1:1 or higher clock speed?

Spartacus

Spartacus

Spartacus

Associate
Joined
28 Oct 2005
Posts
733
Ok guys, what would be best performance wise, I'll give 2 examples...

pc6400 (800mhz) ram running at 711mhz but 1:1

or change the ratio and run the ram at 890mhz. (both with timing of 4-4-4-12)

Clock speed over 1:1 here, im not too up on whats better about running 1:1, hope someone can educate me here hehe

Im asking becuase I think I'll have problems getting my motherboard to go to 400fsb, so no 800mhz 1:1 :(
 
Last edited:
Well 890mhz 4-4-4-12 is going to be a little faster than 711mhz 4-4-4-12. But not so fast that you'd actually notice it outside benchmarks.

Have you tried 3-3-3-10 timings at 711mhz? The Ballistix should be able to do it.
 
The simple answer is this:

Push each combination (1:1 ratio and max clockspeed) to its limits, and then run an appropriate suite of benchmarks. Ie game benchmarks if you're into gaming etc.

I suspect that for most applications the higher clockspeed option will be the faster. Although, I imagine for super-pi etc the lower frequency setting with tighter timings will give better results. Bear in mind that the c2d chips are not nearly as sensitive to synchronous memory settings as the A64s were.
 
So really its gunna come down to whats best out of 711mhz at possibly 3-3-3-10 timings (havnt tried this yet) and 1:1, ....or 4-4-4-12 on the higher clock speed, but not 1:1.

How important is 1:1 for the systems performance? People seem to like running there ram like this.
 
1:1 isn't important.

It's just a case of benchmarking to find out which setting is the fastest, tho the difference will be minor.
 
Spartacus said:
So really its gunna come down to whats best out of 711mhz at possibly 3-3-3-10 timings (havnt tried this yet) and 1:1, ....or 4-4-4-12 on the higher clock speed, but not 1:1.

How important is 1:1 for the systems performance? People seem to like running there ram like this.
Click to expand...

Just give them both a try, and find out. It's going to be highly system- and program- dependent.

There are too many variables to say for sure. Like I said, you need to check on the specific applications you're running. When you get down to this level of fine-tuning (particularly with memory bandwith and latencies) different applications may benefit from the different settings. Either way, the performance difference will be small.
 
What exactly is: Row refresh Cycle time (tRFC)? ...this seems quite high 49T here.

Ive set timing 4-4-4-12 @890mhz as this gives the best benchmarks here , memorys latency in everest is very good now ..52.3ns (havnt tested stability yet though)
 
Last edited:
Forget timings, set em to 5-5-5-15 and clock the nuts off em. Maybe look at reducing timings once you've found a nice stable fsb.

Core 2's just love faster frequencies, and are'nt really bothered about timings.
 
The thing is though, the PC is mainly used for running a computer chess engine, and Ive been advised (by someone whos tested this) that the engine im using seems to prefer timings over clock speed ...were ram is concerned.
 
What kind of system have they tested it in? If it's an AMD system then what they say is correct. But Intel systems tend to prefer high mhz over low timings.

But again just go for the one that gives the best benchmark results from a number of different tests. Try the Fritz Chess Benchmark.
 
The tests were done on an Intel Quadcore @4.2ghz

This is what I was advised about the chess engine im using...

"Tested different RAM configs on my Asus P5K del. with QX9650 @4.2 GHz. I use 2 GB of Cellshock PC 8000 4-4-4-12.
Best setting for that RAM is 4-4-4-11 @ 1066 MHz. It is faster than 5-5-5-15 @ 1200 MHz.
So latency is more important than MHz.
You should only use 2 memory modules, as with 4 speed goes down."


Anyone know if this is important.. Row refresh Cycle time (tRFC) 49T

Fritz benchmark @890mhz 4-4-4-12 is 8951
Fritz Benchmark @711mhz 4-4-4-12 (1:1) was 8778
 
Last edited:
tRFC usually doesn't need to be changed. It has minimal effect on speed, but can affect stability if changed.
 
Thanks Cob, i'll leave that alone then hehe.

One strange thing, I tried 711mhz (1:1) with 3-3-3-10 and on everest the benchmark scores were slightly worse than with 4-4-4-12 @711mhz.

890mhz with 4-4-4-12 was much better though on the everest benchmark and better on the Fritz bench, so think i'll leave it at that for now. Just need to run orthos again now ive upped the ram from 711 to 890.
 
Spartacus said:
The thing is though, the PC is mainly used for running a computer chess engine, and Ive been advised (by someone whos tested this) that the engine im using seems to prefer timings over clock speed ...were ram is concerned.
Click to expand...

How much horsepower does a game of chess need??? :D
 
You must log in or register to reply here.
Back
Top Bottom