Memory Speed

winds0r · 12 Jun 2011 at 13:32

Hi,

I have 8GB of Corsair Vengeance Blue memory and it says on the site that it should be 1600Mhz. However, in CPU-Z it shows the Frequency as something else?

I'm not particularly worried (unless I should be), but is there a reason why it does this? Just so I can add the answer to my growing bank of computer knowledge...

Surveyor · 12 Jun 2011 at 13:37

Your memory is running a 1333MHz (2 x 666.7 as it's DDR - Double Data Rate).

You need to change the speed in the BIOS if you want the full 1600MHz.

If you don't know how to do this you'll have to tell us what motherboard it is before we can advise further.

winds0r · 12 Jun 2011 at 13:40

Thanks for that mate.

Changing it to the full 1600 Mhz - what are the benefits of that?

Apologies for the questions, I just like to know why things happen..

Surveyor · 12 Jun 2011 at 13:43

1600MHz should make things "faster" than 1333MHz.

I don't know what sort of system you have but this will give you an idea of what difference RAM speed makes for Intel Sandybridge:

The Best Memory for Sandy Bridge

winds0r · 12 Jun 2011 at 13:45

Here is my system...

I'll try and fiddle in the BIOS and see if I can get it to 1600Mhz. Wish me luck...

Surveyor · 12 Jun 2011 at 13:49

Good luck.

You want this page in the BIOS:

winds0r · 12 Jun 2011 at 13:51

Aye, I took a picture with my phone but you beat me to it. I have set it to DDR-1600MHz.

Out of curiosity, with my CPU overclocked to 4.8GHz, will this affect the memory in any way?

Surveyor · 12 Jun 2011 at 13:54

winds0r said:
Aye, I took a picture with my phone but you beat me to it. I have set it to DDR-1600MHz.

Out of curiosity, with my CPU overclocked to 4.8GHz, will this affect the memory in any way?

No.

The CPU and the memory speeds aren't linked as long as you're overclocking via the CPU multiplier.

If you overclock using the BCLK then that would increase the memory speed as well. But you're not supposed to be doing that anyway.

winds0r · 12 Jun 2011 at 13:55

Perfect, thanks very much for your help

Surveyor · 12 Jun 2011 at 13:56

winds0r said:
Perfect, thanks very much for your help

No problem.

DennisMenace · 12 Jun 2011 at 23:39

You might be lucky and get to 1866MHz.

I have the same RAM (except it's blue) and once I'd worked out how to set it to spec (1600MHz 9-9-9-24 1.5V) I thought I'd try my luck with 1866MHz. It needed "loosening" the timings, so I copied the spec from the faster Vengeance kit (10-10-10-27, leaving other settings at "Auto") and it runs at 1.5V.

So far it's stable with prime95blend for 8 hours, but I might try some other memory stressers just to make sure.

Definitely worth a go IMHO, what have you got to lose?

Surveyor · 13 Jun 2011 at 00:18

DennisMenace said:
You might be lucky and get to 1866MHz.

I have the same RAM (except it's blue) and once I'd worked out how to set it to spec (1600MHz 9-9-9-24 1.5V) I thought I'd try my luck with 1866MHz. It needed "loosening" the timings, so I copied the spec from the faster Vengeance kit (10-10-10-27, leaving other settings at "Auto") and it runs at 1.5V. So far it's stable with prime95blend for 8 hours, but I might try some other memory stressers just to make sure.

Definitely worth a go IMHO, what have you got to lose?

There's no real benefit to be obtained running 1866MHz with looser timing.

The Best Memory for Sandy Bridge

DennisMenace · 18 Jun 2011 at 01:34

Surveyor said:
There's no real benefit to be obtained running 1866MHz with looser timing.

Personally, I would consider 8% faster read speed (2% write) and 8% lower latency a "real benefit". Maybe not all that noticeable, but still real.

That article emphasises the greater benefit of frequency over latency. In most tests, the difference between CL7, CL8 and CL9 is marginal (at the same frequency), so I'd assume that CL10 would not be drastically worse than CL9. But the jump from 1600MHz to 1866MHz shows a marked improvement in a lot of the (admittedly synthetic) tests.

Whether that improvement is worth paying a substantial premium for is debatable (the article would suggest not). But if you can get it for free, surely it's a no-brainer? After all, this is overclockers.co.uk, not leaveyourcomputeratbogstandardsettings.co.uk.

PS. It's now passed 24+ hours prime95blend and 8+ hours memtest86+. :cool:

Surveyor · 18 Jun 2011 at 02:10

DennisMenace said:
Personally, I would consider 8% faster read speed (2% write) and 8% lower latency a "real benefit". Maybe not all that noticeable, but still real.

That article emphasises the greater benefit of frequency over latency. In most tests, the difference between CL7, CL8 and CL9 is marginal (at the same frequency), so I'd assume that CL10 would not be drastically worse than CL9. But the jump from 1600MHz to 1866MHz shows a marked improvement in a lot of the (admittedly synthetic) tests.

If you compare CL8 1600MHz against CL9 1866MHz (as they don't test CL10) then the "real World" gains of the 1866MHz RAM are:

Image editing +0.8%
H.264 Video Encoding -1.0%
Multi-tasking Test +3.2%
Overall Score +0.6%
Gaming - no difference

As I said "there's no real benefit".

But if you're after that extra tiny percentage then go for it.