Source: Tom's Hardware US – Keywords: memory, scaling, ddr3
Categories: Overclocking, Intel, AMD/ATI
Syndication:
|
This article sums up and adds to three memory scaling articles that we've published in recent weeks: DDR3 Memory Scaling on AMD’s Phenom II X4 Core i7 Memory Scaling: From DDR3-800 to DDR3-1600 DDR3 Memory Scaling on Core 2 Quad In all three cases, we used the same benchmarks and cross-tested the systems with memory speeds from DDR3-800 to DDR3-1600, applying both relaxed and tight timings. The purpose was to take popular memory configurations and show their effective performance levels. In each case, we also added a benchmark run that shows the impact of increasing the processor clock speed instead of investing in high-end memory. In general, the benefit of expensive enthusiast memory is at least questionable. We found that tight timings are typically more beneficial than high memory clock speeds. But in the end, it’s important to look at the benchmarks that are relevant to you. This picture story compares the performance benefits of using faster DDR3 memory across all three popular platforms: AMD’s Phenom II X4 955 (3.2 GHz), Intel’s Core 2 Extreme QX9770 (3.2 GHz), and the Intel Core i7-975 (3.33 GHz), which is Intel’s current flagship solution for enthusiasts and professionals. We wanted to know how much each of these three architectures can benefit from faster memory. Which platform benefits more, and where might high-end memory be negligible? Let’s find out. The following results show DDR3-1600 performance with quick CL8 timings normalized to DDR3-1066, which equals 100%. Please note that the results list Intel's Core i7 at 3.47 GHz. We manually set the clock speed to this level to get consistent results, as Intel’s Turbo mode would automatically overclock individual cores to this speed under heavy processing loads. |
|
- 1 / 2
- Next
-
Best offers
|
16GB Cruzer USB Flash Drive | $79.99 STAPLES More info |
|
16GB Ultra II SDHC Card | $89.99 STAPLES More info |
|
512MB PC800 RDRAM RIMM Dual Channel... | $25.95 UpgradePCSolution.com More info |
|
16GB Memory Stick PRO Duo Card | $79.98 OneCall.com More info |
|
4GB Cruzer Titanium Plus USB Flash Drive | $35.00 MemorySuppliers More info |
- Upgrade Q6600 or Buy New i7 920?
- AMD Phenom II 940 "Xtremely" Benchmarked
- Why the scaling of my core i7 920 is so poor, even worse than Q6600?
- Nehalem memory limitiations?!?!
- Hi all, need some advice about CPU not very expensive for gaming
- my new mobo o/c's great
- 1.5k-1.6k Gaming Build
- ~$3,000 Mid- to High-Range Gaming System - Help, Suggestions, Comments
- 3K Budget for new rig, need suggestions!
- Build A Computer
- memory with different timing
- i7 versus quad core
- core 2 extreme qx9770
- intel core i7 versus core 2 quad
- core™ 2 extreme qx9770
- turbo mode core i7 how to
- what memory for phenom ii
- best ddr3 memory for i7
- what ram for core i7
- graphics card benchmarks 2009
- core 2 quad with ddr3
- apple core i7
- why is some memory for core i7
- what ddr3 for core i7
- processor benchmarks 2009
im sorry, but are these charts for real. 103.7 (min:sec) for a divx conversion?
114.9 (min:sec) to compress using winrar? what is this?
Hmmm a QX9770 is $1,400, an i7 975 is $999 and an PhenomII 955 is $215, just in case anyone was wondering about the cost differences of the tested processors.
hmmm how about DDR2 to DDR3 difference on the same phenom II processor ? one in AM2+ and one in AM3 MB ? they can even use the same 790GX/FX/SB750 chipset for easier comparison.
indeed nice review, good point also
im sorry, but are these charts for real. 103.7 (min:sec) for a divx conversion?114.9 (min:sec) to compress using winrar? what is this?
I am confused as well but I think the 114.9 is a percent improvement over DDR3 1066. The article stated that this meant a 14.9 percent increase. So for example, if your compression time was 5 secs, the faster memory would drop it to about 4.25 seconds (someone please check me on this). I agree the having the time on the Y-Axis is confusing. Unless I am wrong in my assumption. Author, help here?
Edit: I meant X-Axis
those numbers are the performance increase not the actual times / fps / score to the benchmarks.
"Please note that the results list Intel's Core i7 at 3.47 GHz. We manually set the clock speed to this level to get consistent results, as Intel’s Turbo mode would automatically overclock individual cores to this speed under heavy processing loads.Please note that the results list Intel's Core i7 at 3.47 GHz. We manually set the clock speed to this level to get consistent results, as Intel’s Turbo mode would automatically overclock individual cores to this speed under heavy processing loads."
You can turn it off in the BIOS.
It would be interesting to see the exact same charts for the same benchmarks and processors, but comparing CAS9 vs. CAS7. All can be DDR3-1333; only vary the CAS.
I guess no one can read charts too well, for example, the cost is not a factor here,
First of all: they are just using the best of each type of processer to show if you have a certain kind whether it will be beneficial to shell out more $$ for faster memory
Second of all: the normalized results are the percent difference, not the FPS or time or whatever the hell you are thinking, it is just the difference to what the result was before to what the result now, whether it be faster or slower
Thirdly: if the answer is 114 percent difference,,,that is 14% MORE, grab out your calculator & try it, take a number x 114%,,it will be 14% higher not 2x higher,,that is 200% more
OLD ARTICLE
Nice article, i will keep this in mind when people are asking about ram for new builds. I would also like to see one with a 955 dropped into an AM2+ board and an AM3 board to see if the premium for an AM3 board and DDR3 is worth it.
A second question to be asked is how does ram speed affect the ability to overclock a i7 920? Because this is dine by raising the BCLK value, it also overclocks the ram. Would not a ram capable of 1600 allow the BCLK to be raised to a higher value than 1333 ram?
I wish they had included DDR2 results compared with the DDR3 results with the C2Q and X4.
I thought the normalized results was a nice touch.
I guess no one can read charts too well, for example, the cost is not a factor here, First of all: they are just using the best of each type of processer to show if you have a certain kind whether it will be beneficial to shell out more $$ for faster memorySecond of all: the normalized results are the percent difference, not the FPS or time or whatever the hell you are thinking, it is just the difference to what the result was before to what the result now, whether it be faster or slowerThirdly: if the answer is 114 percent difference,,,that is 14% MORE, grab out your calculator & try it, take a number x 114%,,it will be 14% higher not 2x higher,,that is 200% more
The reason why the first poster and myself were a little confused is because why did read the chart. If you like at the X-axis it states either FPS, Time [mm:ss], etc. That throw me off until I actually read the article again. If the X-axis was changed to "Percent increase in " it would have been clearer. Not complaining, just trying to help the first poster.
Edit: because we
I'll just stick with my ddr2 533 cas4 thank you.
P.S. My RAM vendor owns Crucial/Corsair/Kingston etc. anyday... Produced by Hyundai. =D
-_-
They should have used the same clock speed for all 3 processors
What's the margin of error, and result spreading for different runs?