Spaceman' wrote:
| is CAS latency really that important? i see very expensive modules out
there
| with a cas 3 latency, and less expensive ones with a cas 2 latency. so i'm
a
| bit confused.
_____

You seem to have CAS 3 latency and CAS 2 latency reversed. CAS 3 latency is
cheaper and slower than CAS 2 memory. Any price difference that makes CAS 2
memory cheaper than CAS 3 memory must be based on some other factor.

As for the EFFECT of CAS 3 and CAS 2; it depends.

1. CAS latency only comes into play when the initial address is sent out to
memory. For sequential fetches that follow, CAS latency plays no part.
2. If the data or instructions needed are already in the L1 cache or L2
cache, main memory CAS latency plays no part at all. Because data and
instructions tend to be localized in time and space (recently used or in
nearby locations) caches can greatly reduce the affect caused by main memory
being clocked and such a much lower rate than CPUs. This is a complex
dependency and varies according to the type of application, the techniques
used in programming, compiler functions, and CPU design. As a rule of thumb
a 2 way associative 16 KByte L1 data cache the next data needed will likely
be in the L1 cache (requiring no access to main memory) 95% of the time.
Similarly, a 2 way associative 512 KByte L2 data cache will have the next
data needed almost 99.5% of the time (requiring no access to main memory.)

The result is that the difference in performance for a program that fits all
data and instructions into L1 and L2 caches is almost zero between a system
equipped with CAS 3 versus CAS 2 main memory.



You can check the effect of cache performance vs. memory performance using
the comparisons in SiSoft Sandra 2004.

Phil Weldon

"Spaceman" > wrote in message
...
| is CAS latency really that important? i see very expensive modules out
there
| with a cas 3 latency, and less expensive ones with a cas 2 latency. so i'm
a
| bit confused.
|
|

Bob Doran
March 14th 06, 05:43 AM
"rex roland" > said: "****ing idiot" in response to the
imbecilic posting:
CAS3 is slow, CAS2 is faster. You want as lower figure as possible.
Conor

Give him a break. Even ****ing idiots do something right once in a long
while (although I can't imagine what it could be)...

Ed Light
March 14th 06, 05:05 PM
"Bob Doran" > wrote in message
...
> "rex roland" > said: "****ing idiot" in response to the
> imbecilic posting:
> CAS3 is slow, CAS2 is faster. You want as lower figure as possible.
> Conor
>
> Give him a break. Even ****ing idiots do something right once in a long
> while (although I can't imagine what it could be)...

Pathetic ranting.

--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at

Thanks, robots.

Bring the Troops Home:
http://bringthemhomenow.org

Fight Spam:
http://bluesecurity.com

Conor
March 14th 06, 06:13 PM
In article >, rex roland
says...
> Conor wrote:
> > In article >, Spaceman
> > says...
> >> is CAS latency really that important?
> >
> > Absolutely.
> >
> >> i see very expensive modules out there
> >> with a cas 3 latency, and less expensive ones with a cas 2 latency. so i'm a
> >> bit confused.
> >>
> > CAS3 is slow, CAS2 is faster. You want as lower figure as possible.
> >
>
> ****ing idiot
>
Explain then clueless one.

--
Conor,

Same ****, different day.

Folk
March 14th 06, 06:36 PM
On Mon, 13 Mar 2006 01:39:24 -0800, "Ed Light" >
wrote:

>Yipes, I found another article that shows more differences.
>
>http://www.anandtech.com/memory/showdoc.aspx?i=2392&p=1

Here's another good article on the subject:

http://techreport.com/etc/2005q4/mem-latency/index.x?pg=1

Ed Light
March 15th 06, 03:52 AM
"Folk" > wrote
>
> Here's another good article on the subject:
>
> http://techreport.com/etc/2005q4/mem-latency/index.x?pg=1

Ah, that's the really clear one.


--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at

Thanks, robots.

Bring the Troops Home:
http://bringthemhomenow.org

Fight Spam:
http://bluesecurity.com

Inglo
March 15th 06, 05:18 AM
On 3/12/2006 9:21 PM Spaceman brightened our day with:
> is CAS latency really that important? i see very expensive modules out there
> with a cas 3 latency, and less expensive ones with a cas 2 latency. so i'm a
> bit confused.
>
>
>
Anyone have useful articles on the high latency, 4 and 5, modules that
are clocked really high. Like PC5300 memory. What's that really fast
stuff for? I'm looking at upgrading an old Athlon XP system to an
Athlon64, is that super fast memory just for Intel chips? I'd like to
buy some new RAM (2Gb, 2x1 Gb) that I can use for a couple of months on
my old motherboard before I go out and get a socket 939 board and a 64
bit CPU.
I'm looking at newegg and see 5 PC3200 CAS2 2Gb dual channel sets all
for about $160 (all requiring rebates), the CAS 2.5 and 3 RAM is only
marginally cheaper, though they don't require mail in rebates.
Is the faster clocked, higher latency memory just a gimmick? I
understand faster memory increases overclocking potential, but it
doesn't seem priced correctly for it's performance.

--
"I think it's much more interesting to live not knowing than to have
answers which might be wrong"

Richard P. Feynman

Steve ¤ģIngloĢ¤
www.inglostadt.com

Ed Light
March 15th 06, 07:36 AM
The fast memory lets you overclock without setting your memory slow (back to
166 or 133 so that it comes back up to near its normal 200 speed when you
raise the clock above 200). Instead it gets faster than 200, in synch with
the clock.

For those who want the very fastest system, not necessary for a "mainstream"
system.
--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at

Thanks, robots.

Bring the Troops Home:
http://bringthemhomenow.org

Fight Spam:
http://bluesecurity.com

Folk
March 15th 06, 06:14 PM
On Tue, 14 Mar 2006 18:52:45 -0800, "Ed Light" >
wrote:

>
>"Folk" > wrote
>>
>> Here's another good article on the subject:
>>
>> http://techreport.com/etc/2005q4/mem-latency/index.x?pg=1
>
>Ah, that's the really clear one.

As are most articles from Tech Report. It's my favorite enthusiast
site.