G.SKILL Ripjaws Z Series 32GB (OC) ??????

[Cole]

Well......Already ordered from New Egg. But since then I've been educated by reading some Tradesman posts regarding DRAM description by MOBO manufacturers. I've ordered the 2400(OC)**F3-19200CL10Q-32GBZHD** flavor for a build on the ASUS P9X79 DELUXE with an I7 3930K. I do plan to over clock but only modestly since I won't use the rig for gaming but for audio recording and virtual instruments. If I have to order slower RAM i will but just wondering if i should expect that this combo won't work out of the box **reliably** with some auto over clocking. Stability is just as important for my uses as speed. If this will likely cause some issues then this kit will go for cheap and I will replace it with the REAL recommended DRAM. Any thoughts? Has this this been a troublesome set up historically? Thanks.

[Tradesman]

Will want the latest BIOS and to have in mind that not all 3930Ks can carry 32GB of 2400, but some can. With that set, at worse should be able to drop to 2133 and tighten timings to CL9 and have near same performance as 2400/CL10. Think we should have them working optimally with little to no fuss....When they first arrive, make sure you have the latest BIOS, install, enable XMP and set profile to 2400 and give it a go. Any problems, give us a shout, I'm in and out every day

[Cole]

Will want the latest BIOS and to have in mind that not all 3930Ks can carry 32GB of 2400, but some can. With that set, at worse should be able to drop to 2133 and tighten timings to CL9 and have near same performance as 2400/CL10. Think we should have them working optimally with little to no fuss....When they first arrive, make sure you have the latest BIOS, install, enable XMP and set profile to 2400 and give it a go. Any problems, give us a shout, I'm in and out every day

Very........very cool. I thank you Sir. Man I hope that every product in this build has this level of support and knowledge sharing behind it. No wonder why so many people recommend G. Skill

[Techmeister]

Regardless of what frequency the RAM is rated at, you wouldn't need to order "slower RAM" as you can run the faster RAM at lower frequencies without issue, so you're good either way.

There are very minimal system performance gains above 1600 MHz. RAM speed on typical Intel and AMD desktops, so no matter what frequency you run above 1333 MHz. you won't see much system performance change, except in PCs running an APU instead of a CPU.

[Tradesman]

On Techmeister's 2nd point, I both agree and disagree, as far as benchmarks, as far as the DRAM itself it's really a combination of the freq and the CAS i.e. in basic steps going from 1600/CL9 is close to 1866/10 which is close to 2133/11 and on to 2400/12, where you will primarily see the increase is in bandwidth. On the other hand someone with budget 1600/CL10 will see a much larger difference going to a 2133/9 set (and that 2133/9 set can typically run down at 1600 with a CL of 7 or lower as the ICs used for the higher freq ICs are stronger than the lower rated ICs used by a 1600/10 set.

Another thing that plays in with DRAM and system performance more and more in this day and age is how folks use their computers, most benchmarks that revolve around DRAM go back years and are strictly centered on a base test of READ/WRITE?COPY/LATENCY and originated with how people used their systems running one or two apps at a time....in today's world where more and more real-time multitasking is going on both the higher freq and greater amounts of DRAM can greatly increase performance.

[Techmeister]

I respectfully disagree that faster than 1600 MHz. RAM offers significant increases in system performance because I and many other people have tested this many times with real applications, multi-tasking, stress testing, etc.

You will find that RAM benches show a significant theoretical gain with higher frequency RAM because the benches assume full time saturation of the RAM which never happens even when stress testing a PC or RAM. The benches in fact exaggerate the performance gains that will happen with real apps. All people need to do is run the test themselves with real apps.

Typically going from 1600 MHz. to 2000 MHz. in a video game that is memory intense and runs 33 FS with 1600 MHz. RAM with run at 36 FPS with the same RAM running @ 2000 MHz. thats 3 FPS or ~9% which sounds great except for the fact no one can see 3 FPS difference in a video that is running @ 33 FPS. If it were +3 FPS in a video that ran @ 10 FPS @ 1600 MHz. then that would be useful because 10 FPS is almost an unusable game frame rate.

Changes in latency make even less change in system performance than increased frequency. This is because as the frequency increase one clock cycle becomes less in real time. Thus is you go from a CL 10 rated DIMM to a CL 7, there is almost no change at all even in the benches that inflate the real results. Test it for yourself with the RAM you have. Run it at it's fastest CL and then increase the CL 2-3 points and test again.

I know RAM makers don't want folks to know this information and I'm not trying to hurt G.Skill sales, but I think people should educate themselves and not assume that faster RAM equals significant system performance increase. Significant for this discussion means a tangible difference that you can see or feel when using your PC. It does not mean minute gains in benches that inflate RAM performance gains.

Running your own tests will tell you what works best for your application. Tom's and Anandtech have run extensive benches with all sorts of software and the only significant system performance improvement with faster RAM than 1600 MHz. was with systems using APUs where the GPU section could use the added bandwidth. Their own words were: There is no significant gains going to the faster RAM. The tests are on their websites for those interested in reviewing a lot of test results.

Buying faster RAM for the fun of overclocking it is perfectly fine IMO as this is entertainment that you are paying for. People also like to have the fastest ____________ (Fill in the blank) for bragging rights. If that's you, go for it. No harm, no foul. The information I provide is to help those who desire to be techncially informed. You can use it or discard it. That's perfectly fine by me.

[GSKILL TECH]

Techmeister, did you overclock the CPU when overclocking DRAM?

I don't think G.Skill breaking DDR3-3900 was for pure entertainment, nor was it the same system performance as DDR3-1600. How embarrassing would we be to sell RAM exponentially higher in price that has no performance advantage over standard DDR3-1600?

Take a look at XtremeSystems.org, you can find many extreme overclockers and their results. Compare it to wherever you get DDR3-1600 benchmarks and then come back to tell us how it goes. =)

We hold many processing speed records like for HyperPi. If there was no performance advantage, what kind of records are we holding? Everyone would be holding the record with DDR3-1600 RAM.

The only reason people don't think it is a significant difference is because they have no purpose for such high performance, or they are using it simply for gaming. The technology we are dealing with is way beyond an average person's use. For fast gaming, all one needs is a basic Dual Core CPU. The Z77, X79, and latest technology are all essentially super computers that we don't even need. DDR3-2400+? You're right, it's definitely an over kill if people can't understand how to use it.

Thank you
GSKILL TECH

[Techmeister]

Objective scientific testing refutes the false beliefs that there are any SIGNIFICANT system performance benefits in using RAM above 1600 MHz. on a typical desktop PC be it Intel or AMD. Feel free to read the extensive testing at Tom's and Anandtect. They had hoped to find a justification for buying more expensive/faster RAM but there simply isn't any. They were disappointed as they are overclockers like most enthusiasts and the test showed there was no significant system performance gains other than with APUs.

This was from two independent, reputable sources who have no direct financial interest in the test results. If anything they'd want the RAM to show value as they have advertisers who sell more product when these PC hardware sites report positive results on a tested product. That's why it's good to not only read extensively to be informed, but to conduct your own tests and see with your own eyes the results.

People are certainly free to buy whatever makes them happy including buying the fastest RAM on the planet. As far as system performance however, there will be no significant performance gain with RAM frequency above 1600 MHZ. other than in systems using APUs. Test after test with overclocked and OE frequency CPUs proves this, which is no surprise with DDR3 RAM @ 1600 MHz. not being a system bottleneck. I recommend that people with 2000+ MHz. RAM run their own test with the RAM @ 1600 MHz. and then @ 2000+ MHz. and see for themselves.

My goal is NOT to reduce sales of G.Slill RAM, but to explain to those who don't understand why the 2400 MHz. RAM that they bought doesn't improve system performance substantially. Most PC enthusiasts mistakenly think that because the RAM runs at a higher frequency that this increased frequency or the increased bandwidth will make an appreciable improvement in their CPU based desktop system and it does not. If the system was bottlenecked with 1600 MHz. RAM then the higher frequency RAM would show a significant improvement as it does with an APU that has a data bottlenecked GPU section.

[Cole]

Well very humbling that a clueless guy like me has some very technical insight provided in this post. Thanks to the both of you.........Now I have the 32gb kit installed but the XMP profile fails the overclock. I've tried several manual tweaks that I've found on the internet for nearly 48 hours straight with no luck other than system speed showing 3202 a couple of times. TM you said something about running the ram at 2133. I've tried that as well and still no luck. I expected the XMP to be more or less automatic. Any experiences you could share? Thanks.

[Tradesman]

Do you have the latest BIOS? I know I mentioned it before, so just checking. Can you post up your DRAM timings (bath basic and advanced) along with your system voltages