Ket

generally no, its just synthetic marks that state a huge gap

Ade1965

Another good thread in the Anand Tech forums in April talked about this same issue in relation to the A64's.
http://forums.anandtech.com/messagev...VIEWTMP=Linear
Their findings suggested that in "real world" test the difference between the two settings was minor and could be overcome with the ability to up your MHz more with 2T.
This means that the A64's can be overclocked pretty well with value RAM. You will never squeeze the max out of your system but with the cheaper RAM you will have more to spend on other items, like a better graphics card.

ncsa

An interesting thread too, thxs. As we found the real thing to achieve with the AMD CPUs is Mhz and if possible at High FSB's giving the bandwidth. Which makes running 2T OK as you will really not notice the difference esp. if this allows you a higher stable OC .. this is got to be a good thing.

Ade1965

Agreed
I just wish I knew these findings when I built my system in December. I would have gone for cheaper RAM due to the A64's being less "picky".
On the plus side though I have been able to get my systems HTT much higher so cancelling the effects of the 2T.
Also a separate issue with my motherboard (Asus A8N-SLI Deluxe) means that the “bug” which prevents the HTT going over 245/250 when the command rate is set to 1T is not so important.

ncsa

Yes I moved off the A8N-SLi as I was unable to get the OC I was looking for but will be rebuilding this soon and it will be a nice back office workhorse but still plays games ok when needed to. There is meant to be a modded BIOS for the Asus that supports higher Vdimm (>3.0v) which would also help allow more memory options such as UTT/BH5

Ket

something interesting to note, everest consistently gives me scores i get with 1T and sandra reports around an 980MB\s difference, my thoughts on this? none yet. it appears as though 1T is only really important in skt A systems

ed- ran 3dmark01SE and scores plummeted to 23.4k with the same OC settings, thats around a 3000pts drop on my normal scores. i also have some system maintinence to do so ill run the tests again, but as of this moment, the 2T results are quite self explanatory.

ncsa

One of the things that made me curious and hence running some tests was that all the various reviews around the place quote these tools and while it allows comparison between components it does not reflect the reality of the system operation.

If you compare some BW benchmarks between different CPUs say a 3500 Winnie and an FX55 with the same HW & mem modules you will find that the Winnie will show a higher BW for the same settings... this does not make any sense and which is why the BW tools must be taken with a grain of salt.

1T or 2T only has a small real difference and if you can get a higher stable OC with 2T then that will be better for any system - more Mhz's got to be a good thing.

It would be good to see if there is a real difference for the Skt A, so why not try and run some controlled tests and compare here.

Ket

after some maintinence on the system the results remain, 3dm01SE still reports the same, interestingly though the system test FPS are the same, but the mother nature test drops around 140FPS when at the water part.....unsure why, but its there. ill test things with doom and 3dm05 tomorrow. AM3 reports 74k vs 76.5k so not a huge difference there, super PI records a 3 second lag however, so 2T isnt what media creators want to consider using.

ncsa

Correct, so now for me if I can find another 7-10% increase in Mhz running 2T and is stable then it will be worth it. Afterall AMD likes Mhz and this inturn results in more FPS or crunching a file or F@H....

If you get a series of results for the Skt-A post them up and I will link then in from the first post so it is easier for others to find.

Ket

i'll be able to get them soon hopefully, i'm building a pretty meaty skt A system for a mate. on running 3dm05 the result was a score of 5307, this seems low for my system but for some reason its always been kinda low on my system, last time i ran 05 with 1T timings i got a score of around 5500, the key aspect to investigate now is if switching to 2T and lowering the multiplier while upping the "FSB" results in anything posative, but at the same time for me it may not as BH-5 has always topped out @ 250.