New WR DDR636 2.5-3-3-5 1T PI 32M

**gigi77** · 10-25-2005, 10:26 AM

mr. Onepagebook you have my pvt

luigi

**LoKi2k** · 10-25-2005, 10:57 AM

alright, here i am:

mission done:

final pic:

some one want to buy these sticks?

**x3ro** · 10-25-2005, 11:06 AM

WTF!! Thats amazing m8 :| realy.. Congrats with ur WR.... What sticks are you using?

Regards

**$ilver** · 10-25-2005, 11:11 AM

G.Skill 4400LE i think

**x3ro** · 10-25-2005, 11:15 AM

nice for pc4400 sticks..

**LoKi2k** · 10-25-2005, 11:18 AM

jep 4400 le

bios is 703-1

**bachus_anonym** · 10-25-2005, 11:19 AM

@LoKi2k,

I'm not trying to be an a$$, but I have to tell you that even considering that CPU-Z was open during the run, 23:49 is pretty bad time for 3.2Ghz and such high 6-3-3-2.5-1T clock.

**szukalski** · 10-25-2005, 11:28 AM

Originally Posted by ianocean

100Mhz 1.95us -what TREF is this displayed in DFI BIOS

4708 I think.

**LoKi2k** · 10-25-2005, 11:28 AM

hmm... i dont know.... my 1m-time with thise setting is 25.89x. is that also a bad time? i donÂ´t know.

**amrgb** · 10-25-2005, 11:51 AM

You have a camera. Do like AndreYang.

Take shots every 2-3 steps.

Not saying that you've done it, but it is possible to run 23 loops with say 300-310 and then change the clock to 320 with clockgen for the last one (or a timing with the tweaker) and voila.

You have to realize that we are now at a point where for some people is hard to believe in higher clocks (specially when they come in almost a daily basis).

**Zeus** · 10-25-2005, 11:58 AM

Ugh!

Like Bachus said, time is very bad for 10x320 but this could be caused by the Tref.

Could you try again with Tref 100MHz 1.95us?

Please take a few pics during the run with CPU-Z open, memory tab visible.

Still, very very nice result!

**LoKi2k** · 10-25-2005, 12:22 PM

umpf!

ok, tomorrow

**bachus_anonym** · 10-25-2005, 12:37 PM

Update: Scratch that 305Mhz I said above... 303Mhz will be just fine, If run with one CPU-Z window open...

**ianocean** · 10-25-2005, 12:52 PM

Zeus did 32M 23.03 at 10x313Mhz Mhz so i think is not normal at 320Mhz to get such bad time

**s7e9h3n** · 10-25-2005, 12:57 PM

Originally Posted by bachus_anonym

Ummm michal,
Look again...you're not exactly making a "fair" comparison. Check your tweaker table again ->
Your trtw =2, his is =3
Your bypass max = 4x, his = 7x
Your MAL,.PR = 8,5.0, his = 9,7.0
Your Drivestrength = Normal, his = Weak
On top of that, you're running a way higher Vcore than him - which will ultimately give you a much faster time even with the same settings (with the assumption that the cpu's scale similarly with similar voltages). So in the end, your 305 shouldn't be anywhere near his time if you run the exact same timings......

**Zeus** · 10-25-2005, 01:03 PM

That cannot make such a big difference.
Maybe his OS is a bit worn?

Let's wait for his run with pictures tommorow, it's too early to judge yet.

**AndreYang** · 10-25-2005, 01:17 PM

Originally Posted by s7e9h3n

Ummm michal,
Look again...you're not exactly making a "fair" comparison. Check your tweaker table again ->
Your trtw =2, his is =3
Your bypass max = 4x, his = 7x
Your MAL,.PR = 8,5.0, his = 9,7.0
Your Drivestrength = Normal, his = Weak
On top of that, you're running a way higher Vcore than him - which will ultimately give you a much faster time even with the same settings (with the assumption that the cpu's scale similarly with similar voltages). So in the end, your 305 shouldn't be anywhere near his time if you run the exact same timings......

So i say dont use bad A64tweaker run 32M.

If i use his tweaker , maybe i can run 320+.........

**TheMeatFrog** · 10-25-2005, 01:27 PM

His Trc = 10, Max Async Latency = 9, Read Preamable = 7 and his tref is one of the slower ones when it comes to PI. These are just a few performance killers, but he isn't going for best time just max MHz so it doesn't really matter.

His time seems about right to me after taking the above into consideration.

Congrats on a very very nice clock

**amrgb** · 10-25-2005, 01:38 PM

Originally Posted by AndreYang

If i use his tweaker , maybe i can run 320+.........

What are you waiting for? Claim the WR again

I would like to know where this will stop. One of these days the 2.5-3-3 WR would be higher than the 2.5-4-3 one

Edit: went check at the ranking, 5 more Mhz and both WR will be leveled

**s7e9h3n** · 10-25-2005, 01:58 PM

Originally Posted by AndreYang

So i say dont use bad A64tweaker run 32M.

If i use his tweaker , maybe i can run 320+.........

Agreed, your timings above are superb, only thing more I would hope to see is trfc @ 13

But it's hard to get everyone to "standardlize" their timings and if people want to make comparisons, they should at least use the same settings, that's all.....

**bachus_anonym** · 10-25-2005, 02:51 PM

@s7e9h3n and TheMeatFrog,

You guys put way too much money into claim that Read Preamble or MAL are some sort of huge performance boosters. Those values can be very crucial to stability but not performance, especially when 32M is concerned.

My batteries went out so no DC shots, but I took 4 screen grabs during 302x10=3020 run. I cannot match exactly same settings (TRWT=2, Drive Strength=Normal and BypassMax=4 are the factors in my setup that allow mems run above 300Mhz 6-3-3-2.5-1T) but those 3 settings ARE NOT speeding up 32M calculation that much, Only TRWT=2 does it, but it's a matter of ~2s. Anyway, at 302x10 6-3-3-2.5-1T, with CPU-Z open, Normal Priority and no-tweak OS I get 24:03. If you guys still think that LoKi2k's time is still "about right" for such high CPU and Mem clocks, then with all due respect, you're wrong.

What I'm trying to say, is that at this point it's really hard to put a lot of trust in such high clock/ bad time results. Especially, that it is PC4400LE... It's just unheard of to achieve anywhere near that with those modules. I will stand corrected If it's otherwise

**s7e9h3n** · 10-25-2005, 02:52 PM

Originally Posted by Zeus

That cannot make such a big difference.
Maybe his OS is a bit worn?

Let's wait for his run with pictures tommorow, it's too early to judge yet.

Yes it can...I just did the math (I think) that Bachus used to calculate his estimation of 303-305 FSB with his timings to run the same spi time and his timings definately gives him an advantage.....
But if for example, Loki2k decided he wanted to LOWER his htt to run the same 24m14s time as Bachus, he would still need to run @ 314 HTT using his current timings to equal that time.....
If my math is right, with his timings, Bachus is running @ about 8-9% faster per clock than Loki2k

**dinos22** · 10-25-2005, 03:01 PM

bachus if you check out Opteron OC times you'll see that they lose out to San Diegos in 32M runs due to MAL/Read Preamble as people need to loosen those with Opterons to get them stable

Originally Posted by bachus_anonym

@s7e9h3n and TheMeatFrog,

You guys put way too much money into claim that Read Preamble or MAL are some sort of huge performance boosters. Those values can be very crucial to stability but not performance, especially when 32M is concerned.

My batteries went out so no DC shots, but I took 4 screen grabs during 302x10=3020 run. I cannot match exactly same settings (TRWT=2, Drive Strength=Normal and BypassMax=4 are the factors in my setup that allow mems run above 300Mhz 6-3-3-2.5-1T) but those 3 settings ARE NOT speeding up 32M calculation that much, Only TRWT=2 does it, but it's a matter of ~2s. Anyway, at 302x10 6-3-3-2.5-1T, with CPU-Z open, Normal Priority and no-tweak OS I get 24:03. If you guys still think that LoKi2k's time is still "about right" for such high CPU and Mem clocks, then with all due respect, you're wrong.

What I'm trying to say, is that at this point it's really hard to put a lot of trust in such high clock/ bad time results. Especially, that it is PC4400LE... It's just unheard of to achieve anywhere near that with those modules. I will stand corrected If it's otherwise

**bachus_anonym** · 10-25-2005, 03:41 PM

@dinos22,

That was initial information that TeamJapan gave all of us... Take a look at AndreYang's screenshot... He's running Opteron and his MAL and RP are not high. Also, there's quite a few good results in Opteron thread with also "normal" MAL and RP. Anyway, this is setting that, like I said, might have large impact on stability but rather insignificant on performance (unless, one is going into >10ns values...).

Ohm btw... Just in case someone would like to say that I run Server2003 so it might not be fair either, since someclaim that Server2003 is faster

I just re-run 302x10 at same settings and conditions, but under WinXP SP2 with default services. In other words, totally standard, non-tweaked OS... How about a shocker, 24:00 - 3secs faster than on Server2003

Enjoy...

Now that i lived up to my title, time to go out and grab a beer

**poiuy223** · 10-25-2005, 05:10 PM

Originally Posted by Onepagebook

hi Master gigi77, your run and everything is more impressive

, at least you won't get the shxty ram program them first(oh ya , pry the heatspreaderand from geil reinstall on the shxty ram) then return to store and cost you nothing. I should learn this huh?

wow. a little harsh on someone just because they've taken the new WR with memory not from g.skill

Thread: New WR DDR636 2.5-3-3-5 1T PI 32M

Thread Tools

Search Thread

Rate This Thread

Display

Bookmarks

Bookmarks

Posting Permissions