The truth about Tras and its dangers....

[Poisonsnak]

I ran across that very mushkin article about a month ago and brought up this exact same issue, only we didn't have 4 pages of posts on it hehe. THunDA pointed out he had run a tRAS of 00 before and believe me he knows what he is doing.

 

But if you read the article it states:

Any tRAS setting lower tRCD + CAS + 2 cycles will allow the memory controller to close the page “in your face!” over and again and that will cause a performance hit because of a truncated transfer that needs to be repeated. Along with those hassles comes the self-explanatory risk for data corruption. That one is not a real problem as long as the system is kept running but in case it is shut down and the memory content is written back to the hard disk drive, the consequences can be catastrophic. For the drive, that is.

 

So this poses a data corruption risk only when you shut down according to mushkin. Not only that, you have to have corrupt data in RAM when you shut down, this is just pretty much guaranteed by the low tRAS, again according to them.

 

But if you can memtest and prime 100% stable then how can the data in the RAM be corrupt in the first place?

 

edit: reference link: http://www.dfi-street.com/forum/showthread...ption#post76633

[Guest shaolin95]

Once again, i justed posted the link...now its your call.

[Sharp]

Hello,

 

I think it is more about memory efficiency.

 

Allowing the memory to seach for a specific time interval before closing the search and starting a new one.

 

During the search it may find some DATA or it may not.

If it does and during the read/write the link is closed the data sent to the memory buffer is lost/cut. (That is the corruption).

 

Usually when this happens the memory controller will try again and as it knows where to look this time and the small tRAS time may be enough.

 

And that may be why you get no errors on memtest, as it will eventually get the data.

 

But the small tRAS will effect Bandwidth/efficiency, if you have to repeat the same search so many times it will kill your data rate.

 

If you use a program like RMMA, you may be able to see the effects.

Test 4 5 6 and watch the graph, compare your scores at the end.

http://www.dfi-street.com/forum/showthread.php?t=8823

[Mastakilla]

intresting post!!

 

ill try to increase mine from 7 to 8 and then do a low level format + winxp reinstall...

 

i REALLY hope that it solves my unexplainable, and reproducable blue screen i get now and then

 

first it was the daemon tools driver causing the blue screen (d346bus.sys or something), after uninstalling it, i now get a blue screen caused by ntkrnlpa.exe

 

my system is 100% stable (hours occt, prime, memtest, ALOT of games, pifast, superpi, sisoft, 3dmark, other 3d demos, etc ,etc)

none of those can reproduce the blue screen

 

it happens while copying from usb while listening to music fe, or viewing a movie, or installing something with remote desktop connection, etc, etc, really random ....

[murmeli]

tras 5 is the old bh5 type memory prefered for older platforms, using any lower decreases bandwith.

 

Also using cas 1,5 vs 2,0 is nearly equal bandwith, no point using 1,5

 

Some other mem settings like trrd >2, trc >7, trc >12 cause problems even with good bh5

[Guest JaY]

Ran the quick tests with RMMA, the program sharp recommended. Here are my results. All tests were done with 1.5-2-2-x, x being tras, the only number that changed.

 

Looks as if tras 5 is the fastest overall for me, the bandwith dropping due a tras setting lower than the formula doesn't seem to be true.

[Guest shaolin95]

The formula aint about bandwidth dropping please re-read....

[Guest JaY]

Well, bandwith would drop if it had to go back and get the data again over and over.

 

edit: to explain better

 

"Any tRAS setting lower tRCD + CAS + 2 cycles will allow the memory controller to close the page “in your face!” over and again and that will cause a performance hit because of a truncated transfer that needs to be repeated. Along with those hassles comes the self-explanatory risk for data corruption. That one is not a real problem as long as the system is kept running but in case it is shut down and the memory content is written back to the hard disk drive, the consequences can be catastrophic. For the drive, that is."

 

I didn't notice a performance hit from tras being lower than trcd + cas +2, if what they're first saying isnt even valid about a truncated transfer being repeated/lowering performance, how can i expect the second to be valid. Especially when it still passes tests to check for any errors in calculations which would be data corruption.

[Guest shaolin95]

The way I see it is like this, at some extremely thight timings I can SuperPi 1mb and get my best times....then I try 32MB and I get a worst time than I do a more relax timings if I can pass the test at all...I think thats what you are experiencing. To me is pretty simple and the explanation is clear. It will happen sooner or later, remember when your mom use to say dont play with fire or you will get burn? Same thing...it may not happen today, tomorrow in 4 months but as long as that "leak" is there the time bomb is ticking....boy I am very dramatic today :nod:

[Ubermann]

Yeah and after 5 month computer crash..do you then reboot it or do you start tearing your hair off and get suicidal thoughts that you can never again press the reset button cuz some strange force from some dark side of some planet keeps you from putting your tiny starved little finger from reaching that little button thats now has 30 000 fangs round it and whispers -everything floats down here- ? :drool:

[Guest shaolin95]

Yeah and after 5 month computer crash..do you then reboot it or do you start tearing your hair off and get suicidal thoughts that you can never again press the reset button cuz some strange force from some dark side of some planet keeps you from putting your tiny starved little finger from reaching that little button thats now has 30 000 fangs round it and whispers -everything floats down here- ? :drool:

 

Dude is your brain on? Do you even have a remote idea of what we are talking about? How do you know that when it crashes it will be as simple as reboot to fix? BTW, your attempt at comedy is on par to a 1541 drive's reliability... atrocious! You know whats really scary...that your post was actually edited so who knows how much dumber it actually was in its original form....the humanity! :confused:

[GNU]

Very interesting posts. I see the performance for the drive is better with a low tras like 2,3 or 4. I have 2.5-4-4-8 without any problems. It's bad for my drive?

I have no idea zu set tras lower than 7 (8 is still testing) because my system crashes.