Memory Performance Change from 590i to 650i

[PimpYo]

Yesterday I had a Foxconn 590i motherboard and an AMD X2 4000+ OCd from 2ghz to 2.66Ghz. Vista 64bit Experience index of 5.9 for my Corsair 2GB 8500C5 memory.

 

Now I am using the ASUS P5N-E SLI 650i motherboard and a Q6700, with the same Corsair 2GB 8500C5 memory, but Vista 64bit only rates the memiry at 5.3 now.

 

Is there that much of a performance loss going from 590i to 650i, or is the ASUS not setting up my memory correctly?

 

ASUS board is set to run Linked 1:1 1066mhz for FSB and Memory.

 

5-5-5-15-21-2T @ 2.178V (21 is tRC not tRRD)

[RJLeong65]

Yesterday I had a Foxconn 590i motherboard and an AMD X2 4000+ OCd from 2ghz to 2.66Ghz. Vista 64bit Experience index of 5.9 for my Corsair 2GB 8500C5 memory.

 

Now I am using the ASUS P5N-E SLI 650i motherboard and a Q6700, with the same Corsair 2GB 8500C5 memory, but Vista 64bit only rates the memiry at 5.3 now.

 

Is there that much of a performance loss going from 590i to 650i, or is the ASUS not setting up my memory correctly?

 

ASUS board is set to run Linked 1:1 1066mhz for FSB and Memory.

 

5-5-5-15-21-2T @ 2.178V

 

This is because the Intel processors don't have a memory controller integrated onto the CPU die. All Intel processor-based systems use a separate bus for the memory controller, which is located on the motherboard. And the interconnecting bus to the on-motherboard memory controller will never be as efficient as a memory controller that's integrated on the processor die.

[PimpYo]

I had an intel DG33TL mobo with this same Q6700 for about 2 days and it registered this memory as a 5.9 in vista.

[PimpYo]

I have made some changes.

 

I have the CPU OCd to 3.25Ghz with a 1300FSB, and a 5:4 ratio so my memory runs at 1040mhz.

 

5-5-5-15-21-2T

 

and I also changed my twTR from Auto to 15.

 

now my memory shows as a 5.7

 

I am going to try 5-5-5-14-20-2T twTR 12 and see what I come up with.

[PimpYo]

5-5-5-14-20-2T twTR 12 is a 5.8

 

now to try 5-4-4-14-20-2T twTR 10...

[PimpYo]

Success! 5.9!

 

Wow, what a difference this has made with nothing but tweaking ram settings.

 

ASUS P5N-E SLI 1st setup my memory as 5-5-5-18 (and the rest auto), 1.8V 800Mhz. and a vista rating of 4.8

 

Tweaking it to 5-4-4-14-20-2T and twTR 10, 2.1V 1040mhz has bumped me all the way to a 5.9

 

oh, and to clarify if someone is reading this 100 years from now, the 20 in 5-4-4-14-20-2T is my tRC not tRRD! I left tRRD on Auto.

[Xtreeme]

So....what your saying is you got it = only by tweaking the timings like no tommorow? Thats good, but that didnt really change anything-if you ran that on the other board it would have higher index than 5.9 then....the diff is the MC at = timings and mhz the ram should have lower latency on athlon due to the IMC. The intel northbridge will add latency no way around it. You can use everest to verify it by checking the mem latency benchmarks. Look at how good a little am2 4000+ scores. Second from best in list heh. :):

[RJLeong65]

So....what your saying is you got it = only by tweaking the timings like no tommorow? Thats good, but that didnt really change anything-if you ran that on the other board it would have higher index than 5.9 then....the diff is the MC at = timings and mhz the ram should have lower latency on athlon due to the IMC. The intel northbridge will add latency no way around it. You can use everest to verify it by checking the mem latency benchmarks. Look at how good a little am2 4000+ scores. Second from best in list heh. :):

 

Not only that, but I got a 5.8 to 5.9 for the WEI memory score on an x2 3800+ Socket 939 system with 2GB of DDR400 2-3-2-6 rated memory in a dual-channel configuration...

[Yellowbeard]

Does anyone know:

 

A: What parameters Vista bases this memory score on?

B: If this score has any effect on real world performance?

 

I am not sold on this seemingly random Vista number. :confused:

[Xtreeme]

Does anyone know:

 

A: What parameters Vista bases this memory score on?

B: If this score has any effect on real world performance?

 

I am not sold on this seemingly random Vista number. :confused:

 

A: I have no clue! heh.

B: I imagine like most synthetic benches, the only "real" thing is a higher number is better! heh.

[RAM GUY]

Well all that would be great if it were a bench mark, but its not; I would suggest looking at something like Sy-Soft Sandra or Everest Home to bench the system.

[Xtreeme]

Ah, good point. I did overlook the obvious there, its a rating system not a benchmark (it doesnt test anything) so its literally just M$ opinion not really any basis other then what they rate x hardware as. LOL. I use everest, have since it came out (amd forums). Seems to give more info.:):

[RAM GUY]

Exactly and to be honest its hard to give any merit to anything that is not well defined no matter who its from. No Offense Mr Gates Sike! :eek::D:

[Yellowbeard]

Good point RAM GUY, I am certainly not criticizing them per se. I just don't know what that number means. So, it's very difficult to know if their rating has any real benefit.