New Posts  All Forums:Forum Nav:

DDR -> DDRII -> DDRIII - Page 4

post #61 of 77
4/4/05 at 3:34pm
Thread Starter 
Here is a list of all the DDR2 brand and their latency setting

The following DDR2 533/400 SODIMMs have passed system-level validation on Mobile Intel® 915GM/GMS, or 910GML Express chipset family reference motherboards.

http://developer.intel.com/technolog...mm_results.htm

Quote:
Listed below are the results of testing a small sample of DDR2 533/400 SODIMM modules on Intel® reference systems. We are providing this information as a guide to module performance. This testing is not intended to replace the normal OEM module qualification process. For results on specific Intel® or OEM production motherboards, please refer to the OEM's list of qualified memory suppliers.
Reply
Reply
post #62 of 77
4/4/05 at 3:51pm
Thread Starter 
http://www.corsairmicro.com/corsair/...h/trg-cas.html

CAS Latency: What Is It, and How Does It Impact Performance?
Quote:
First of all, what is CAS?

"CAS" is short for "Column Address Strobe". A DRAM memory can be thought of as a matrix, kind of like a spreadsheet with memory cells instead of numbers and formulas. Like the spreadsheet, each cell has a row address and a column address (like "AA57" or "R23C34" in the spreadsheet). As you might have guessed, there is also a RAS signal, which is shorthand for "Row Address Strobe".

And, what do you mean by "latency"?

Latency refers to the time that you are waiting to get what you need. Merriam-Webster dictionary defines it as "the interval between stimulus and response".

Now, how does CAS work?

To understand this let's walk through a simplified version of how the memory controller actually reads the memory. First, the chip set accesses the ROW of the memory matrix by putting an address on the memory's address pins and activating the RAS signal. Then, we have to wait a few clock cycles (known as RAS-to-CAS Delay). Then, the column address is put on the address pins, and the CAS signal is activated, to access the correct COLUMN of the memory matrix. Then, we wait a few clock cycles -- THIS IS KNOWN AS CAS LATENCY! -- and then the data appears on the pins of the RAM.

So, for CAS-2 you wait 2 clock cycles and for CAS-3 you wait 3 clock cycles?

Bingo!

So, CAS-2 is 33% faster than CAS-3?

Whoa, not so fast! There are a LOT of other factors in the memory performance. Here are a few of the main ones:

Sometimes you have to move to a different row in memory. This means activating RAS, waiting RAS-to-CAS delay, then doing the CAS latency thing.
Other times, you do a "burst" read, when you pull in a lot of data in one big block. In that case, CAS is only activated ONCE, at the beginning of the burst.
Also, don't forget the most important thing: processors have big caches! The cache is where the processor stores recently accessed instructions and data. The cache "hit rate", i.e., the percentage of times the processor finds the information it needs in its own cache, is typically greater than 95%!
OK, OK, so what's the bottom line?

So, the bottom line is, moving from CAS-3 to CAS-2 will offer a percentage performance increase in the low single digits for most applications. Programs which are known to be memory intensive (you gamers might know of some...) will see the best improvement.

The other thing to keep in mind is that CAS-2 memory will run FASTER ( some review sites have taken it to 160MHz!) than CAS-3 memory. So, if you're thinking of overclocking your system (now or in the future), CAS-2 is your best bet for speed and stability.

So, the Ram Guy sez...

Buy CAS-2 if [1] you want to wring the last bit of performance out of your system, or [2] you're thinking of overclocking, either now or in the future, or [3] it costs the same as CAS-3, which it sometimes does...

Otherwise, CAS-3 memory should meet your requirements
Reply
Reply
post #63 of 77
4/4/05 at 4:48pm
Quote:
Originally Posted by smilepak
http://www.corsairmicro.com/corsair/...h/trg-cas.html

CAS Latency: What Is It, and How Does It Impact Performance?
Here is an interesting article that might shed light into the pc3200 pc4200 performance...

clicky
Reply
Reply
post #64 of 77
4/4/05 at 6:18pm
Thread Starter 
Reply
Reply
post #65 of 77
4/4/05 at 6:22pm
Thread Starter 
http://www.pcstats.com/articleview.c...id=1573&page=4

Quote:
What about Latency issues?

Latency is one area where the DDR2 specs fall short of regular DDR memory, at least on paper. The JEDEC definition of DDR2 allows for CAS latencies of 3,4 or 5, as compared to DDR's 1.5,2 and 2.5. While some manufacturers have released 400MHz DDR2 at CAS latency 3, all 533MHz DDR2 chips currently shipping use a Latency timing of 4-4-4. As 533MHz DDR2 is likely to be the first widely used speed for the new standard, these are the timings that will be the default standard.

Write latency is also considerably greater with DDR2. While DDR1 allows a single cycle for write latency (1T), DDR2 defines write latency as read latency-1. This means that with a CAS latency of 4, DDR2-533 should have an effective write latency of 3T. Again this seems like bad news on paper.

To be fair, it may be difficult to compare DDR2 to DDR memory directly in terms of latency and performance. The various new features that DDR2 implements may make the difference in memory latency less significant. For more information, check out PCstats' great article on the benefits of low memory latency versus memory speed with respect to DDR, and for a (much) more technical examination of the latency issues of DDR2 memory, look here.

As it stands, it seems unlikely that DDR2 memory will have any performance advantage over DDR of an equivalent speed. The slower speed of the DDR2 core is cancelled out by its ability to move more data each clock cycle. Manufacturers claim that the various new signal enhancement features that DDR2 possesses will add to effective bandwidth by reducing interference, but then DDR2 takes more time to perform data actions (higher latency). Only proper lab tests will decide this. Stay tuned.
Reply
Reply
post #66 of 77
4/4/05 at 6:23pm
It's too bad that the Sonoma doesn't support DDR-II @ 667 MHz
Reply
Reply
post #67 of 77
4/4/05 at 6:30pm
Thread Starter 
Quote:
Originally Posted by MichaelX30
It's too bad that the Sonoma doesn't support DDR-II @ 667 MHz
Sonoma 2.0? LoL!
Reply
Reply
post #68 of 77
4/6/05 at 2:25pm
If I get DDR2 533 MHz RAM with CL4 timing, can that RAM also run at 400 MHz but at faster CL3? I read somewhere that when you use slower frequency (533 to 400), that conversely the timing is automatically sped up (CL4 to CL3).
Reply
Reply
post #69 of 77
4/6/05 at 2:54pm
Thread Starter 
Quote:
Originally Posted by bfavre444
If I get DDR2 533 MHz RAM with CL4 timing, can that RAM also run at 400 MHz but at faster CL3? I read somewhere that when you use slower frequency (533 to 400), that conversely the timing is automatically sped up (CL4 to CL3).
Not sure about notebook, but on my desktop I was able to adjust the latency around on the memory through the BIOS. I don't believe that feature is available in the notebook BIOS.

So in theory, at least on the desktop end yes you could do that and tweak your memory around to get an optimal setting for it.
Reply
Reply
post #70 of 77
4/7/05 at 1:56pm
2 (yes, that's two) gig notebook memory stick...

http://www.tomshardware.com/hardnews...06_135503.html
Quote:
Infineon intros 2 GByte notebook memory, hints to 1 GByte graphic cards
By Wolfgang Gruener, Senior Editor
April 6, 2005 - 13:55 EST

Munich - Infineon announced several new high-density memory products for tight spaces in notebooks and graphics systems - including a 2 GByte dual-die SO-DIMM and 512 Mbit GDDR3 devices for graphic cards with up to 1 GByte of memory.

Based on the technology of the 4 GByte desktop memory module launched in February, Infineon today said it will extend its dual-die product line and released samples of a high-density device for notebooks.

The 2 GByte DDR2 dual-die SO-DIMM target high-end notebooks and is manufactured with 8 dual-die 2 Gbit DDR2 components. The result is a 1.8 volt part with a density of 2 GByte and a thickness of 3.8 mm at the standard 30 mm height, Infineon said. The basic principle of dual-die memory is based on the concept of stacking two identical dies within one BGA package.

Due to the limited sample production, pricing of the modules is far from being affordable for the consumer at this time. Infineon currently lists a price of $1700 for each sample.

The company also launched new memory for performance graphics cards. The 512Mbit GDDR3 chips are built in a 16Mbit x 32 structure and are clocked 800MHz, enabling data bandwidths of up to 51.2 Gbit per second and memory. The devices allow graphics card manufacturer to build their products with frame buffer sizes of 512 MByte or even 1 GByte (256 MByte in notebooks), according to Infineon. Volume production of the memory is scheduled for the second half of this year.

Another new product is a 1 GByte DDR2 Micro-DIMM module that could the manufacturing of lighter and smaller sub-notebooks and enhance battery lifetime. Infineon said the modules are about 65 percent smaller than common SO-DIMMs, consume about 50 percent less power, but are compliant to standards set by the Joint Electron Device Council (JEDEC). Samples of the modules have been shipped in PC2-3200 and PC2-4300 speeds to "select customers", according to the company.
Reply
Reply
post #71 of 77
4/7/05 at 2:00pm
Thread Starter 
Hmm what system could use beyond 2GB?
Reply
Reply
post #72 of 77
4/7/05 at 3:07pm
Any current laptop could use up to 4GB. A 64-bit machine would be able to use >4GB.
I run out of RAM sometimes while working with my 2GB space, wouldn't mind having 4GB every now and then.
Reply
Reply
post #73 of 77
4/28/05 at 5:29pm
Thread Starter 
sending this back up
Reply
Reply
post #74 of 77
5/5/05 at 7:16pm
Thread Starter 
sending this back up
Reply
Reply
post #75 of 77
5/6/05 at 12:40am
Seems nobody has anything new to say here...
Reply
Reply
post #76 of 77
5/11/05 at 2:01pm
Thread Starter 
For those who are interested in seeing the differences between Single Channel DDRII vs Dual Channel DDRII

http://www.notebookdiscussions.com/showthread.php?t=239
Reply
Reply
post #77 of 77
5/11/05 at 7:49pm
I would like to point out that there is no large hardware site doing a through comparison between DDR and DDR2. Games don't test memory performance very well and neither do synthetic benchmarks. What we need is some real-world applications that happen to be very latency sensitive for a true comparison of latencies. Also DDR2-400 is actually running at 100MHz compared to DDR-400's 200MHz. So ddr2 is not doing as bad as people make it out to be. (tangent-if you want to be technical, ddr2 also has the advantage of having twice the bus frequency than ddr at a given core frequency).

So DDR2-800 would probably be a more apt comparison to DDR-400, at least that's my opinion.

BTW, DDR3 doesn't exist as a standard. GDDR-3 is just a modification of DDR2 for graphics cards.
Reply
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Notebook Forums - General