Xtreem UCCC memory OC to 257MHz :(

[Bladesinger7x]

I just recently exchanged the 1GB Xtreem DDR600 TCCD memory for
2GB Xtreem DDR500 UCCC memory. But when I atempted to overclock
the memory the max was 257Mhz (on a 166 divider). When set to 263Mhz
(HTT322) Super Pi 1M errorrs immediately

I was under the impression that Xtreem memory is premium stuff and yields
quite nice overclocks (I was hoping to achieve round 275Mhz), since one of
my friends managed to OC his generic PQI ram with UCCC chips to ~270Mhz.

Am I using too tight timings...voltage is 2.6v (more voltage doesn't help)?


Also when testing with Memtest (Bios/DOS version) at 263Mhz 1 pass of test
#5 gives 4 errors, but when swithching the sticks it produces like 500 errors??

Maybe a different board would do the trick..I have another Ultra-D board (rev. AD0),
the one in use is AA0 (made in China) with the latest official BIOS.

Update: Well, I just got news that Prime blend test failed after 2hours. The memory
was running at 254.5Mhz (HTT311/divider 166) and relaxed timings (I loaded the
optimized defaults and set 3-4-3-8 manually, all other settings are on auto).
Can cpu voltage set to low affect this...the voltage is 1.4125v (1.25vx113%)?

[Bladesinger7x]

Are these good timings in try to determine if the memory is
stable:

[Clint]

Oh well...

[Xtreem Tech]

we will also try for you than feedback to you as soon as possibile.

[Bladesinger7x]

we will also try for you than feedback to you as soon as possibile.

Thanx...eagerly waiting for your reply.

If it helps...the memory date is week 17(maybe 16) 06.
Do the UCCC chips with newer dates clock as well as the
older ones?

I was kinda hoping to achieve ~270-275Mhz (so 166 divider
could be used), but right now, I would just like to get them
stable at stock speeds (with divider 166 or 150).

What would be the appropriate procedure for testing so that
I can consider the memory stable (i.e., which test programs
should be used to determine Windows stability; Super Pi 1M/32M,
3DMark05, Prime Blend test, Memtest for Win, gaming)?

[Onepagebook]

Thanx...eagerly waiting for your reply.

If it helps...the memory date is week 17(maybe 16) 06.
Do the UCCC chips with newer dates clock as well as the
older ones?

I was kinda hoping to achieve ~270-275Mhz (so 166 divider
could be used), but right now, I would just like to get them
stable at stock speeds (with divider 166 or 150).

What would be the appropriate procedure for testing so that
I can consider the memory stable (i.e., which test programs
should be used to determine Windows stability; Super Pi 1M/32M,
3DMark05, Prime Blend test, Memtest for Win, gaming)?

Did you try 3-4-4-8 instead?
I hope that would work!
I do apologize that 3-4-3-8 is by using 0552 after.
but ya, try 3-4-4-8 and the setting I set, should not be the prob.

[Bladesinger7x]

Did you try 3-4-4-8 instead?
I hope that would work!
I do apologize that 3-4-3-8 is by using 0552 after.
but ya, try 3-4-4-8 and the setting I set, should not be the prob.

For now I tried 3-4-3-8 (other timings on auto after load optimized defaults)
and memory speed 250MHz and it passed Prime blend test for 8 hours:


Tomorrow I will run it at 255MHz (same timings), and then 260/265 to see how
far I will get with 3-4-3-8. Then I will change to 3-4-4-8.

BTW the memory production date is week 18, year 06. Which chips are used
on this date ...and how do they perform?

[Clint]

I did gave you a full description in my earlier post, did you ever try it?

[Bladesinger7x]

I did gave you a full description in my earlier post, did you ever try it?

Yes, these are the default timings : ) ...only 3-4-3-8 is set
manually (if on auto it's 2.5-4-4-8 and no boot); all other
timings are on auto...but this is 250Mhz on a 166 divider
(cpu@2750MHz; (250MHz on 1:1 gives cpu speed at only
2250).

But so far I have not entered these timings manually in
bios...first I wanted to see if it will be Prime stable this
way...but I also raised vcore...so maybe this is the
solution...that memory controller needs more volts for
memory to be stable (with 2x512MB TCCD cpu@2800
needed only 1.4125v to be stable, now the vcore is set
to 1.45v ...ignore cpu-z voltage reading ...it is off).

[Clint]

Well, I pointed out very clearly that you shouldnt use auto timings

If you can't take an advice you are on your own.

good luck, over and out.

[Bladesinger7x]

Well, I pointed out very clearly that you shouldnt use auto timings

If you can't take an advice you are on your own.

good luck, over and out.

Yes, I know you pointed that out...but what would be difference bettwen auto and manually setting the same
values?

[Xtreem Tech]

ok may I know your peripheral situation?

[Bladesinger7x]

Sorry for not replaying sooner...I have take some time off...from everything

Well, I got another pair of TG memory with UCCC chips and this is better, but as
far as overclocking it is almost the same ...the fun end around 260Mhz...but this
pair is at least Prime stable at 255Mhz (with 406 bios default timings which are
pretty tight...at least compared to the previous timings). But I also upped vcore
a bit ...so maybe this is the solution...that the memory controller needs more
juice for 2x1GB sticks.

In the week ahead I will also try Prime with lower vcore and also 257/259 Mhz
in order to determine where the Prime stable wall is.