General question about O'clocking (RAM)

[AnyOldName3]

I don't think you understood the quotes you quoted.




Making RAM run faster than it is rated is more likely to cause problems with the RAM than to remove problems with it.

If the RAM can't get data to the CPU fast enough, the CPU will wait for it, and then perform the correct calculation. It won't try to perform the calculation without the data, and therefore get a garbage result, wich is what someone was suggesting.

The stability of a chip is a measure of how likely it is to perform the correct calculation, and return the correct result, when given an instruction. It has nothing to do with how long it actually takes for the chip to return the answer.

[omega_rugal]

Maybe i must have said unstable PERFORMANCE (sometimes faster than at stock speed, sometimes no speed gain, sometimes even slower.)

back on the original post

if you meant DDR3-1333 RAM (which is rated at a maximum of 666-667 MHz) running at 900 MHz...then try it, does your PC boot?

[NaturalViolence]

omega_rugal Wrote:high latency memory overclocked (forced) to work faster may lead to that, we are talking about overclocking aren`t we? simply pumping up the RAM clock and/or the voltage without adjusting the RAM timings will either lead to poor performance or crashes due memory corruption.

I agree with this. But this is the exact opposite of the claim that you made. You stated that slower memory would cause system instability. You even implied that it would cause a GPU to crash. When I repeatedly stated that this was not true you responded with:

omega_rugal Wrote:Oh boy...

Fine, whatever you say sir...

Implying that you still disagreed. I don't see how I could have been any clearer in the way I worded my post. Yet you're still twisting my words.

omega_rugal Wrote:Maybe i must have said unstable PERFORMANCE

Yes you should have. This makes a lot more sense. When you say something is unstable that means that it is crashing or generating incorrect results.


Also good job AON3. I couldn't have explained it better myself.

@omega_Rugal

I'm going to assume for now that you never intended to make that claim and just kept repeatedly wording your posts incorrectly somehow.

I mean no offense by asking this question but out of curiosity is English your first language?

[admin89]

Quote:if you meant DDR3-1333 RAM (which is rated at a maximum of 666-667 MHz) running at 900 MHz...then try it, does your PC boot?

Yes , I can make it run even slower than 900 MHz , CPU is stable at 3.9GHz with Prime95

[omega_rugal]

Quote:I mean no offense by asking this question but out of curiosity is English your first language?

No, spanish.

Quote:You even implied that it would cause a GPU to crash

Mine does, it`s an APU, uses system RAM to work. RAM speed it`s the only diference between the system the guy`s at Tomshardwareguide and mine, which makes me suspect that RAM is a bottleneck here.

a discrete GPU obviusly won`t crash or anything

like i said i will get a 1866 and see if the gpu can go at over 900 mhz

[NaturalViolence]

So apparently you do still believe that claim.

omega_rugal Wrote:Mine does, it`s an APU, uses system RAM to work.

Just because it uses main memory doesn't mean that is the source of the crashing. If the memory is not overclocked then that would make no sense.

omega_rugal Wrote:RAM speed it`s the only diference between the system the guy`s at Tomshardwareguide and mine,

No it isn't. Even if every component in both systems is the same model you still have different chips. And different chips overclock differently even if they belong to the same model. The same thing applies to motherboards and other components. You also have different chassis environments and ambient temperatures to account for. So don't assume it's the ram. You have very little to go on here.

Also a sample size of 2 is hardly strong evidence.

omega_rugal Wrote:which makes me suspect that RAM is a bottleneck here.

We've been over this. Yes it is the bottleneck. But that has nothing to do with stability.

Using slower ram does not cause microprocessors to magically become unstable. If it did this would be a nightmare for system vendors and application developers. System vendors wouldn't be able to guarantee the stability of their systems since a wide range of different memory is used with each microprocessor and application developers would have to worry about invalid issues from users whose microprocessors were randomly producing incorrect results due to instability caused by slow memory.......even with a perfect chip. But of course application developers will never ask you about your memory speed when you're filing a bug report because they all know that it has absolutely zero impact on stability. They may ask whether the system is overclocked or not since that can invalidate reported issues that were actually caused by a bad OC.

As far as CPU/GPU overclocking is concerned slower memory raises overclocking potential and faster memory reduces overclocking potential. This is widely known.

As long as the memory isn't overclocked it won't cause instability just because it's slow.

omega_rugal Wrote:a discrete GPU obviusly won`t crash or anything

Of course it won't. And for the same reason. Discrete GPUs also have DDR based memory. The only difference is they usually have a small amount of dedicated memory not directly accessible by the cpu (only by DMA) with a wide high frequency high latency bus. If slow main memory causes IGPs to become unstable then slow video memory should cause discrete GPUs to become unstable. But of course it doesn't because that's ridiculous. You can easily test both without buying new ram by underclocking your memory.

[omega_rugal]

Whatever...

i`ll keep toying with my A8

get one and do the same.

maybe i should mention that my IGP does boot at 900 MHz, but crashes everytime i try to run anything gpu intensive...

[NaturalViolence]

Increase the voltage to the GPU then. Faster memory isn't going to make it more stable.