• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

ECC vs non-ECC

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

emericanchaos

Member
Joined
Sep 13, 2002
Location
Williamsport, PA
ok what is ECC anyways? error correction? what's the differences between them and which is better (faster). i would think non-ECC was if ECC is error correction. if it doesn't stop to fix itself it should move along faster?

isn't this what EDO was. error correction for fast page memory?

i'm asking because i've been browsin around and i found a site that gives me a choice and i'm stumped.

here i'll do this for easier replies since i ask lots of questions at once.

1.

2.

3.
 
EDO was Extended Data Output - it came in three speeds, 70, 60, and 50ns to my knowledge, I think it change the way that memory is accessed which made it faster than FPM, I believe FPM had to flush out whatever was currently there before moving onto the next instruction - where EDO could just use available space.

FPM or Fast Page Mode - was pretty much what I described towards the end but vice versa, I think it came in 70 and 60ns and basically once it had done a task - had to stop and remove all that was stored which was a bit of a slow down...

ECC Error Checking Code sends more bits for checking - where parity would send 1bit this sends 5 for 8bit instructions, 7 for 32bit Instructions and 8 for 64bit instructions...it apparently downs performance by 3% - but I have never had any to find out :(

on the stopping to fix themselves I dont know - I know with parity it will simply start over - I think it might do the same thing with ECC
 
Yeah you've got it right, basically ECC gives you a little performance hit, but it much more stable in it's performance level and can handle heavy memory load more easily, which is why it's usually only used in server applications
 
Parity-checking memory will detect, but not fix, single-bit errors in any given memory section. ECC memory will detect and fix single-bit errors, and will detect, but not fix, two-bit errors in a memory section.
 
of course, ECC does you no good if your board don't support ECC RAM.
in some (most) cases, it'll be seen as regular PC-2100 or whatever speed/type just non-ECC...
the RAM is still (usually) compatible, follow me?

but in some rare *other* cases, such as certain Shuttle AK31 revisions, the board rejects ECC *type* RAM altogether.

basicly it costs about 30% extra for ECC (approx.)
if you have a ECC supporting board,
and you're mainly concerned with stability then it's worth it.
it isn't great at high FSB and it isn't geared for performance.

if you have a plain DDR mobo and want to overclock,
then spend the extra money on good, faster Corsair XMS RAM.
the extra money is better spent on speed,
than on features which wont work,
and overall less performance than equivelent speed non-ECC RAM
(if based on overclock ability alone)
 
when you run in ECC mode, it's slower due to the time it takes to read and correct errors...
naturally there will be some delay there.

in non-ECC mode, there should'nt be any delay, and it should run just like any other RAM, altho it may not be as accepting to aggressive timing, higher FSB's, etc...
IMO, Corsair makes RAM for that purpose.

my Crucial ECC PC2100 does about 150FSB no probs, and that's fine by me.
have i pushed it further or really tried to get more? no need to really.

as i said, it isn't designed for performance or overclocking,
it's purpose is to be more reliable, to stop errors from occuring,
which in a server environment, is rather important.

in your EPoX nforce2 mobo (etc...) you're looking for speed, O/C, and performance, so obviously the extra money is better spent on hi-perf. RAM instead of ECC.

as i said, Corsair XMS series is designed For that purpose,
and will make you happier in the long run i'm sure.
 
Last edited:
Back