RAM mixing question

it can work but its not guaranteed to do so you do it at your own risk of getting it to work and stable

any of them memory companies like g-skill will tell you that . only memory sold in full kits are guaranteed to work and run stable



[from gskill]
Q:
I want to install more memory, in addition to my existing memory kit. What are my options?

A:
We do not recommend mixing memory kits, regardless of brand or model.

By mixing memory kits together, there may be compatibility issues such as unable to boot or unable to operate at rated specifications.

Each of our memory kit are thoroughly tested to ensure compatibility within each memory kit. And because we have not tested our memory kit with your existing memory kit, we cannot guarantee compatibility when multiple kits are used.

[ good read with that covered ]
http://www.tomshardware.com/reviews/ddr-dram-myths,4155.html#p2

hould I just return these and get a 2x16gb or 4x8gb pack?

Click to expand...

The 1st option....tho I have read that KL CPUs seem to get a boost from 4 kits ... it was just mentioned as an aside in an article ... have been looking for details.

''btw.. do you guys actually feel like there will ever be a gaming use for 32gb of ram? I've been playing with emulators lately but nvidia has a bug where it uses A LOT of memory and 16gb seems to be just slightly too little.''

only program I got /use that can use up to all the 16gb I got is Y-cruncher nothing else comes close to using the full 16gb

you can monitor that with something like hwinfo64 running in the background to see your daily min and max used

like from that you see physical memory used and available the highest you use for the day or pre program will be the MAX reading .

http://imgur.com/NdJQWMi

''I've been playing with emulators lately but nvidia has a bug where it uses A LOT of memory and 16gb seems to be just slightly too little.. 24gb would be the sweet spot but then again this is nvidia issue and they are working on it.. so I'm not really sure if this upgrade would be worth anything. ''

if you see from that hwinfo 64 that it is using up to or dang close to the full 16gb then yes its worth going more don't forget your os requires up to at least 3.5 gb for its self so I feel you all ways need that buffer regardless or you will go in to ''swap'' off the harddrive [that can bogg things down /slow]

just invest in a 32gb quad factory matched kit

Well at one time ... the mantra was "anything more than 8GB is a waste". That's has changed only recently.

There's also the issue that using DDR3 / DDR4 using 4 sticks outside of quad channel boards has been deemed less than a wise choice as it put a greater load on the memory controller which oft had a negative impact on OC ... folks OCs were found no longer stable after replacing 2 sticks with 4.

There's also the difference between memory needed / used and memory allocated which led to much misinformation being bandied about regarding VRAM. In alienbabeletech's infamous test, the game max payne would not load at a setting of 5760 x 1080 w/ the 2 GB 770 ... so they swapped in a 4 GB 770 and it went fine. Then, on a lark, they put the 2 GB card back in and it ran at the same fps, same quality, same everything as the 4 GB model.

I have never seen this "real used vs. allocated" issue better explained then here:

http://www.extremetech.com/gaming/213069-is-4gb-of-vram-enough-amds-fury-x-faces-off-with-nvidias-gtx-980-ti-titan-x

GPU-Z: An imperfect tool

GPU-Z claims to report how much VRAM the GPU actually uses, but there’s a significant caveat to this metric. GPU-Z doesn’t actually report how much VRAM the GPU is actually using — instead, it reports the amount of VRAM that a game has requested. We spoke to Nvidia’s Brandon Bell on this topic, who told us the following: “None of the GPU tools on the market report memory usage correctly, whether it’s GPU-Z, Afterburner, Precision, etc. They all report the amount of memory requested by the GPU, not the actual memory usage. Cards will larger memory will request more memory, but that doesn’t mean that they actually use it. They simply request it because the memory is available.”

Click to expand...

So ... if i used HWiNFO to monitor max memory, is it being allocated solely cause "it is there" or is it being allocated because it is needed and used ?

The only way to know is to test the subject program in both configurations and measure the impacts in the program. In the referenced GPU test, there was no **real** impact in any of the games tested between 4 GB and 8GB at 4k because the games were bottlenecked by the GPU making RAM amount irrelevant. Do we care if we get 15 fps w/ 4 GB and 18 fps w/ 8Gb if the game is unplayable either way ? Any game that the GPU could keep up with and deliver satisfactory frame rates performed virtually identically at both VRAM amounts.

The other considerations are a) it's harder to get 4 sticks stable at higher speeds than 2 sticks and b) 4 x kits sometimes bear a rather steep price, especially at the higher speed levels.

I guess a lot comes down to how strong/ stable the memory controller in the CPU is and how well the boards bios's handle any thing you put to it

this kit I'm using in this build [ gskill 2x8/16gb 1600 ] just did not work out in a AM3+ build I bought it for at all , but works great in this one intel z87 ?

my buddy used 4x8 1866 but had to down clock it to make stable his son use 32 gb [don't recall the speed ] and it was set xmp and forget and ran first boot full spec ??

in the end you never know what you got till you fire it up

I guess if me I would stay in intels specs of there cpu's

[for a example ]
Memory Specifications

64 GB
DDR4-2133/2400

https://ark.intel.com/products/97128/Intel-Core-i7-7700-Processor-8M-Cache-up-to-4_20-GHz

http://ark.intel.com/products/97150/Intel-Core-i5-7600-Processor-6M-Cache-up-to-4_10-GHz

any thing used over that is overclocking and not intes reasonability as overclocking is all your own risk

Don't read too much into it ... If you have a system that can really, really push the OC... let's say 4.8 - 5.0 Ghz w/ Kaby Lake ... then you need to put more thought into RAM selection. Now there if you are using 2400, it's far less of a concern than if 3200 and over. In recent years, it has always been considered wiser to uses a 2 x 8Gb set rather than x 4 x 4GB set. But, as i said above, I read any article which said, in passing, something to the effect of "ignoring the boost that ya get from 4 sticks w/ Kaby lake" ... this was said as if it was obvious to everyone ... but it was not something I ever saw mentioned before and have been looking to verify.

I like to say that "no one is ever wrong, they have simply been misinformed" and that statement is no more appropriate than when you read "RAM speed has no effect in gaming".

1. When you are talking about effects ... how many effects are you monitoring ? When you see an article (popular youtube topic) that they tested 3 games and RAM speed had no effect, is that a thorough test ? How does this sound to you ? "The people who say McDonalds is bad for you are wrong, cause I ate there 3 times and I didn't die " ? But are we happy with just looking at avg fps ... what about minimum fps ? ... ya know, those leggy spots in a game when we see some stuttering. We do see improvements here with faster RAM. What about in SLI ? ... again we see that two cards oft benefit from faster speeds.

2. To truly know the impact, you must make sure that the system is not being bottlenecked elsewhere or yes RAM speed will have no effect. Most times it will be the GPU, So you have to make sure that your system does not exceed the capabilities of the test system when considering these types of reviews.

3. With a small sampling of games, are you getting enough of a spread to qualify the validity of the tests ? As in politics, folks often start out with a re-conceived conclusion and then will reference things that agree with the conclusion and reject / ignore any information that contradicts it. The STALKER series was memory intensive as are games like F1.

In the THG article this image was taken from, on the same page was a test on Metro 2033 which showed no improvement from DDR3-1600 to 2400 ... then on F1, there was an 11% improvement.



My approach is this ... price obviously will increase with speed ... the price / performance curves starts out relatively flat at first as yields are good ... but at a certain point, when yields start to drop the curve starts increasing in slope where the performance gain gets harder and harder to justify give the incremental increase in price. Right now the 'break point" is about 3200 and that's what we recommend

well yes and no ... it is officially overclocking but not because Intel says so but because the JEDEC specification only goes so far. Intel XMP is officially supported by Intel and they maintain certified compatibility listings.

http://www.intel.com/content/www/us/en/gaming/extreme-memory-profile-xmp.html

You can load predefined and tested Intel® XMP profiles through your computer’s operating system, using BIOS or a specific tuning application.... The Intel® certification program lets memory vendors test their products for compatibility against Intel® Extreme Memory Profile (Intel® XMP). You can view compatible memory DIMMs from leading vendors.

Click to expand...

On the list below that quote you will find Intel certified compatible memory kits up to 4133 and 1.4 volts.


-In Intel certifies a kit as compatible, they have no claim for not honoring CPU warranty if you use it
-If MoBo manufacturer advertises a MoBo as supporting a certain RAM speed, they have no claim for not honoring MoBo warranty if you can't run it because of the MoBos failings
-If RAM manufacturer advertises a RAM package as supporting a certain RAM speed, they have no claim for not honoring RA<M warranty if you can't run it because of the RAMs failings

And as for anything over 1.2 being scary scary ... Intel states otherwise

http://www.legitreviews.com/what-is-the-safe-voltage-range-for-ddr4-memory-overclocking_150115

while lower voltages are better ... if a RAM kit is on Intel's certified compatibility list at 1.4v, they have no cause to deny a warranty claim. Is it necessarily safe ? have to answer the question in 2 ways

No... maybe. it will blow up ... lower is always better.
yes, if it does blow u, they gotta give you a new one.