Faster DDR4?

Anyone using 2400, 2666 or 2800 DDR4 in their laptop? I am thinking of moving to 2 16Gb sticks since I do VM work, but wondering if moving to faster RAM would actually have any benefit also.

 

These are great...a little pricey but all good. Definitely worth it if you're running VM..

 

That's over 30% more than any of the kits on newegg.

 

I know...and the other reality is that sometime in the very near future, these will be half the price if not more.

I guess it's just one of those things of paying a premium for having the shiniest now. Then there's the brand loyalty that comes in the play....

Have you found any other brands that are compatible 16x2 RAM sticks for our good ol' Alienware's? I haven't even looked, but am interested as I'll be needing to buy some more for another 17" R3 soon...

Cheers

 

I think he asked about faster memory. Those are standard 2133Mhz. Anyone actually owns a faster 32GB kit? Heard Kingston hyper-x modules are good, they get installed in some clevos i looked at.

in terms of benefits, I think it's worth it. I overclocked stock ram to 2400MHz and got 5-10% improvement in benchmarks.

 

So I take it no one has actually tested faster RAM?

I'm wondering what would be more beneficial, tighter timings or more MHz.

 

You posted this on reddit too didn't you? Listen, it's your money, but unless you're doing something like converting huge (10GB+) files with handbrake or enormous work on a professional level, faster ram isn't going to do anything for you, more physical ram might but 16GB is enough to do a lot of work, probably more work than is reasonably capable of being done on a quad-core laptop honestly.

Techspot did a good piece in comparing how applications reacted to 4/8 & 16 GB of system ram from a gaming perspective but they also included several productivity tests in there that you should look at. In any case, faster ram is almost certainly not going to do anything noticeable(or even barely measurable) for you as much as more ram will, if you actually need more ram.

Synthetic benchmarks are just that though.. benchmarks, he's looking for real-world results.

When Anandtech did a piece of 1866 DDR3 CL9 vs 2133 CL15 DDR4 ram in their skylake review, the average delta was < 0.5% between the two on just about every recognized real-world productivity test. At best it was around 4% and at worse DDR4 was acutally slower than DDR3 by 2%.

The only legitimate area where faster ram makes sense is if you're gaming off the IGP, but that's never going to be the case in a laptop with a dGPU.

They go hand in hand. Faster clocked ram can read/write more quickly but often have to have higher delays in requests, which slow it down. Slower clocked ram might take longer to read/write but they are not as burdened by as long of a delay between requests from a very basic perspective. It's partially why in anandtech's review the DDR3L did so well, it had been running (very) tight CL9 timings which helped it against the more loosely timed DDR4 ram. Should the DDR3L had been a more JEDEC compliant CL13 the results would have been (slightly) different but not by much. TBH it doesn't *really* matter outside of benchmarks.

Ideally though, you want a good combination of clock speed and tight timings. Given the limited options in the R3 to control clock speed & timings, If you actually need to expand the system's memory, you'd want to look at enthusiast ram, almost anything in say, the Kingston Hyper X lineup is going to be capable of running lower latencies than JEDEC standards.

 

I tested my Micron RAM at 2400 CL15 using a few benchmarks, got the best improvement in CB15 (+10%) which clearly isn't supposed to reflect the gaming scenario.

For data-intensive apps (rendering, graphics, compilation, data analysis etc), I think there will be a significant benefit to faster RAM.

 

It's interesting that your results are completely different from say, anandtechs results with CB15...
Here's Anandtech's review of 2133 vs 3200(CL16) mhz on Haswell-E
http://www.anandtech.com/show/8959/...-3200-with-gskill-corsair-adata-and-crucial/5
There is just a 2% delta between 2133 and 3200 mhz ram in their testing with multi-threaded CB15. In Single-thread there's 0 difference. The same scenario plays out for basically every single test they run, and they're comparing a much bigger difference in speeds here.

Here's their skylake review which supports their same conclusions;
http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/12

There is nothing I have seen from any source which reflects that faster ram provides a meaningful benefit.

 

Thanks for the posts guys, I'll just stick with the stock memory unless I see my VM work would benefit from the upgrade to 32GB.