DDR3 1333 vs 1600

killadoob said:
Rift you are wasting time and money for next to no improvement man.

Honestly even 2000mhz memory with lower latency is not going to give you a speed boost that justifies the large amount of money spent.

Rather get a raptor or SSD drive for the same money if possible, not sure how much you got to spend.
Click to expand...

Yip, I know. Was just sought of wondering about it and needed some advice. Seems like I've got it.

However, with my mentality I now want to know more. :D Pretty interesting stuff, especially once the marketing boys/girls get hold of it.

SSD would be biggest improvement, but much more cash outlay than I was thinking of. I wouldn't be dumping the current RAM in the bin, but add a couple of hundred if it was going to improve things. With an SSD it'd be a big improvement but at a massive expensive. Can't wait for those buggers to drop in price. Still out of my price bracket especially considering the size one gets.

But thanks to those who gave good and quick answers. :D Much appreciated. However, no thanks to my stupid brain for now leaving me with more questions on modern architectures.
 
TheRift said:
Everything! :D
Click to expand...

Well here is the Everest benchmark of DDR3-1600 in triple channel for now:

DDR3-1600_Triple_Channel.jpg
 
The_Techie said:
I'll see if I can keep CPU frequency, BCLK and timings constant, then the only difference should be RAM speed.

Which tests would you like me to run?
Click to expand...

We already have review on tests as you suggest. It makes only a 3-5% difference at best. My take is that if you want to use higher speed memory you have to increase the FSB/speed of the system to benefit. Thus if you use DDR 1333 then adjust the FSB accordingly or overclock the system to the max. Then run the real world tests as per the articles posted. Then install the DRR1600 then adjust the FSB accordingly or overclock the system to the maximum agin and then run the same tests all else reamining constant.

Then we look at the results again in comparison!
 
Y-Cruncher 25M DDR3-1600

Y-Cruncher_25M_DDR3-1600.jpg


I am Penguin said:
We already have review on tests as you suggest. It makes only a 3-5% difference at best. My take is that if you want to use higher speed memory you have to increase the FSB/speed of the system to benefit. Thus if you use DDR 1333 then adjust the FSB accordingly or overclock the system to the max. Then run the real world tests as per the articles posted. Then install the DRR1600 then adjust the FSB accordingly or overclock the system to the maximum agin and then run the same tests all else reamining constant.

Then we look at the results again in comparison!
Click to expand...

Well to run anything above DDR3-1066 (I believe, might be DDR3-1333) on the X58 chipset you have to increase the BCKL regardless. But you're talking about overclocking headroom between different RAM speeds, I want to show whether there's a difference going from DDR3-1333 to DDR3-1600 as I believe that's what the OP asked.

killadoob said:
Super pi tests cpu though? Does it test mem as well?
Click to expand...

We'll soon see. I'm running everything, just as TheRift requested.
 
Last edited:
The_Techie said:
Well to run anything above DDR3-1066 (I believe, might be DDR3-1333) on the X58 chipset you have to increase the BCKL regardless. But you're talking about overclocking headroom between different RAM speeds, I want to show whether there's a difference going from DDR3-1333 to DDR3-1600 as I believe that's what the OP asked.
Click to expand...

That was the original question.

The_Techie said:
We'll soon see. I'm running everything, just as TheRift requested.
Click to expand...

Thanks. :)
 
I am Penguin said:
Thats what he said! as I understood him.
Click to expand...

I have Kingston DDR3 1333MHz (2GB x 2) in there. Was considering pulling them out, selling and upping to the Kingston DDR3 1600MHz (2GB x 2). The difference would be around R200-R300.
 
OK with my current setup I won't be able to reach DDR3-1333 without changing the BCLK so I settled for DDR3-1200 but at CL8 instead of CL9 so it should be close.

Everest DDR3-1200:

DDR3-1200_Everest.jpg
 
Y-Cruncher 25M DDR3-1200:

Y-Cruncher_25M_DDR3-1200.jpg


So the conclusion that we can draw from the 3 tests is that it only really makes a difference in read bandwidth but DDR3-1200 CL8 already offers more than what is needed by these 3 tests.
 
you would be able to sell your current RAM for close to its market value?
 
The_Techie said:
OK with my current setup I won't be able to reach DDR3-1333 without changing the BCLK so I settled for DDR3-1200 but at CL8 instead of CL9 so it should be close.

Everest DDR3-1200:
Click to expand...

Looks like 2GB down on the read, but the rest is minimal difference.

I installed Everest 5.30 now, had a go with it via RDP. That EPU thing from Asus messes stuff up a bit. Says RAM is 668.9MHz, FSB is 200.7MHz and the ratio is 20:6 :P Need to turn all that off this evening and have another run. I hate seeing weird numbers. :P

Damn that i7 is fast!
 
The_Techie said:
So the conclusion that we can draw from the 3 tests is that it only really makes a difference in read bandwidth but DDR3-1200 CL8 already offers more than what is needed by these 3 tests.
Click to expand...

And considering the fact RAM usage is a big mix of reads and writes, etc, one shouldn't see much of difference in normal use.

Thanks, man, appreciate those tests. Don't appreciate you showing off the i7 though. Makes me jealous. :D

Wonder my clock numbers are so odd.
 
TheRift said:
And considering the fact RAM usage is a big mix of reads and writes, etc, one shouldn't see much of difference in normal use.

Thanks, man, appreciate those tests. Don't appreciate you showing off the i7 though. Makes me jealous. :D

Wonder my clock numbers are so odd.
Click to expand...

Personally I'd just stick with DDR3-1333. The only reason I went for the DDR3-1600 was because I wanted to overclock my CPU to 4GHz which required a 200MHz BCLK which results in DDR3-1600 RAM speed by default.

No problem. It is rather fast, yes :)

It might be the EPU messing around, or even SpeedStep (only seen SpeedStep interfere with Everest once, though).
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X