RAM compatibility

>>299233

The RAM is working fine. What were you even expecting to "feel"?

>>299233
You really only "feel" RAM when it is lacking. Open up 30 tabs while gaming and watching a 1080p video. Then remove 4GB and repeat. You will probably feel a difference.

But once you have enough for whatever you are doing, it is just that - enough.
Having 3GBs of unused memory does not make whatever is stored in the other 5GB any faster, for example.


There are many games in which the minimum fps can increase with more RAM or which flat out need the 8GB and can even profit from having 16, but as I said, you only really notice this kind of thing if you don't have enough.

were you expecting dual channel? both modules are different base speeds and probably different latencies, for dual channel to work you need to tweak them to "be the same", and given its a laptop (since they are sodimms, there are exceptions though) you wont have much luck since most laptops lack the settings for it.
and even if you have a fully functional dual channel setup you wont notice it in any "regular" applications. and like >>299257 said, you will most likely only notice that the machine doesnt slow down as much during memory pressure situations (lack of available ram)

>>299544
No they don't, and they haven't since 2004. Not even the size has to be the same: modern chipsets simply run the overlapping bit dual-channel and the protruding bit single-channel.

Intel calls this "Intel Flex Memory", but AMD can do it too.

http://www.intel.co.uk/content/www/uk/en/support/boards-and-kits/000005657.html#flex

(You'll note that even outside of flex configurations, Intel explicitly states that the brand, timing, and speed *don't* matter and *can* be mixed).

>>299545
This is 99% true. Binned speeds and timings do not matter at all for compatability. But not all ICs will work on every board, and not all ICs will work well with any other. For example, if you run Elpida BDSE together with Hynix MFR, you will often find that the BDSE sticks will not get recognized at startup at all.
But once they actually run, and programs like CPU-Z say "yes you got 8GB", then they are compatible.

>>299545
>flex mode
that might be true for most desktop motherboards, on laptops its an entirely different story. just the default speed difference alone will make it revert back to single channel (if the bios doesnt sync both slots to the lowest common denominator, which is very uncommon on laptops)

well thank you very much guys for this info. Much appreciated :D

>>299562
>Intel literally, specifically and explicitly says that difference in speed doesn't matter, on intel.com
>Yet I know better than Intel how Intel chipsets work

You don't have a username to get defensive over. You can just say "I was wrong, I'm not going to argue with Intel's own documentation".

>>299588
>Intel's own documentation
literally, the first line of the article is
>Several types of memory modes can be configured on Intel® Desktop Boards
notice the "desktop" in there? (inb4 same)
>get defensive
uh?

>>299622
Oh, I'm terribly sorry, here's a datasheet for one of their ancient mobile chipsets that says THE EXACT SAME THING. Y'know because all Intel chips since DDR2 have been able to do this, which is why 3GB and 6GB laptops are a thing.

http://www.intel.com/assets/PDF/datasheet/320122.pdf#page62

>>299642
like ive mentioned before, that only pertains stick size (memory amounts), speed and latencies are another subject
the laptop with physical 3gb limits exist because of the limited bus memory address width (24bit instead of 48bit?!? i cant recall exactly it was quite some time ago)

also, why so condescending? you that same rage fueled guy from the other day? you might wanna get off your high horse and get that checked man...

>>299645
No it doesn't, it exists because they liked 3GB as a price/performance point. You absolutely could (and still can) put 4GB in them.

But if they couldn't run dual-channel because they had unmatched DIMMs, that would be a performance disaster, it would have got the laptop panned in reviews, and there's no way these laptops would have ever existed in the first place.

If being told you're wrong triggers you, maybe try not being wrong?

File: 4L_hFSbrjVv.jpg (56KB, 724x217px) Image search: [Google]

56KB, 724x217px

>>299645
>the device technologies may differ
Now here's you being explicitly contradicted, yet again, by documentation direct from Intel. Yes, it's documentation about laptops. Yes, it's talking about device manufacturers and timings. Yes, it's right there in the documentation you just claimed was not relevant. No, you don't get to do any more wiggling.

I'm not going to go get a signed affidavit from Gordon Moore, so if you still don't think you're wrong, it's time for you to find some actual evidence that supports your position instead of directly contradicting it.

>>299650
>You absolutely could (and still can) put 4GB in them
ofc, some of them would just not boot if you do, but you could
>they couldn't run dual-channel because they had unmatched DIMMs
you keep saying that, but its just the stick size that matters, if you use pc2 6400 and pc2 5300 (even if they are the same size) if the bios does not match the latencies and speed to the lowest common denominator you will not get dual channel
>being told you're wrong
uh? arent you the one saying stuff as if it was part of the documentation?
some laptops didnt have the necessary bus width for some stuff (less data/address lines == cheaper pcb), what that has to do with memory flex mode i dont know, your introduced it you could explain it. also could you point where, in the various documents you linked, where is the unmatched latencies and unmatched speeds dual channel mode?
>try not being wrong
ayy, maybe ill learn something if you show me ?!?

>>299654
those are the rules, for activating dual channel mode, you must set them, or let the bios do it for you, since most laptop bios dont do it, and neither possess the settings to change it manually im failing to see the point of your post

please tech me ?!?

>>299658
>if the bios does not match the latencies and speed to the lowest common denominator you will not get dual channel

>In all modes, the frequency of system memory will be the lowest frequency of all SO-DIMMs in the system
>In all modes

Do you have some kind of problem reading?

>>299658
>some of them would just not boot if you do, but you could
[citation needed]
You're just making shit up now.

>>299666
>problem reading?
im still not seeing where its written that it can be different values for speed and/or latency. if you are expecting the spd to have the same info on all sticks then that means they are similar/same sticks with similar/same ics, no? or maybe i do need reading glasses
>[citation needed]
im unable to. i did see it happen multiple times: some more known branded laptops (dell, hp and clevo rebrands mostly) worked just fine with 2gb+2gb=3gb, others less known branded ones, any configuration higher than 3gb would lead to bios load failure (some wouldnt even beep code). take it as you wish, which will be "fake and gay"/"made up shit" i assume judging by your rhetoric

File: 4L_Eae9pWKD.jpg (88KB, 750x579px) Image search: [Google]

88KB, 750x579px

>>299659
>the value programmed here must match the CAS latency of every SO-DIMM in the system
>must match
>every SO-DIMM
Now I know you really, really, really want to be right, but this chipset is physically not capable of running different banks at different speeds.

The BIOS must "set them", or none of the RAM will work. It is physically impossible to run different timings. The timings for each bank are all the same because it is impossible for them to be different.

Thus, as stated in the very first source, timings HAVE NOTHING TO DO WITH DUAL-CHANNEL, as it is actually impossible to run the banks at different timings.

If the computer boots, and the RAM is detected, it is a GUARANTEE that it is running at the same timings.

This means it will do dual-channel with either NONE, or ANY AND ALL identically-sized DIMMs that actually boot, and that frequency and timings have nothing to do with dual-channel, and that you. are. wrong.

>>299672
Stop pollutiing this thread you fucking faggots.

> others less known branded ones, any configuration higher than 3gb would lead to bios load failure (some wouldnt even beep code).

This is not a thing, you are lying or incompetent. This is not a capacity problem, but a general incompatability issue between board and module, which can happen as stated in like 2nd or 3rd post here.
Everything important to op has been said, just fuck off if you only want the validation of not being wrong.

>>299672
RAM doesn't complain about being run too slow, and the memory controller can only run at one speed. It doesn't run each bank at its own special snowflake speed*, because it can't.

It has to run every bank at the same speed, so it picks a speed that works for all banks and runs every bank at that speed.

Which is why timings don't matter for dual-channel.

* I think this is the root of your misunderstanding: I think you think it can run each bank with independent timings, and then having different timings on each bank would somehow muddle-up the memory controller and so it wouldn't be able to do dual-channel. This is wrong, because a) the hardware can't do that, and b) even in "single channel" mode it's still doing dual-channel accesses, it's just not interleaving the physical address across banks.

>>299678
>>299681
>>299683
jesus christ all this vitriol...

consider the following scenario, two sticks of ram, one pc2 6400, one pc2 5300, each have only one spd table entry for their rated speed, for this particular case lets make them one 1gb other 2gb, since you are so keen on flex mode, so you stick them in, and the bios should set them to the lowest speed and highest timmings right? but one of the ics on one of the sticks does not support those settings. so what do you get at the end: failure to boot in dual channel or working single channel? laptops without settings to change latencies/timings (which is a big chunk of them) dont set dual channel on different sticks (read: different spd entries), jesus...

>pollutiing this thread
the thread has served its purpose on >>299563 , should be fine, at least its contained

>>299699
You get failure to boot, as discussed earlier in the thread.

Your confusion lies in thinking of dual-channel asymmetric as "single channel" just because that's the vernacular term for it. Even if the address space isn't interleaved, there's still two channels. If the SPD timings in one or more of the DIMMs are wrong, the computer won't boot, because the only way it can operate is with every bank set to the same speed and timings.

I don't know why this is so hard for you to understand; I presume it's because understanding it involves you admitting that you were wrong.