OK, I was previously running 2x256Mb at 2-2-2-5 timings. Yesterday, I installed 2x1Gb Crucial Ballistix PC4000 monsters. First boot, I ran it at 3-3-3-8 timings, everything else identical to before. Played LFS, and :eye-poppi, 50fps. But wait, it was at 60fps before (OK, +/- 2 FPS, average, 60 ) So anyway, today I return from work, put the RAM at 2-2-2-5 timings, and sure enough, the same 60fps (average) as before.
10FPS seems a lot. As in, ~ 16.5% drop. Ah well.


Just thought i'd let you all know, as there's often threads about maxing FPS :up:

Quote from letdown427 :

OK, I was previously running 2x256Mb at 2-2-2-5 timings. Yesterday, I installed 2x1Gb Crucial Ballistix PC4000 monsters. First boot, I ran it at 3-3-3-8 timings

Hang on....you put what I presume to be faster RAM in, and slowed down the timings? illepall

Nice report though :up:

Yeah , i had 1GB CL2.5 3-3-6 2T. Later i set them to CL2 2-3-5 1T in the bios. And had a boost by about 10Fps (LFSbench max.qual. max. FPS CL2.5 92fps, CL2 104fps). But now i got 2GB CL3 3-3-6 ! and just had a minor performance loss (LFS Bench max. fps=101 FPS) and i can't explain why ?!

What kind of CPU do you have?

Just to explain my apparent stupidity, that RAM is quoted to run at 3-3-3-8, and so for the first boot i just wanted to check that it actually worked so put everythin at stock and just ran a few programs.

@M.Mos, maybe it's the last number that has the biggest effect then? as mine dropped from 5 to 8, and yours now only dropped from 5 to 6?

I have a 4400+ @ 4800+ clock speed, if you have an athlon or some such, they're apparently affected a lot by 1T / 2T, although there's a big thread on anandtech competely disproving that with lots of evidence, so that's probably worth ignoring.

[sarcasm]What an amazing discovery. Your game stays the same with the same settings/equal hardware as before.[/sarcasm]

most games get a boost in fps from tightening up ram timings , however some modules will run at 2-2-2-5 and some do not ,also the command rate can give a boost when lowering to 1 from 2 but again not all ram works like this ,there is often a "sweet spot" to be found not allways the lowest, its the same for overclocking , better quality ram = better timings at greater speed, I have allways found Mushkin ram (the level 2 BH5 stuff ) works at the tightest timings and overclocks great without having to increase timings.

WARNING!!! Changing RAM timings can mess up your PC bigtime unless you are confident with this and know the risks please dont try it..

SD.

should this thread be under "hardware"?

Don't worry, i'm English, I can figure out when sarcasm is implied, no need for the [sarcasm]...[/sarcasm]

The 'discovery' was more about just how much it dropped because of the RAM timings, even with 4times as much memory, and I apologise if my exclamation mark gave you the impression I considered it an amazing discovery, I don't, just thought I'd mention it in the General Discussion forum. As for the hardware, name brand high quality RAM of four times the size, isn't equal to the cheap generic rubbish I had in there before.

SD, I didn't realise the boost was so big normally, so thought maybe it was a trait of LFS. My apologies for wasting everyones time.

Whilst I think it's obvious ram timing would effect any game, I didn't think it'd be as much as 10fps. So I'm surprised. I'd have though loading would have been the big winner though - any timings of loading LFS?

Tristan I think LFS takes different amounts of time to load depending on what track area (ie Fern bay or aston ect) was loaded last time you exited it . so I am not sure you could get completely comparable results , I can say for sure HD speed (raid ect) increases load time far more than Ram timings

SD.

Yeah, the amount of RAM mainly influences loading times and that also only on memory intense games. It can help with the fluidity of the game if there are insane amounts of textures used (BF2 anyone?), so with a too low amount of RAM pagefile shuffling would constantly occur during play.

The RAM timing on the other hand is the thing that changes fps, although you can't really change much of the timing anyways. Setting the timing too low is likely to screw up your PC, or atleast cause severe instability.

As people always say, "it seems quicker"

I don't have any figures for how long it took to load before or after, and I'm not in the mood to go about swapping around timings and such, once I have a full day, I'll probably be doing some more in depth comparisons of the two sets of RAM, and may even try it with both (2.5Gb), although this may force a 2T command rate, which may or may not have a big affect.
[waffle]
I'm still not convinced by the whole '2T is the anti-christ' thing that people often post on overclocking forums. The thread I referred to on anandtech.com was started by one guy(as threads so often are) who did a lot of comparisons and benchmarking purely changing command rate (yes he had an Athlon 64), and he saw so little difference (most was normally around 3%) that his results encouraged many more to try it out, and so they did, and the results seemingly disproved the whole anti-2T thing, but as a hopefully well balanced observer, I have to say that the anti-2T opinion must have started from somewhere, and I presume must have had some good examples to back it up, so yes, I'm still undecided.

[/waffle]

If it's a particularly huge improvement in loading etc, I'll post back in this thread, but anything less than 5-10 seconds = meh in my opinion.

Quote from M.Mos :

and i can't explain why ?!

1t command rate vs 2t ... makes up for quite a few frames as seen in a recent review of a new dfo mobo on anandtech

and for an a64 you should set the rams to 1t with x-x-x-8 or someting else in the 8-10 range

Quote from Shotglass :

1t command rate vs 2t ... makes up for quite a few frames as seen in a recent review of a new dfo mobo on anandtech

and for an a64 you should set the rams to 1t with x-x-x-8 or someting else in the 8-10 range

Yeah sure but, i have higher frames with 2GB CL3 3-3-6 2T then i had with 1GB CL2.5 3-3-6 2T ... explain that

Quote from M.Mos :

Yeah sure but, i have higher frames with 2GB CL3 3-3-6 2T then i had with 1GB CL2.5 3-3-6 2T ... explain that

is there a typo or are really both with 2t command rate ?
cause if they are the double size of your ram should make up for the practically negligable .5 increase in cas

Just remember... sometimes faster isn't better.

For example, though it may seem odd a tras of 11 (what you currently have set to 5) is actually the best for an nForce 2 based system.

Anyway, I'm not surprised by this. Since LFS is so CPU hungry, and of course data hungry and what not. And I'm definately not surprised that you saw a big boost going from 2T to 1T. That change alone is huge.

FYI: Tras 7 is best for A64

Quote from jmkz :

FYI: Tras 7 is best for A64

? dosent that depend on the type of RAM and other factors too?

and Tras 11 for NF2 sys ,I am not sure I completely follow the logic

SD.

Quote from SparkyDave :

? dosent that depend on the type of RAM and other factors too?

nope

nForce2 Tras=11 http://www.madshrimps.be/?acti ... tpage=503&articID=133

nForce3 Tras=10

Quote :

With nForce3 motherboards, the fastest performance on AMD Athlon 64 was at a Cycle Time or tRAS of 10. However, the nForce4 behaves a bit differently with memory. We ran a complete set of Memtest86 benchmarks with only tRAS varied to determine the best tRAS setting for these memories, and found the best bandwidth at tRAS settings ranging from 5 to 8. Therefore, a tRAS setting of 7 was used for testing.

nForce4 Tras=7 http://www.anandtech.com/memory/showdoc.aspx?i=2560&p=4

This post is a good note to people concerned with getting that little extra from their computers. The understanding that not all ram is created equal is another good concept. Some people buy cheap ram, some pricy name brand. Others look at timing specs.

The numbering system is in order of importance, performance wise. The CAS timing has the greatest effect on overall speed. A CAS of 2 will be a good bit faster than a CAS of 2.5. The others fall in order of importance behind CAS. Getting ram that can do 2-2-2-5 costs some cash. At one point, I was considering getting some...for over $300. The timings have a greater impact on performance than bus speed. Buying high bus speed ram and just overclocking the bus and ram way up does provide a benifit. However, the latency timings will create a larger difference.

I've come across the trans thing over on the Nforce forums some time back. The idea's kind of interesting, the lowest isn't always the fastest. However, as far as I'm aware, it only seems to be that one setting.

The only ram I buy now is the one with the lowest CAS Latency. Too bad I can't get CAS 2 for my laptop...

There are memory testers available. They allow you to test memory performance and stability. You can find out if your computer is capable of running at lower settings or not. Either the memory stress tests will fail or your computer won't boot and you have to reset your BIOS.