Am I going to notice the difference?

R

r3dh3adkid

New Member
I plan to upgrade from my current memory (2x512 3200) to 4x512 3200 and just get two more of the same sticks that i already have ( http://www.newegg.com/Product/Product.asp?Item=N82E16820145450 )


My question is would i notice a better preformance jump if i went from 4x512 to 2x1gig? I know the 2x1 gig is better but would i actually notice it that much?

Upgrading to the 2x1 gig costs more than getting 2 more 512 sticks even if i sell the current two sticks that i have.
 
You won't notice any difference at all between 4x512 and 2x1GB. All you will do is free up two slots with the latter, which is never a bad idea.
 
In a number of ways especially when running some of the newer games that just "LOVE MEMORY" you will often see an immediate difference even if it is only a small one there at first. The problem often seen with smalled muliple dimms is what is termed "memory holes" where once the first dimm is filled a gap is seen when data jumps over to the next. I went from a matched pair of 512mb performance dimms to a pair of 1gb value memory and have seen a new machine in the process in more then one way(my old board crapped). But with 2gb I don't see any stalls or locks like what was seen with the 512s.
 
PC eye said:
In a number of ways especially when running some of the newer games that just "LOVE MEMORY" you will often see an immediate difference even if it is only a small one there at first. The problem often seen with smalled muliple dimms is what is termed "memory holes" where once the first dimm is filled a gap is seen when data jumps over to the next. I went from a matched pair of 512mb performance dimms to a pair of 1gb value memory and have seen a new machine in the process in more then one way(my old board crapped). But with 2gb I don't see any stalls or locks like what was seen with the 512s.
Click to expand...

Is it really worth it to spend an extra 114 bucks so that it wont lag a little?
 
Anandtech did a test comparing two DIMM and four DIMM combinations. He tried mismatched and matched pairs. Now don't get me wrong, I don't like Anand one bit, but he did all the homework so I don't have to.

Here are the benchmarks:

http://www.anandtech.com/memory/showdoc.aspx?i=1839&p=4

You will notice the 4 DS modules perform slightly better in the benchmarks. The difference, however, will not be noticeable in every day usage.
 
You'll note that in two out of the three tests there the performance gain was actually seen with the two dimm setup over the four dimm. Anytime you have an increase in memory by simply adding to the existing or swapping one size dimm or dimms out for larger ones there will be a gain in some areas. But at the time the lower amount of larger dimms will see an overall gain in the long tun when you start running different types as well as different apps.
 
Actually I am afraid that is incorrect. Look at the results again. The Sandra test, not the Memtest. They are very, very close with 4 DIMMs taking a small lead.

That small lead will not be noticed in every day use. Trust me. I have played with Sandra, timings, etc for hours and hours. Sometimes I get really damn bored.

I don't know where you got that silly memory hole idea from.
 
You won't notice any difference. 4 dimms might be slightly faster, but you will not notice it. Either is fine, but if you get 2 dimms, that frees up two more slots so you can upgrade later.
 
SirKenin said:
Actually I am afraid that is incorrect. Look at the results again. The Sandra test, not the Memtest. They are very, very close with 4 DIMMs taking a small lead.

That small lead will not be noticed in every day use. Trust me. I have played with Sandra, timings, etc for hours and hours. Sometimes I get really damn bored.

I don't know where you got that silly memory hole idea from.
Click to expand...

"November 2003
Avoiding Memory Fragmentation


Be careful when allocating and de-allocating memory so that you are not creating unnecessary memory fragmentation by leaving many small memory holes that are not usable. These small holes are free memory and will be counted as such, i.e. calling the function Runtime.getRuntime().freeMemory() will return the amount of all free memory including the holes. However, this memory is unusable since the holes would be too small to hold the data objects."
http://developer.sonyericsson.com/s...strickscode/java/p_avoidingmemoryfragment.jsp

"The regions within the STT that contain entries for addresses within unsectored regions are never referenced. Not to be wasted, these “holes” within the STT are made available as additional sectored storage through incorporation into the free list."

http://www.research.ibm.com/journal/rd/452/tremaine.html
 
That is a myth PC eye. A well known myth as a matter of fact. If you had done a quick search on Google before posting that you would have found that out. Windows XP has very effective memory management. The companies that tell you otherwise are trying to sell you something you don't need. I am well aware that memory holes exist, but they have absolutely no effect on performance like you falsely asserted, and that is what I was trying to get at when I said that. In fact, if you optimize your RAM you will degrade performance significantly. You need to do more research before you start flinging links around and unrelated concepts to the OP.

For example, here is just one result at Google:

http://www.4peeps.com/ivb/lofiversion/index.php/t20862.html

Reality - "RAM Optimizers have no effect, and at worst, they seriously degrade performance. Although gaining more available memory might seem beneficial, it isn't. As RAM Optimizers force the available-memory counter up, they force other processes' data and code out of memory. Say that you're running Word, for example. As the optimizer forces the available-memory counter up, the text of open documents and the program code that was part of Word's working set before the optimization (and was therefore present in physical memory) must be reread from disk as you continue to edit your document. The act of allocating, then freeing a large amount of virtual memory might, as a conceivable side effect, lead to blocks of contiguous available memory. However, because virtual memory masks the layout of physical memory from processes, processes can't directly benefit from having virtual memory backed by contiguous physical memory. As processes execute and undergo working-set trimming and growth, their virtual-memory-to-physical-memory mappings will become fragmented despite the availability of contiguous memory." - Source
Click to expand...

http://www.overclock.net/faqs/88781-opinion-does-increasing-amount-available-ram.html

RAM Optimizers don't have any effect on performance, at worst, they can and do cause performance to degrade seriously. It seems like gaining more available RAM is beneficial, it's not. How RAM optimizers get you more available ram is by dumping important and often used processes out. Lets say that you are running a word processor, as an example. When the optimizer dumps the physical memory, the open document's text and the program code that was part of the word processor's working set(in RAM) before the dump. It must be read again from the relatively slow hard disk as you continue to type up your document.
Click to expand...

http://www.techenclave.com/forums/windows-xp-tweaks-that-doesnt-work-6519.html

Memory Optimizers

They claim "Increasing the amount of available RAM improves performance". However, the only thing that these programs can do is harm real system performance.

Although gaining more available memory might seem beneficial, it isn't. As RAM Optimizers force the available-memory counter up, they force other processes' data and code out of memory. However, because virtual memory masks the layout of physical memory from processes, processes can't directly benefit from having virtual memory backed by contiguous physical memory. As processes execute and undergo working-set trimming and growth, their virtual-memory-to-physical-memory mappings will become fragmented despite the availability of contiguous memory.
Click to expand...

http://www.microsoft.com/whdc/driver/kernel/xp_kernel.mspx#ECLAC

Memory Management
Windows XP provides improved memory management. The memory manager provides the system services to allocate and free virtual memory, share memory between processes, map files into memory, flush virtual pages to disk, retrieve information about a range of virtual pages, change the protection of virtual pages, and lock the virtual pages into memory. The memory manager also provides a number of services, such as allocating and de-allocating physical memory and locking pages in physical memory for DMA transfers, to other kernel-mode components inside the executive as well as to device drivers.

Memory management enhancements include the following:

• Logical prefetcher for faster boot and application launch

• Enhanced memory management for better scalability

• Reduced paged pool usage

• Increased number of system Page Table Entries (PTEs)

• Support of giant drivers
Click to expand...



Memory holes are clearly not an issue here and you should have never confused the issue by even bringing it up in the first place, IMO. You will never see an impact on performance from memory holes. There will be no more or no less memory holes with two sticks verses four. The only improvements you saw were either in your head or as a result of getting rid of a crappy memory controller on the old motherboard.
 
SirKenin said:
That is a myth PC eye. A well known myth as a matter of fact. If you had done a quick search on Google before posting that you would have found that out. Windows XP has very effective memory management. The companies that tell you otherwise are trying to sell you something you don't need. I am well aware that memory holes exist, but they have absolutely no effect on performance like you falsely asserted, and that is what I was trying to get at when I said that. In fact, if you optimize your RAM you will degrade performance significantly. You need to do more research before you start flinging links around and unrelated concepts to the OP.

For example, here is just one result at Google:

http://www.4peeps.com/ivb/lofiversion/index.php/t20862.html



http://www.overclock.net/faqs/88781-opinion-does-increasing-amount-available-ram.html



http://www.techenclave.com/forums/windows-xp-tweaks-that-doesnt-work-6519.html



http://www.microsoft.com/whdc/driver/kernel/xp_kernel.mspx#ECLAC





Memory holes are clearly not an issue here and you should have never confused the issue by even bringing it up in the first place, IMO. You will never see an impact on performance from memory holes. There will be no more or no less memory holes with two sticks verses four. The only improvements you saw were either in your head or as a result of getting rid of a crappy memory controller on the old motherboard.
Click to expand...

Performance
This should be reasonably easy to determine, data flow at 800MHz should be faster than data flow at 266MHz, right? Well, it isn't all that simple. Because EDO DRAM, SDRAM, and RDRAM are based on the same core-memory technology, their internal device timings are nearly identical. Thus, the differences among memory subsystems that affect the latency include the rate at which the system can move the address and control information to the DRAM and the rate at which the DRAM can move data from the DRAM to the memory controller. We know what you're thinking .. huh? The simplest and easiest explanation is that the speed is entirely dependent upon how the systems, consisting of the motherboard components such as the memory controller etcetera, are designed and constructed.
In a DDR or SDRAM system, each DIMM is connected, individually and in parallel, to the data bus. So whether you have a single DIMM or multiple DIMMs, the amount of time it takes to initiate a data transfer is effectively unchanged.
In a Rambus system, RIMM modules are connected to the bus in a series. The first data item transferred must pass through each RIMM module (and/or CRIMM module) before it reaches the bus. This makes for a much longer distance for the signal to travel, but remember it is also traveling 3+ times faster. Again, which memory you choose is entirely dependent upon how you intend to use the system. All too often people purchaser computers without first determining whether their choice will meet their needs. http://www.dewassoc.com/performance/memory/ddr_sdram.htm
 
PC eye said:
Performance
This should be reasonably easy to determine, data flow at 800MHz should be faster than data flow at 266MHz, right? Well, it isn't all that simple. Because EDO DRAM, SDRAM, and RDRAM are based on the same core-memory technology, their internal device timings are nearly identical. Thus, the differences among memory subsystems that affect the latency include the rate at which the system can move the address and control information to the DRAM and the rate at which the DRAM can move data from the DRAM to the memory controller. We know what you're thinking .. huh? The simplest and easiest explanation is that the speed is entirely dependent upon how the systems, consisting of the motherboard components such as the memory controller etcetera, are designed and constructed.
In a DDR or SDRAM system, each DIMM is connected, individually and in parallel, to the data bus. So whether you have a single DIMM or multiple DIMMs, the amount of time it takes to initiate a data transfer is effectively unchanged.
In a Rambus system, RIMM modules are connected to the bus in a series. The first data item transferred must pass through each RIMM module (and/or CRIMM module) before it reaches the bus. This makes for a much longer distance for the signal to travel, but remember it is also traveling 3+ times faster. Again, which memory you choose is entirely dependent upon how you intend to use the system. All too often people purchaser computers without first determining whether their choice will meet their needs. http://www.dewassoc.com/performance/memory/ddr_sdram.htm
Click to expand...

Ummm.. Again, that is pretty much irrelevant, except it further goes to show that memory holes don't matter because the DIMMs are connected in parallel, not in series.
 
"How Much Do You Need?
It's been said that you can never have enough money, and the same holds true for RAM, especially if you do a lot of graphics-intensive work or gaming. Next to the CPU itself, RAM is the most important factor in computer performance. If you don't have enough, adding RAM can make more of a difference than getting a new CPU!
If your system responds slowly or accesses the hard drive constantly, then you need to add more RAM" http://computer.howstuffworks.com/ram2.htm

If you read the article in full that I would have loved to have added here once relocated you would understand that XP plus the board you run averages 196mb of ram reserved from a 512mb dimm and that once filled data was then extended to the next dimm produced memory "holes" as it was illustrated in the article seen last year. What that was describing was when an initial 196mb of the first of two 512mb dimms was used and you run a program that requires 512mb you subtract the 196 from the 512mb to end up with 316mb available on the first dimm. With a 1gb dimm the 196mb system usage and 512mb needed by the program are readily available. That alone sees an overall better performance.
 
Spectacular gaming
A new graphics card is nice, but it isn't enough for today's cutting edge gaming. Revolutionize your 3D gaming experience by adding more memory. Added RAM supports color at higher resolutions and gives you the power for enhanced 3D rendering, with amazing character realism and texture versatility. Enjoy crisper, brighter images and excitingly fast 3D video and animation for a powerful, virtually real gaming experience.
[FONT=Verdana, Arial, Helvetica, sans-serif]
Enhanced multimedia presentations
Intense graphics use and memory-laden programs, such as multimedia, publishing and graphics arts, demand more RAM for the optimal computing experience. Flash and ROM take up a lot of space. You're going to need at least 128MB, with 256MB or more needed for best performance. Once again, more is better. The applications respond much quicker with faster image drawing. http://www.computermemoryupgrade.net/why-more-ram-is-better.html

The thing you seem to miss is how the article illustrated what happens when XP and the system board chew an average of 196mb of ram on a system with two 512mb dimms when running a program needing 512mb. On the first dimm regarding at that time single channel boards only 316mb of ram was available for the program when that saw the first dimm filled with data. The remaining 196mb of the required was then taken from the next dimm. When two 1gb dimms were installed as the article described the system performance improved greatly. This can be said for having larger dimms on each channel on a newer dual channel board where larger apps are given priority on the primary while the background services and other apps are then pushed to the secondary channel according to the theory there.[/FONT]
 
lol :D Ummm. It doesn't matter if all the info is stored on DIMM 4. How can I stress enough what it means that the DIMMS are hooked up in PARALLEL? Your little pet theory that information is retrieved off the first DIMM faster would only be applicable if the DIMMS were hooked up in series, and the only DIMMS that are like that are RDRAM. Even then you are looking at times in the tiniest of fractions of a second.

Your "theory" is not applicable at all to DDR and the benchmarks prove you wrong.
 
First of all it "wasn't" my theory. It was a well written explaination with illustrations by a pro at technical writing. The article was seen sometime early in '05 well after XP with an improved memory management came out. In fact if I'm not mistaken SP2 was already out then. What you are getting confused on is that this factor doesn't effect every app and game. But outlined what can be seen when you have problems running a specific program. From having run muliple dimms of one size and then going with larger sized dimms there has been a noticable performance gain seen especially on a dual channel board.
 
Then you're going to have to provide your benchmarks to substantiate your claims, because it is very clear that benchmarks available on the internet completely annihilates them. Not to mention the theory you presented is absurd considering the circumstances, considering we are talking nano and micro seconds. No offence meant.
 
[ Memory Modules / A0 ]

Memory Module Properties:
Socket Designation A0
Type DIMM
Installed Size 1024 MB
Enabled Size 1024 MB

[ Memory Modules / A2 ]

Memory Module Properties:
Socket Designation A2
Type DIMM
Speed 50 ns
Installed Size 1024 MB
Enabled Size 1024 MB

[ Memory Modules / A3 ]

Memory Module Properties:
Socket Designation A3
Type DIMM
Speed 50 ns
Installed Size Not Installed
Enabled Size Not Installed

[ Memory Devices / A0 ]

Memory Device Properties:
Form Factor DIMM
Size 1024 MB
Speed 400 MHz
Total Width 64-bit
Data Width 64-bit
Device Locator A0
Bank Locator Bank0/1
Manufacturer None
Serial Number None
Asset Tag None
Part Number None

[ Memory Devices / A2 ]

Memory Device Properties:
Form Factor DIMM
Size 1024 MB
Speed 400 MHz
Total Width 64-bit
Data Width 64-bit
Device Locator A2
Bank Locator Bank4/5
Manufacturer None
Serial Number None
Asset Tag None
Part Number None
Memory Timings:
CAS Latency (CL) 3T
RAS To CAS Delay (tRCD) 3T
RAS Precharge (tRP) 3T
RAS Active Time (tRAS) 8T
Row Cycle Time (tRC) 11T
Row Refresh Cycle Time (tRFC) 14T
Command Rate (CR) 2T
RAS To RAS Delay (tRRD) 2T
Write Recovery Time (tWR) 3T
Read To Write Delay (tRTW) 5T
Write To Read Delay (tWTR) 2T
Write CAS Latency (tWCL) 1T
Refresh Period (tREF) 200 MHz 7.8 us
DQS Skew Control Disabled
DRAM Drive Strength Normal
DRAM Data Drive Strength 4 (No Reduction)
Max Async Latency 7 ns
Read Preamble Time 5.5 ns
Idle Cycle Limit 16
Dynamic Idle Cycle Counter Enabled
Read/Write Queue Bypass 8
Bypass Max 4
32-byte Granularity Disabled

seen with Everest Ultimate Edition trial version
 
Not that. Actual benchmarks. Get SiSoftSandra and bench all your different combinations that we have talked about and post your results here for examination.
 
I am all too familiar with SiSoftware's Sandra already. But when downloads and installs of both Sandra Lite 2005 and 2007 there is always one driver missing in it. The last one that worked here was the 2003. With that in mind the benchmarks results are PF=2706 as you can see at <a href="<A href="http://imageshack.us"><img">http://imageshack.us"><img src="http://img247.imageshack.us/img247/1397/emsaflexinfoqr8.jpg" border="0" alt="Image Hosted by ImageShack.us" /></a>
 
You must log in or register to reply here.
Back
Top