Read more.Quote:
But we test it in a way that may surprise you.
Printable View
Read more.Quote:
But we test it in a way that may surprise you.
This platform seems so pointless and such a waste of money. Anything beyond a 2500k and some average speed RAM seems totally unnecessary, hell I'm still using an old phenom II with DDR2 and I see no reason to upgrade :)
That's only looking at it from your point of view, I understand it's total overkill for the casual user and would be simply for the e-peen extension :p, but it's overclocked prowess makes it attractive to anyone who does CPU intensive work. Basically if I can process my work faster I can improve the product or increase the output and charge more, so the value is justified.Quote:
Originally Posted by MustardCutter
cheers
brasc
Bandwidth tests based only on SiSoft Sandra are skewed. It appears only Sandra is able to provide for tangible benefit of three and four channel memory over dual channel. None of the other leading memory tests (i.e. AIDA, MaxxMem, etc.) provides evidence of bandwidth increase scaling with more then two DIMMs. Is it possible to add additional bandwidth benchmark for good measure?
maube try these in an AMD machine?
Can you add to this test?
How would the 8GB dual channel results compare to 8GB Quad channel 4 x 2GB of the same spec?
Something like, corsair-ddr3-vengeance-jet-black-pc3-12800-(1600)-non-ecc-cas-9-9-9-24-xmp-1.5v for example if G-Skill don't have a product at that point?
As for more memory not yielding better performance, that really depends on your workload. My laptop (Thinkpad W520) runs a couple of instances of SQL Server, a couple of VMs, IIS, and adding Visual Studio and Outlook to the mix, I dread to think what that would be like with anything less than 16GB of RAM!
and what about mixing sets. in DDR2 and DDR world you could mix sets without an issue, there's a myth growing up that you can only have 'sets' and that 2 single of the same type are not the same as a dual channel kit. and a dual kit + a single does not equal a triple kit. Could we please confirm or bust this myth. There will be failures to be compatible but these should be minimal. RAM on certain timings at certain voltage should be brand independent else it is not conforming to specs?
@13thmonkey -- As I understand it, the biggest issue with mixing sets is with the XMP profile used to define the higher speed grades. XMP profile requires that you identify the number of modules in the kit, so if you mix two sets of dual module kits (for four total modules), the XMP profile will identify each as part of a 2-up kit. Depending on *your* motherboard/BIOS/module combo, the memory may fail at that speed or the BIOS may drop the XMP profile completely and revert to the stock timings. If that happens, you would have to tweak the memory settings by hand to get them to the performance they are spec'd for.