[Techie007]

Nice test. I would be curious to see the difference between single-channel and dual-channel on at least 1333 MHz or 1600 MHz if you could add that to the test.

 

[BradleyW]

Quote:

Originally Posted by axizor

I believe your GPU driver uninstall guides were the first guides I've ever looked at on this site, and have referred to them countless times since joining.

(I would usually print out the first post in the thread then place it next to me when I went offline to reinstall/uninstall drivers)

Your excellent work here does not go unnoticed.

That means a lot to me. I appreciate your comment. Thank you very much.

 

[nleksan]

I may have to run some of my own tests again. I had found that, for 1 GPU on SB-E ([email protected]), 2133/2200 9-10-10 gave me the bestrresults, while for IVB ([email protected]) 2400 9-10-10 got the best results, in terms of both avg and min fps.

 

[Blameless]

Quote:

Originally Posted by BradleyW

RAM speeds and timing information:

1333MHz @ CL 8
1600MHz @ CL 8
1866MHz @ CL 9
2133MHz @ CL 10
2400MHz @ CL 10

Would it be possible for you to post more detailed timing information, including secondary and tertiary timings?

 

[HeadlessKnight]

Thanks Bradley. That's a very informative post, always thought that RAM clock results in a negligible increase in performance. The Jump from DDR3-1600 to DDR3-2400 is very decent.

 

[Neb9]

Interesting results, so high clock speed on ram definitely has a small affect on performance.
Good info for computer building in the future.

REP +

PR: I gave the REP + a while ago.

 

[TomcatV]

Nice job Bradley ... as stated there are a ton of variables in ram OC'ing but I find real world results always interesting ... +R

 

[TheReciever]

One of those oldies but goodies thread.

Makes me feel better about the 1600 ram I have in my laptop lol

 

[BradleyW]

Quote:

Originally Posted by TomcatV

Nice job Bradley ... as stated there are a ton of variables in ram OC'ing but I find real world results always interesting ... +R

Quote:

Originally Posted by TheReciever

One of those oldies but goodies thread.

Makes me feel better about the 1600 ram I have in my laptop lol

 

[BoredErica]

I have 1866 ram which I clocked to 2200 10-11-9-26-177-2T.

I play Oblivion which is a single-threaded CPU bottleneck in a box. I'm assuming HT is useless for this game. Do you think a ram overclock will help?

I haven't thoroughly pushed the timings apart from 2133 10-11-9. Does the 26 and 177 matter? I heard not all ram timings are equal, so CAS is like the biggest performance changer... I'm thinking of upping my ram voltage from 1.65 to 1.7 to see what I can get.

 

[BradleyW]

Quote:

Originally Posted by Darkwizzie

I have 1866 ram which I clocked to 2200 10-11-9-26-177-2T.
I play Oblivion which is a single-threaded CPU bottleneck in a box. I'm assuming HT is useless for this game. Do you think a ram overclock will help?

I haven't thoroughly pushed the timings apart from 2133 10-11-9. Does the 26 and 177 matter? I heard not all ram timings are equal, so CAS is like the biggest performance changer... I'm thinking of upping my ram voltage from 1.65 to 1.7 to see what I can get.

Hello,
I don't really think there is anything more you can do. If the game is single core threaded, the only thing you can try is to overclock the CPU more, to increase single core performance. Or reduce any settings that are CPU intensive. The fast RAM stuff kind of helps when you are using a multi threaded game.

 

[Prophet4NO1]

Its a bit subjective, but I have noticed a bit of a fps jump in my tri sli surround setup with higher speed ram in the same sorts of scenarios. I have used 1866, 2000, 2133, and 2400 speeds. Don't remember al the CL timings off hand. CL9 at 2133 and 2400 on my current kit. When I get back from work in a few weeks I will have to remember to do some actual testing to get some solid numbers to add to this discussion.

 

[BradleyW]

Did you get that testing done?

 

[Prophet4NO1]

Just got home this morning. Going to test tonight if I get a chance.

 

[BradleyW]

Quote:

Originally Posted by Prophet4NO1

Just got home this morning. Going to test tonight if I get a chance.

Happy hunting!

 

[TheReciever]

Funny we have this thread come up since we had another thread saying DDR2 cl5 is better than DDR3 cl9

 

[Crouch]

Wow, this is awesome!

 

[MagicBox]

Can you update your OP with the actual latencies? You can derieve them from the formula Tcas / Clock. Example: 1600MHz with a Tcas of 6 (the first parameter of the timings numbers) gives 3.75 nano seconds.

This is the number what it's about actually. Tcas is the determining performance indicator when we are talking about memory operations that are random, single reads across various locations in memory. This is the bulk of memory access in an executing program.

The bandwith, which you calculate from the clockspeed and channel width is what affects the other time of memory access: sequential bulk access. Like copying textures from system memory to VRAM.

So we have two performance indicators for two specific performance areas.

1) Program execution (variabeles access, random access), highly influenced by Tcas (in nanoseconds, smaller is faster).
2) Bulk operations (like copying), highly influenced by bandwidth.

 

[MagicBox]

Quote:

Originally Posted by TheReciever

Funny we have this thread come up since we had another thread saying DDR2 cl5 is better than DDR3 cl9

Can or can not be true, depends on the clocks as well.

 

[Klocek001]

Before I'm gonna talk about what I observed as far as RAM speed -> min FPS impact....
Quote:

Originally Posted by BradleyW

Thank you for the feedback everyone!

Also, I'm lucky my chip can handle 2400MHz. Most are limited to 1866 on SB-E, some are limited to 2133, I'm limited a little higher than that. 2666 is a no go on my system.

I just got Predator X 2666MHz. The PC runs fine, except both XMP profiles,which are actually shown correctly in UEFI, kinda don't work. Both 2400 and 2666 CL 11 profiles set my RAM speed to 2133 CL9. I thought it was some kind of UEFI problem (although I picked my mem sticks from the official supported list). Are you suggesting that Sandy Bridge chips can actually limit RAM speed ? I ain't found nothing like this on the net, although I've been looking...

OK, back to the results.

Setup1: i5 2500k @4,72GHz, 8GB DDR3 @2240MHz CL9 in dual channel mode, R9 290 Tri-X.
Setup2: the same, except for @1066 CL7 dual channel DDR3 memory
Far Cry 3, Ultra preset, 1080p, mission: deepthroat (complete), run twice to ensure the results are precise.

Run1:
@2240 CL9 min. FPS - 57
@1066 CL7 min. FPS - 43

Run2:
@2240 CL9 min. FPS - 53
@1066 CL7 min. FPS - 42

Quite a difference, right ?
The -4 fps difference between run1 and run2 might've been more of a one-sec HDD freeze, most of the time the game is locked at 60fps, no drops at all.