Micron D9 GMH\GKX\JKH etc club

[zsamz_]

Dondan try 5-4-4-9

[radaja]

No one seems to be interested in what I'm posting... hm...

IM INTERESTED
@dondan,very nice P45 results,the read score seems about right to me.
my read scores always averaged about 270-300 below my write scores.

[Hoss331]

Well, here's an Everest run on P45 so you can compare it to the Commando. Should have run PL9 though....

Whats the highest youve been able to run on PL8?

[Don_Dan]

You would think with the latency that tight with Cas 4 your read times would be 12K.

Yes, this is what leads me to think there's something wrong with it.

Dondan try 5-4-4-9

On 1:1? I tried it on 5:6, but this kit doesn't like it, I top out at about 645MHz and scores are worth than CL4, and 5-5-5-15 scales higher and scores higher.

IM INTERESTED
@dondan,very nice P45 results,the read score seems about right to me.
my read scores always averaged about 270-300 below my write scores.

Thanks for your comment!
Take a look at it again though, read is not 300 but almost 1300MB/s less than write and copy.

Whats the highest youve been able to run on PL8?

PL8 and 1:1 is a no go, PL8 and 5:6 I tried up to 530 or 535MHz.

Today I played with it on 5:6 with settings CL5-5-5-15, PL9:



This is the highest speed for Everest, MaxxMem was possible at 685MHz.
Then I used setFSB and managed to get up to 700MHz.

Here's a validation and a screenshot:



I was still able to start MaxxMem but it crashed halfway through.

[radaja]

Thanks for your comment!
Take a look at it again though, read is not 300 but almost 1300MB/s less than write and copy.

@Don Dan,WOW!! my math was way off on that one. i see it now,definitely something not right there.very strange
are you on a x64 OS? or x86?i know with my scores that the x86 would yield much lower copy scores but never low read?

[Don_Dan]

@Don Dan,WOW!! my math was way off on that one. i see it now,definitely something not right there.very strange
are you on a x64 OS? or x86?i know with my scores that the x86 would yield much lower copy scores but never low read?

I'm on XP Professional x86.

I tried the performance enhancements in BIOS ( Standard, Turbo and Extreme ), all they do is tighten the subtimings, of course they have a positive influence so I have been running Turbo for the above tests. Extreme fails to boot.

Any more ideas guys? I still want 12k across the board!

[Hoss331]

You can run PL8 on 2.00D if your using cas4, Ive used it up to ~560fsb for benching but never tried for any bandwidth numbers. The profiles also do a little more than just tighten the sub timings because the bw numbers still go up even when the sub timings are all preset the same for each profile. This is generally only the case with 2x1 kits though as with 2x2 kits theres no difference between standard and turbo. Extreme takes alot of mch to use and only does ~50mbs over turbo so its generally not worth using. If your using a bios newer than F10b you need to switch back to an older one.

[CryptiK]

I found myself in a rare mood of hardware nostalgia, so I thought I'd bust out my old DK X48-T2RSB Plus and some Team Group PC6400 C3's.

I posted this in a separate thread but I thought I put it in here too. Here's the results.

Timings: 5-5-5-15 | tRFC 42 | tRRD 3

Test run - 670 MHz | PL8 | 2.45 vdimm




675 MHz | PL7 | 2.496 vdimm




680 MHZ | PL7 | 2.544 vdimm




I wanted to get >680 MHz SPI32M stable since not many people have managed to do so. I was hopeful this kit could do it, and it didn't disappoint. The board did well too. I'm going to stop here, raising the voltage anymore would be reckless and things weren't scaling well anymore so gains would have been minimal.

SPI32M @ 683 MHz (DDR2-1366 MHz) | 5-5-5-15 | tRFC 42 | 2.656 vdimm

[Shiranui Gen-An]

I managed to eek this out with my newest Team Xtreem DDR2-667 C3 kit using 2.45V BIOS set (=2.49V real). I guess putting a fan over the MCH helps eh?

[radaja]

Any more ideas guys? I still want 12k across the board!

what Hoss331 said.i forgot on the ud3p using the 400 strap 1:1 and cas 4
you can run pl8 stable,but if you set cas 5 it wont boot at pl8
it might just net you the triple 12's your looking for

[Don_Dan]

Thanks for your help guys!

You can run PL8 on 2.00D if your using cas4, Ive used it up to ~560fsb for benching but never tried for any bandwidth numbers. The profiles also do a little more than just tighten the sub timings because the bw numbers still go up even when the sub timings are all preset the same for each profile. This is generally only the case with 2x1 kits though as with 2x2 kits theres no difference between standard and turbo. Extreme takes alot of mch to use and only does ~50mbs over turbo so its generally not worth using. If your using a bios newer than F10b you need to switch back to an older one.

I'm on BIOS F9I.
PL8 with 1:1 divider is not worth it, since I couldn't run PL8 at 590MHz anyway, so I can use 2.00B. I don't think it's strap or divider related, both settings I have shown should be over 12k read.

what Hoss331 said.i forgot on the ud3p using the 400 strap 1:1 and cas 4
you can run pl8 stable,but if you set cas 5 it wont boot at pl8
it might just net you the triple 12's your looking for

As said above already, 1:1 and cas4 is useless to me, since 400 strap, 560MHz CL4 and PL8 will be worse than 333 strap, 590MHz CL4 and PL9. Don't you agree?

[CpmF]

I receive 2*1Gb OCZ Reaper HPC PC8500 this week.

I'll see what TPower i45 is capable to

[Hoss331]

As said above already, 1:1 and cas4 is useless to me, since 400 strap, 560MHz CL4 and PL8 will be worse than 333 strap, 590MHz CL4 and PL9. Don't you agree?

I didnt mean 560 was the max for PL8, thats just the highest ive been able to test it so far since all I use are quads. With a dual it may or may not go to 590, thats what I was hoping you could tell me since you have a higher fsb limit. If it will go to 590 it should do more bw than 2.00B PL9.

[Don_Dan]

I didnt mean 560 was the max for PL8, thats just the highest ive been able to test it so far since all I use are quads. With a dual it may or may not go to 590, thats what I was hoping you could tell me since you have a higher fsb limit. If it will go to 590 it should do more bw than 2.00B PL9.

I need PL9 for 565MHz already, so it's not gonna happen.
I hooked up the same drive to my REX and read is normal, so it's not the OS.
I will revisit this board when it's no longer my 24/7 board, more cold on the NB should help as well.

[Calathea]

This run is made with 2x512 Kingston, probably D9GKX ram.
DDR2 1300 5-5-4-5 2T

I wish I played a little more with maxxmem, 900 isn't too far away.

[Don_Dan]

Congrats, that's a very nice score!

I guess I can post my best backup now, since you have taken it down by far...

[Calathea]

Thanks!
Gained more than 100 marks on switching to dice, I will probably re-reun this benchmark next time cpu tastes cold Must tweak nb more..

[George_o/c]

Are these results any good ? Done with Cellshock 1000MHz 4-4-4-12 D9GKX kit

[Samu 17]

Are these results any good ? Done with Cellshock 1000MHz 4-4-4-12 D9GKX kit

good! how many volts?

[Samu 17]

here is my result with my ocz sli pc8500 d9gmh but i can't get the ram over 1214 cas5 so i have to set it 525 cas4 2.24v. how can i go up with this ram?
ps: i'm on air on q9400

[Don_Dan]

Thanks!
Gained more than 100 marks on switching to dice, I will probably re-reun this benchmark next time cpu tastes cold Must tweak nb more..

I was relieved to see you needed dice to take it down.
MaxxMem with the CPU on ln2 would be very interesting.

Are these results any good ? Done with Cellshock 1000MHz 4-4-4-12 D9GKX kit

Clocks are decent but what's your PL? Bandwidth seems low and latency is very high.

here is my result with my ocz sli pc8500 d9gmh but i can't get the ram over 1214 cas5 so i have to set it 525 cas4 2.24v. how can i go up with this ram?
ps: i'm on air on q9400

You mean on 5:6? Your quad might be limiting you, did you test max FSB?

[George_o/c]

PL is 8 for SPi 1M run and 5 for MaxxMem benchmark
500MHz 4-3-2-3-3 PL8 with "loosened" sub-timings only requires 2.35V. In order for me to tight the sub-timings that much I fed them with 2.65V

Bandwidth is indeed low... Can it be 'cause of the chipset ? Because P35 and X48 have tighter internal latencies than X38

[zsamz_]

PL is 8 for SPi 1M run and 5 for MaxxMem benchmark
500MHz 4-3-2-3-3 PL8 with "loosened" sub-timings only requires 2.35V. In order for me to tight the sub-timings that much I fed them with 2.65V

Bandwidth is indeed low... Can it be 'cause of the chipset ? Because P35 and X48 have tighter internal latencies than X38

when unstable bandwith is lower

[Samu 17]

You mean on 5:6? Your quad might be limiting you, did you test max FSB?

yes on 5:6.. i didn't test my fsb wall but i'm sure that my q9400 get 542 (is my max fsb at the moment)

[halfwaythere]

Some results with a pair of Team Xtream 1066 cl5. GMH if i'm not mistaken.