Ram ns timing question

[Uranium]

Ok. I currently have a 256M PC133 6ns stick in my machine. I've been running the ns timing setting (in my bios) for that bank at turbo with no instability problems. The options are 10ns, 8ns, Normal, fast, and turbo. I ordered a 256M pc133, 7.5ns stick last night. If I set it to turbo also, could it cause instability problems? If I set it to fast or normal would it slow my first(6ns) stick down any in access timings? Does anyone here know the ns timing values for normal, fast, and turbo, I can't find them ANYWHERE on the net!

 

[Guest]

I would expect that information to be in the mother board manual. Have you looked there?

 

[Uranium]

Thiers no specific info on what would happen if I set it to higher. Or what values Turbo, Fast, or Normal are

 

[Guest]

That is really lame. Assuming the selections do change something, it would be helpful to know what it is they are changing. Then one could make an educated decision as to whether or not this setting should be messed with. That is very poor documentation, and would piss me off.

My next best advice is contacting tech support for the motherboard manufacturer. Someone down there better know what is up. If it were me I wouldn't give up until I talked to someone that knows what they are talking about.

 

[bum_jcrules]

I hate it when it posts twice. First they update the community and then it double posts. I need a weekend.<P ID="edit"><FONT SIZE=-1><EM>Edited by Bum_JCRules on 01/04/02 03:57 PM.</EM></FONT></P>

 

[bum_jcrules]

They are either doing one of two things or both, I think, most likely.

PC133 means, it is running at 133MHz. 133MHz translated is 133,000,000 cycles per second. So if there are 1,000 nanoseconds in one second. You come up with 1000/133 ~ 7.519ns.

They could be changing the latency settings to have a better "read" cycle timing. (8ns vs. 10ns, where 8ns is bursted) PC133 is at minimum 7.5ns and when you add latencies it can operate at that higher figure of 10ns.

In SDRAM, there are 5 timings:

1. The time required switching internal banks (tRP);

2. The time required between /RAS and /CAS access (tRCD); (Hence CAS; Column Access Strobe and RAS; Row Access Strobe.)

3. The amount of time necessary to "prepare" for the next output in burst mode (tAC);

4. The column access time (tCAC);

5. The time required making data ready by the next clock cycle in burst mode (read cycle time).

So if they burst the PC133 the time to complete a cycle will be less but it will still be subject to latencies.

You can overcome higher latencies by overclocking the memory slightly. So instead of running at 133MHz you could run at say it at 142MHz; here the time to complete one cycle is 7.042ns minimally. Add in the extra latencies and it is around 8ns-8.5n.

So there is my theory. They are either just bursting the "read" cycle and/or adjusting the timing settings or they are overclocking the module. They could be doing both.

Back to the peanut gallery...

They are either doing one of two things or both, I think, most likely.

1. They could be changing the latency settings to have a better "read" cycle timing. (8ns vs. 10ns, where 8ns is bursted) PC133 is at minimum 7.5ns and when you add latencies it will opperate at a higher figure. But that is what PC133 means, it is running at 133MHz. 133MHz translates 133,000,000 cycles per second. So if there are 1,000 ns in one second. You come up with 1000/133 ~ 7.519. So if they burst the PC133 the latencies will be less but it will still be subject to latencies but they can overcome that by overclocking the memory slightly.

So that is the second thing they could be doing; OC the memory as well.

So there is my theory. They are either just bursting the "read" cycle of the or overclocking the module. Or they could be doing both.

<b>"The events of my life are quite inconsequential.." - Dr. Evil</b> :lol:

 

[Guest]

Well thanks for stepping in. I was clearly flailing.

I haven't worked with S/DRAM before, so that stuff is new to me, and very informative.

It makes sense that bursting the read cycle might be given some special name, but calling overclocking the whole module something else seems confusing (why not just show the various clock speed choices?).

I stand by when I said that this should be described in the manual. Even if your description is perfect how would anyone verify it?

PS OT question: Keeping all those timings straight sounds as if it could be confusing. Is this something handled by the mem controller or something which must be dealt with and kept straight by the programmer, on some level?

 

[bum_jcrules]

Like I said it is only my theory. Calling or e-mailing the techs at the motherboard manufacturer is you only way you will ever know the truth for you situation.

Memory controller is what handles the synchronization of the CPU to the Memory. In most cases run at different speeds and on differing signals. Timing settings can be set in the BIOS like CAS settings. So all settings could be in software languages instead of assembly language, machine language that is.

<b>"The events of my life are quite inconsequential.." - Dr. Evil</b> :lol:

 

[Uranium]

well I got the stick and it works fine @ turbo in my bios. It does run hotter then my 6ns stick as I suspected it would, but my 120cm fan in the side helps cool it down

I tried turning up the cas to 2 once, thing wouldent boot, I couldent find the reset jumper too, I had to flip the top contact off the batterey for a second to get the thing to post