True bandwidth differences.

[dippy_skoodlez]

ALright... Been talking about people doing some extreme OCing with 250fsb and stuff, and someone complain that cas3 is worthless. Well, I did some tests. Heres what I found wisth Sisandra.

@ 133mhz:

Cas 2 = ~1922Mb/s
Cas2.5= ~1920Mb/s
Cas 3 = ~1911Mb/s

@ 166mhz

Cas 2 = ~2432Mb/s
Cas2.5= ~2429Mb/s
Cas 3 = ~2426Mb/s

@200mhz

Cas 2 = ~2912Mb/s
Cas2.5= ~2909Mb/s
Cas 3 = ~2900Mb/s

This is single channel, On a DFI Infinity and 256Mb Pc4000 Adata.

Whats the big deal with Cas 2 Vs cas3? :confused:

[CPL.Luke]

thankyou for prooving that timings do not make much difference

[Hardin]

Hmm whatever happened to RDRAM?

[dippy_skoodlez]

Hmm whatever happened to RDRAM?


It got slow :p

[Bios24]

WOW. Anyone want to back this up? Does timing really make no difference at all???

[dippy_skoodlez]

WOW. Anyone want to back this up? Does timing really make no difference at all???


I have screenies if you want em... one at each FSB and each CAS. All run CAS-3-3-11 single channel on my infinity.

But, it has a crooked socket, and is throwing my SLK way off. RMA time. :bang head

I may be able to stick my WD in this PC and pull the screenies off if you want me to.

I dont see how BH-5 is worth 3x the price of my Adata.... :confused:

[julianba]

Cas isn't as important as trcd and tras in memory bandwidth.



Dippy_skoodles, any chance you could test cas-3-3-11 versus cas-4-4-11? That might show more of a difference in bandwidth.

[TheGame240]

From what I've always understood, the real difference in CAS timings is in access time, not bandwidth. In other words, your findings are exactly as they should be. However if you used a different test (something that accessed memory frequently not just big chunks) you would start to see a bigger gap in CAS 2 and CAS 3 settings. Also this is why memory timings don't play as big of a role in PIV systems. Less work per clock cycle, therefore less memory access per clock cycle, less importance in fast (access wise) ram and more imporatance placed on large bandwidth.

[obsolete]

Yes, Cas 2 vs Cas 3 reflects access times. Load windows at Cas 2 & browse around & load windows at Cas 3 & browse around. If you can't notice the difference...... :beer:

[dippy_skoodlez]

Cas isn't as important as trcd and tras in memory bandwidth.



Dippy_skoodles, any chance you could test cas-3-3-11 versus cas-4-4-11? That might show more of a difference in bandwidth.


I can as soon as DFI sends my mobo back :cry:

[Gautam]

Funny how everyone puts so much faith in such a useless benchmark.

The Sandra memory bandwidth benchmark is a memory bandwidth benchmark. All it does is test raw bandwidth. Of course timings won't make a difference there. Try out PIFast, SuperPI, 3DMark01, and you'll see a completely different story.

Although CAS isn't nearly as big a deal as tRCD.

[meionm]

sisandra measures bandwith and it usualles with timings, use super pi or prime 95 benchmark, my super pi test and result
super pi for 1M
280-2.5-4-4-7------36sec
280-2.5-3-4-7------35sec
280-2.5-3-4-6------34sec