SOLVED 1T vs 2T Command Rate and Timings (Results in Post #8)

[Tyerker]

Does it make sense to anyone else that running something at 9-9-9-25 timings with a 2T Command Rate is every now and then wasting time / clocks? Sorry if it's not that simple, but it seems with a 2T Command Rate wouldn't it be better to run your RAM with even Timings so everything synchronizes? Conversely, wouldn't it also be better to try 1T Command Rate with timings like 9-9-9-25?

 

[Woomack]

You can try CR1 but AMD don't really like it ( depends from other settings and IMC ). I doubt that you will see any difference in performance. CR1 is of course faster than CR2 but not that much and in most cases on new platforms there is almost no difference.
Synchronizing timings works for some other timings but on each platform or memory IC it works a bit different so I won't tell you now how to make it. It's also not that big difference in performance and you can see it mainly on higher clocks ( sometimes there is no difference at all ). It helps more for stability after oc than performance.

 

[Tyerker]

So would I be better off setting timings to like 10-10-10-30 to get maximum Speed on my RAM to have it all sync up? First time I've really tweaked RAM since my old DFI Ultra-D / Opteron 165 build from years back.

 

[Janus67]

The lower the timings the better. For all intents and purposes 2t basically adds 1 to the primary timings so 9-9-9 2t would be about equal to 10-10-10 1t

 

[Tyerker]

So it just makes every timing 1 slower? I thought the Command Rate was more or less how large of intervals the RAM actually performed its job at.

If AMD functions better at 2T I suppose I'll leave it at that, but the SPD Setting has it performing at 9-10-9-24 1866 MHz at 1T. It is Intel-Certified RAM (designed for Z77 and all that) so is that 1T timing just meant for Intel OC settings?

 

[Woomack]

Check if all is running fine at 1T , when it will be fine then keep it .
Also main question that haven't been asked yet ... what memory you have ?

 

[Tyerker]

In the sig. 4x4GB Patriot Gamer 2 Div 4.

Newegg Link: http://www.newegg.com/Product/Product.aspx?Item=N82E16820220650

Stock Settings are 9-11-9-27 DDR1866 @ 1.5V. It apparently runs stock settings at 1.65V, so I'm working with that as a baseline. Currently running it at 1200 MHz with 7-7-7-20 @ 1.5V with this old Rana CPU so I don't blow it's IMC while I try to find the max FSB.

I'll eventually unlock it to the Phenom II x4 it can be detected as when unlocked. Just exploring all this chip can do for the last week or so that I have it until my FX-4100 comes in. Benching the whole way and seeing what the results show.

 

[Tyerker]

I'm about to test 1T vs 2T with my latest OC:

240 FSB x 13.5 = 3240 MHz Athlon II x3 425
2400 HT Link
RAM @ 1280MHz 7-7-7-20

MaxMemm Unganged 2T
12141 MB/s Copy
8758 MB/s Read
8202 MB/s Write
Reached Memory Score = 8.48 GB/s
Latency = 51.2 ns

MaxMemm Ganged 2T
9552 MB/s Copy
9137 MB/s Read
8068 MB/s Write
Reached Memory Score = 8.60 GB/s
Latency = 50.1 ns

MaxMemm Unganged 1T
12049 MB/s Copy
9263 MB/s Read
8569 MB/s Write
Reached Memory Score = 8.98 GB/s
Latency = 49.0 ns

MaxMemm Ganged 1T
9552 MB/s Copy
9361 MB/s Read
8444 MB/s Write
Reached Memory Score = 8.90 GB/s
Latency = 49.1 ns

So here's the conclusions I came to:

1.) Unganged RAM shows a significant improvement in Copy speeds
2.) Ganged RAM Has slightly better Read/Write speeds
3.) 1T Gives Slight overall performance advantages

I have pretty much decided that, for me, I'll use the Unganged 1T setting. I may clock a bit higher (250-265 FSB-ish range) but I am liking the performance I'm getting with an Athlon.

Side note, when I Enabled the Core unlock and my CPU was recognized as a Phenom II, most memory scores when down significantly, even with the 4th core disabled (recognized as a Phenom II x3). Any thoughts as to how it could perform worse as a Phenom II from a memory standpoint?

 

[OptyTrooper]

I'm about to test 1T vs 2T with my latest OC:

240 FSB x 13.5 = 3240 MHz Athlon II x3 425
2400 HT Link
RAM @ 1280MHz 7-7-7-20

MaxMemm Unganged 2T
12141 MB/s Copy
8758 MB/s Read
8202 MB/s Write
Reached Memory Score = 8.48 GB/s
Latency = 51.2 ns

MaxMemm Ganged 2T
9552 MB/s Copy
9137 MB/s Read
8068 MB/s Write
Reached Memory Score = 8.60 GB/s
Latency = 50.1 ns

MaxMemm Unganged 1T
12049 MB/s Copy
9263 MB/s Read
8569 MB/s Write
Reached Memory Score = 8.98 GB/s
Latency = 49.0 ns

MaxMemm Ganged 1T
9552 MB/s Copy
9361 MB/s Read
8444 MB/s Write
Reached Memory Score = 8.90 GB/s
Latency = 49.1 ns

So here's the conclusions I came to:

1.) Unganged RAM shows a significant improvement in Copy speeds
2.) Ganged RAM Has slightly better Read/Write speeds
3.) 1T Gives Slight overall performance advantages

I have pretty much decided that, for me, I'll use the Unganged 1T setting. I may clock a bit higher (250-265 FSB-ish range) but I am liking the performance I'm getting with an Athlon.

Side note, when I Enabled the Core unlock and my CPU was recognized as a Phenom II, most memory scores when down significantly, even with the 4th core disabled (recognized as a Phenom II x3). Any thoughts as to how it could perform worse as a Phenom II from a memory standpoint?

In answer to your last question I imagine the un-lock is not completely stable and the IMC is wonky b/c of it. Just a thought, there are others that know more about AM3 than I.

 

[Woomack]

@Tyerker, just run it @ 1200 or 1333 7-7-7-20 1T and when you move to FX then try higher like 1600 8-8-8-24 / 8-9-8-24 1T or 1866 9-10-9-27 / 9-11-9-27 1T. All these settings will give similar results but when you go higher with CPU-NB then 1866 will be fastest.