Legit Reviews

Common sense tell you that higher memory bandwidth should mean faster results, right? I set out to put this thought to the test looking at just two different memory dividers on my o/c'ed Q6600 system. At a FSB of 333 MHz, the slowest and fastest dividers I could run are:

1:1 a.k.a. PC5300 (667 MHz)
3:5 a.k.a. PC8888 (1,111 MHz)



Just for reference, as they relate to DDR2 memory: The highest divider is 1:2 aka PC10600 (1,333 MHz) and it just wasn't stable with my hardware @ 333 MHz.

All other BIOS settings were held constant:
FSB = 333.34 MHz and multiplier = 9.0 which gives an overall core rate of 3.0 GHz.
DRAM voltage was 2.25V and timings were 5-5-5-15-4-30-10-10-10-11.

You can think of memory bandwidth as the diameter (size) of your memory's pipe. Quite often, the pipe's diameter isn't the bottle neck for a modern Intel-based system; it is usually much larger than the information flow to/from the processor. Think of it this way, if you can only flush your toilet twice per minute, it doesn't matter if the drain pipe connecting your home to the sewer is 3 inches around, or 8 inches around, or 18 inches around: the rate limiting step in removing water from your home is the toilet flushing/recycling and the pull of gravity, not the size of your drain line. The same is true for memory bandwidth.

After seeing the data I generated on a quad core @ 3.0 GHz, I concluded that this toilet analogy is pretty true: the higher memory bandwidth gave more or less no appreciable difference for real world applications. Shocked? I was.

Further, I should point out that in order for my system to run stable in PC8888 mode @ a FSB of 333, I had to boost my NB vcore two notches and raise my ICH to the max (both of which the BIOS colored red meaning "high risk.") The increased voltage means more heat production, and greater power consumption -- not worth it for small gains realized in my opinion. Anyway, the test details and results are below if you want to read on.



Relevant test hardware:

Motherboard: Asus P5B-Deluxe (BIOS 1215)
CPU: Intel C2Q - Q6600 (B3 revision)
Memory: Ballistix DDR2-1066 (PC2-8500)

"Real-World" Application Based Tests

I chose the following apps: lameenc, x264, winrar, and the trial version of Photohop CS3. I ran these tests on a freshly installed Windows XP Pro SP2 machine.

Lame version 3.97 – Encoded the same test file (about 60 MB wav) with these commandline options: (which is equivalent to the old –-alt-preset fast standard) a total of 8 times and averaged play/CPU data as the benchmark.

x264 version 0.55.663 – Ran a 2-pass encode on the same MPEG-2 (720x480 DVD source) file 5 times totally and averaged the results. Without getting into too much detail, the benchmark is 1,749 frames @ 23 fps. Based on these numbers, I reported the time it would take to encode 215,784 frames (which is your average 2.5 h of video @ 23 fps). Why did I do this? The differences of just 1,749 frames were too insignificant.

Shameless promotion --> you can read more about the x264 Benchmark at this URL which contains results for hundreds of systems. You can also download the benchmark and test your own machine.

RAR version 3.62 – rar.exe ran my standard backup batch file which generated about 1.09 G of rars (1,654 files totally). Here is the commandline used: where list.txt a list of all the dirs I want it to back up. Benchmark results are an average of two runs timed with a stopwatch.

Trial of Photoshop CS3 – The batch function in PSCS3 was used to do three things to a total of twenty-nine, 10.1 MP jpeg files:

1) bicubic resize 10.1 MP to 2.2 MP (3872x2592 --> 1800x1200) which is the perfect size for a 4x6 print @ 300 dpi.
2) unsharpen mask filter (60 %, 0.8 px radius, threshold 12)
3) saved the resulting files as a quality 8 jpg.

Benchmark results are an average of two runs timed with a stopwatch.

"Synthetic" Application Based Tests

Just two of these were chosen to illustrate a point about theoretical gains vs. real world gains. Actually, I did SuperPI for the hell of it. WinRAR served to illustrate that point.

SuperPI / mod1.5 XS – The 16M test was run twice, and the average of the two are the benchmark.

WinRAR version 3.62 – If you hit alt-B in WinRAR, it'll run a synthetic benchmark. This was run twice (stopped after 100 MB) and is the average of two runs.

Raw Data - "Real-World" Apps
Lameenc play/cpu (average 8 runs) @ PC5300: 30.7935
Lameenc play/cpu (average 8 runs) @ PC8888: 30.8045
Result: PC8888 is 0.5 % faster

x264 time to encode 2.5 h DVD @ PC5300: 01:48:54
x264 time to encode 2.5 h DVD @ PC8888: 01:46:14
Result: PC8888 is 2.5 % faster

rar.exe back-up (average 2 runs) @ PC5300: 45 sec
rar.exe back-up (average 2 runs) @ PC8888: 44 sec
Result: PC8888 is 2.2 % faster

Photoshop CS3 Trial batch (average 2 runs) @ PC5300: 33 sec
Photoshop CS3 Trial batch (average 2 runs) @ PC8888: 33 sec
Result: PC8888 is 0.0 % faster

So stop right here and ask yourself if a 2-3 % gain is worth the higher voltage and heat.

Raw Data - "Synthetic" Apps

SuperPI/16M test (average 2 runs) @ PC5300: 8 m 8.546 s
SuperPI/16M test (average 2 runs) @ PC8888: 7 m 33.328 s
Result: PC8888 is 7.8 % faster

Winrar internal benchmark (average 2 runs) @ PC5300: 1,515 KB/s
Winrar internal benchmark (average 2 runs) @ PC8888: 2,079 KB/s
Result: PC8888 is 37.2 % faster

...but who uses their system exclusively running internal and synthetic benchmarks? Recall that for my 1.09 gig back up, I only gained about 2 % doing "real work" by using the higher divider. Hardrives are notorious bottle-necks in systems that serve to nullify any memory bandwidth increases. In this case the 37 % theoretical increase was translated into only a 2 % "real world" increase likely due to the hardrive/rar's ability to read/write the data. Again, this seems kinda wasteful to me.

I will admit that there might be special cases where running at high memory dividers may produce more substantial gains: apps such as folding@home or seti@home, etc. may benefit from the higher memory bandwidth since they tend to make exclusive use of the system memory bandwidth and rely much less on the hardrive. I have no data to back-up this though. Also lacking in my experiments are any game data. I'd be interested in knowing if the higher bandwidth can be leveraged by game engines such as UT3, Crysis, etc. but I also didn't look at these here.

Finally, since I held everything else constant, I didn't look at the tighter timings in 1:1 mode that people can often use which may give additional gains. For example, I can get away with 3-3-3-9 @ 1:1 vs. the slower 5-5-5-15 @ 3:5 with this memory.

Anyway, I hope you found this useful and maybe this will inspire someone else to look at the gaps pointed out above (and the gaps I haven't thought of too!)

graysky wrote: Raw Data - "Synthetic" Apps

SuperPI/16M test (average 2 runs) @ PC4300: 7 m 33.328 s
SuperPI/16M test (average 2 runs) @ PC7100: 8 m 8.546 s
Result: PC7100 is 7.8 % faster

Isn't PC4300 faster then as lower in seconds means better performance.

Ah crap... that was just a mix-up on my part. I'll correct it after dinner. Good eye

You just made another Sticky

Now let's see more tests from you and others!

Wow, i missed this first time round.

Thanks for another great insight into benchmarks that we'd all love to test out Graysky. ^_^

Dan

Here is a very good post on this subject:
http://www.thetechrepository.com/showthread.php?t=160

Thanks for the post dude. Megadeth rules by the way (their stuff from 1985-1997 anyway).

Peace doesn't sell...nobody's buying! Yes sir, Megadeth rules. Symphony of destruction is killer!

Funny, my fraternity pledge name is Vic, straight from Megadeth

This is a good post but it is really hard to say that as its posted would be grounds to purchase one or the other, maybe I am missing your point of that post.
I mean the CPU FSB will always play a role in memory performance and running these tests with the CPU at a specific setting to show the difference is like apple's to oranges.
IE running the CPU at a higher frequency which you would have to do with many chipsets to get the memory frequency at 1111MHz will make it another test and not really showing the true difference.
In addition, the platform you are using may or may not have a bearing with the results. It is interesting and I am sure quite accurate just I don’t know if it’s relevant to anything.

I think its best to keep an open mind with memory and use the memory speed to get the best over all performance of any system.
IE the memory frequency you choose should always be based on the CPU FSB and the MB chipset used and if you plan on over clocking the system or not. You want the system to run at its peak with little or no bottle necks in the different system Buss's.
I am not trying to be rude just add maybe a different out look

JohntechUPi wrote:This is a good post but it is really hard to say that as its posted would be grounds to purchase one or the other, maybe I am missing your point of that post.
I mean the CPU FSB will always play a role in memory performance and running these tests with the CPU at a specific setting to show the difference is like apple's to oranges.
IE running the CPU at a higher frequency which you would have to do with many chipsets to get the memory frequency at 1111MHz will make it another test and not really showing the true difference.
In addition, the platform you are using may or may not have a bearing with the results. It is interesting and I am sure quite accurate just I don’t know if it’s relevant to anything.

I think its best to keep an open mind with memory and use the memory speed to get the best over all performance of any system.
IE the memory frequency you choose should always be based on the CPU FSB and the MB chipset used and if you plan on over clocking the system or not. You want the system to run at its peak with little or no bottle necks in the different system Buss's.
I am not trying to be rude just add maybe a different out look

He kept the same 333MHz FSB, he just changed the RAM divider from 1:1 to 3:5

JohntechUPi wrote:He kept the same 333MHz FSB, he just changed the RAM divider from 1:1 to 3:5

I see, so as I said the results are great to know but how useful is it?
Faster memory will always show better results we knew that at least I hope we all knew that.

Well at any rate it was a nice experiment and I do not want to make it sound like its not appreciated because it was; the results take time and effort to compile. I am just saying that before you use these results you would need to put them into context and how it would relate to the system one might plan to build.
So to better put this in context let me put it this way.
Based on this result posted by graysky (great work by the way) and how this would relate to current chipsets on the market.
If you have a non Intel chipset MB that will support ratios on the memory frequency and a CPU that can help to take advantage of the added memory performance then get the fastest memory you can afford. And if it is an Intel Chipset as most if not all will not support Ratios with greater number for the memory and be stable get the best latency memory and a 1-1/2-1 Speed for the CPU you plan to get. Also some chipsets will perform better than other s and have better memory functionality. Some testing to show this would be really helpful in my opinion, what say you o forum gods/Buddha's of legit reviews??? LOL

You called?
(lol, I'm not that big-headded really )
What?!? No I'm not!

What results we have here showed that there is very little real world performance gain from getting the fastest memory you can.

I think the best compromise is to only buy as fast RAM as you can when you plan to overclock (so that you are not held back by the memory). Otherwise DDR2 800Mhz works fine (I think I have something psychological about not wanting 667Mhz ram , a friend has 4Gig of the stuff combined with a Q9450 and 8800GTX, and he is running along great on it).

Dan

Yup. I'll have to show this chart to a friend of mine...He swears his 1066MHz Cosair Dominators are faster than my 800MHz OCZ Reapers.

If I recall correctly, there was an X-Bit Article on memory. However, I think the comparison was DDR2/DDR3 with differing speeds/latencies. However, the improvment was minimal.

One quick question: I've always read that a 1:1 ratio gives you higher bandwidth than, say, a 3:5 divider. Could that affect your results?

~Ibrahim~

EDIT: Found it! http://ixbtlabs.com/articles3/cpu/ddr2- ... -1333.html

Here is why the benchmarks didn't show any gains:
FSB became a bottleneck.

First, note that Intel LGA775 chipsets sit between the CPU and the RAM. So any data going from the RAM to CPU and vice-versa must first go through the FSB.

Bandwidth between CPU and chipset @ 333MHz FSB (QDR-1333):
FSB bus width for LGA775 platforms is 64bit. Thus, bandwidth between CPU and chipset is:
1333MHz × 64bit = 85312Mbit/s = 10.7GB/s

Bandwidth between RAM and chipset @ dual channel DDR2-667:
667MHz × 128bit = 85376Mbit/s = 10.7GB/s

Uh-oh...
FSB1333 is already saturated by DDR667...

Bandwidth between RAM and chipset @ dual channel DDR2-1111:
667MHz × 128bit = 142208Mbit/s = 17.8GB/s

Bandwidth of FSB333 << bandwidth of DDR2-1111. Meaning, the excess bandwidth DDR2-1111 offers never actually reaches the CPU and any gains are more due tighter latencies at DDR2-1111 in your setup.
Wonders of Intel LGA775 platform...


edit:
And I just realized after posting this that the thread is positively ancient.

Thanks for the insight though

Dan

here some memory bandwidth comparisons:
http://www.maxxpi.net/pages/result-brow ... lusive.php

best

good work. pretty interesting results.

With the new Sandybridge platform all of the above is untrue. Here the higher the rated speed - the higher is the bandwidth. Even the timing has not much significance anymore. With 2133 memory I get 28GB/s in sandra, while 1600mhz would give me about 20-21GB/s, 1333 about 17GB/s. So you need to mark all of the above being related to core and earlier platforms.