Micron D9 GMH\GKX\JKH etc club

[zsamz_]

It looks like I have DDR2-1250 stable (or close to it) at 521FSB PL8, but I had to use 2.42V BIOS set with my Ballistix 8500 16FD5.

use what voltage you comfortable with
but i would not use over 2.3 for 24/7
why not 1225mhz you should be able to do it with 2.05v
@ 510 fsb you will go over 11k in everest

[fatguy1992]

I quickly got my CellShock PC2-8500's @ 1066MHz 5-5-5-15 2.0v and 800MHz 4-4-4-12 1.9v. They were both stable enough 2 play games for ages.

[Shiranui Gen-An]

use what voltage you comfortable with
but i would not use over 2.3 for 24/7
why not 1225mhz you should be able to do it with 2.05v
@ 510 fsb you will go over 11k in everest

I just find it odd you, Cryptik and others hae such success with high clocks on low volts. I have some of the same types of kits (Cellshock D9GKX, Ballistix D9GMH) and I know performance will vary, but still...

[zsamz_]

I just find it odd you, Cryptik and others hae such success with high clocks on low volts. I have some of the same types of kits (Cellshock D9GKX, Ballistix D9GMH) and I know performance will vary, but still...

use only ballistix with ut p35 they dont like d9gkx takes way too much volts

[Shiranui Gen-An]

I'm wondering if my Ballistix aren't GKX considering how much voltage they need to go over 600MHz...

[eva2000]

I just find it odd you, Cryptik and others hae such success with high clocks on low volts. I have some of the same types of kits (Cellshock D9GKX, Ballistix D9GMH) and I know performance will vary, but still...

also has alot to do with board/bios used as well and there's very very early D9GMH as well that didn't do as high as later revisions

Got some of the later stuff as well as old stuff..

later/newer 16FD5 and 16FD3 tracers



small segment of my D9GMH collection

[WaterFlex]

Holy
Sell this all and buy a car,dude

[Shiranui Gen-An]

also has alot to do with board/bios used as well and there's very very early D9GMH as well that didn't do as high as later revisions

Got some of the later stuff as well as old stuff..

later/newer 16FD5 and 16FD3 tracers

That's just it, I have a kit of 16FD3 Tracers and 16FD5 regular Ballistix (both 8500), Cellshock 8000 C4 and C5, Team Xtreem 9600, Dominator 6400C3 (two kits) and 10000C5, XMS2 8500 (D9GMH), Reaper 8500 (old version with D9GMH), Geil 9280 C4 and some other stuff. Boards I'm using are Maximus Forumla with Rampage latest BIOS, Maximus II Formula, DFI LT P35, P5B Deluxe. You'd think SOME combination of the above could net me 650MHz+ stable

[eva2000]

how you define stable ? pi/3d benching is easy on different boards different levels of stability

[WaterFlex]

Outstanding results! Good job, Eva.

[Shiranui Gen-An]

how you define stable ? pi/3d benching is easy on different boards different levels of stability

Well, it needs to pass at least 2-3 full loops in Memtest86+ before I even think about trying to boot into Windows to do anything.

[WaterFlex]

Well, it needs to pass at least 2-3 full loops in Memtest86+ before I even think about trying to boot into Windows to do anything.

Windows MemTest 3.8 is the best utility to stress RAM<--->NB.
Eva2000. Can you run some memtest with those setups?

[Shiranui Gen-An]

Windows MemTest 3.8 the most stressfull for RAM.
Eva2000. Can you run some memtest with those setups?

Well if it can't even get through DOS MemTest, I don't think Windows will even be able to boot...

[zsamz_]

I'm wondering if my Ballistix aren't GKX considering how much voltage they need to go over 600MHz...

its verry possible they might be d9gkx
my 16fd3 and my 16fd5 sets do 1200 with 2.0v
i cant even do 1200 with d9gkx 2.3v

[eva2000]

Outstanding results! Good job, Eva.

thanks

Well, it needs to pass at least 2-3 full loops in Memtest86+ before I even think about trying to boot into Windows to do anything.

yeah i usually do at least 2 full passes and loop test #5 for around 10-20 passes before booting into windows

Shiranui, since DFI P35-T2R is board we have in common, back then I found clk fine delay crucial to breaking the barriers with Micron D9..

take a read of http://i4memory.com/f80/dfi-lp-ut-p3...ne-delay-4743/ and http://i4memory.com/f80/dfi-lanparty...eenshots-4487/ for DFI P35

For DFI X38 http://i4memory.com/f80/dfi-lanparty...friendly-5226/

Windows MemTest 3.8 is the best utility to stress RAM<--->NB.
Eva2000. Can you run some memtest with those setups?

with right combination of clocks and vdimm probably but i usually don't do that as the aim is for super pi 32m/dual 32m or 3dmark stable so unnecessary stress on system/ram would be wasted as i don't plan on running that 24/7 anyway.

its verry possible they might be d9gkx
my 16fd3 and my 16fd5 sets do 1200 with 2.0v
i cant even do 1200 with d9gkx 2.3v

or very very old/early D9GMH. I had one of the first released Team PC2-6400C3 no heatspreader D9GMH which was capable of high 4-4-4-5 up to around 610Mhz at very high vdimm 2.65+ but 5-5-5-15 didn't go any higher

[neo mike]

Wow Eva, where did you get that warehouse full of memory ?, any of that up grabs ?

[chad]

also has alot to do with board/bios used as well and there's very very early D9GMH as well that didn't do as high as later revisions

Got some of the later stuff as well as old stuff..

later/newer 16FD5 and 16FD3 tracers



small segment of my D9GMH collection

Hi eva2000.
Do you want to sell any of kits you have in the picture? Pm me with details of what you want to sell.

[CryptiK]

I just find it odd you, Cryptik and others hae such success with high clocks on low volts. I have some of the same types of kits (Cellshock D9GKX, Ballistix D9GMH) and I know performance will vary, but still...

Don't stress about it, with the benches I have shown, I was nowhere near totally stable, just stable enough for whatever I was doing. None of the benches would have been SPI32M stable, I don't run memtest first I just guess volts/speed and boot and bench.

I am just now getting the time to run some SPI32M and the vdimm required for stability is much higher than required for everest/SPI1M.

Sure I can do 600 MHz at 1.98 - 2.02v with my GKX/GMH totally stable, but most GKX/GMH can. More than that and the required vdimm for stability increases rapidly.

Hi eva2000.
Do you want to sell any of kits you have in the picture? Pm me with details of what you want to sell.

Get in line (behind me) I however doubt eva would sell any of his beloved ram, but if he does, he knows I'll take it

[Shiranui Gen-An]

Don't stress about it, with the benches I have shown, I was nowhere near totally stable, just stable enough for whatever I was doing. None of the benches would have been SPI32M stable, I don't run memtest first I just guess volts/speed and boot and bench.

I am just now getting the time to run some SPI32M and the vdimm required for stability is much higher than required for everest/SPI1M.

Sure I can do 600 MHz at 1.98 - 2.02v with my GKX/GMH totally stable, but most GKX/GMH can. More than that and the required vdimm for stability increases rapidly.

That makes be feel a bit better. I thought my memory sucked, or I just sucked at overclocking them. 600MHz is pretty easy to get stable with most of my kits, requiring 2.05~2.15V but some of the lower binned stuff needs more like 2.35~2.45V (my G.Skill HZ 6400 and Ballistix 5300/6400 come to mind). That being said, my CellShock kits are just weird. They don't seem to like too much voltage; if I go above a certain amount, like 2.25V with the 1000 C4 kit, memtest just locks up.

[eva2000]

Wow Eva, where did you get that warehouse full of memory ?, any of that up grabs ?

ram addict that i am.. bought quite a few kits when they were available - i manually test each module properly match them in dual channel pairs - hence masking tape labels on heatspreader

Hi eva2000.
Do you want to sell any of kits you have in the picture? Pm me with details of what you want to sell.

plan to sell some locally in Australia but not sure what the going rate/price for these are outside of OZ ?? And well shipping would be expensive ...

Don't stress about it, with the benches I have shown, I was nowhere near totally stable, just stable enough for whatever I was doing. None of the benches would have been SPI32M stable, I don't run memtest first I just guess volts/speed and boot and bench.

I am just now getting the time to run some SPI32M and the vdimm required for stability is much higher than required for everest/SPI1M.

Sure I can do 600 MHz at 1.98 - 2.02v with my GKX/GMH totally stable, but most GKX/GMH can. More than that and the required vdimm for stability increases rapidly.

1M vs 32M - 32m would need at least 0.1-0.15v more vdimm than 1M usually.

That makes be feel a bit better. I thought my memory sucked, or I just sucked at overclocking them. 600MHz is pretty easy to get stable with most of my kits, requiring 2.05~2.15V but some of the lower binned stuff needs more like 2.35~2.45V (my G.Skill HZ 6400 and Ballistix 5300/6400 come to mind). That being said, my CellShock kits are just weird. They don't seem to like too much voltage; if I go above a certain amount, like 2.25V with the 1000 C4 kit, memtest just locks up.

memtest lock ups on DFI boards can potentially be overcome via clk fine delay tweaking http://i4memory.com/f80/dfi-lp-ut-p3...ne-delay-4743/ . Example, I had D9GMH which had troubles stablizing at 600mhz 5-5-5-15 2.19v at first but clk fine delay tweaks allow same ram to shine at 640mhz 5-5-5-15 at 2.19v and 645mhz at 2.23v 32M stable http://i4memory.com/f80/dfi-lp-ut-p3...743/#post82457 !



Or if your ram is D9GKX, then it may not play as nice on DFI heh

[neo mike]

Well shipping couldnt be that much to usa, they dont weight much im interested in some, let me know.

[zsamz_]

That makes be feel a bit better. I thought my memory sucked, or I just sucked at overclocking them. 600MHz is pretty easy to get stable with most of my kits, requiring 2.05~2.15V but some of the lower binned stuff needs more like 2.35~2.45V (my G.Skill HZ 6400 and Ballistix 5300/6400 come to mind). That being said, my CellShock kits are just weird. They don't seem to like too much voltage; if I go above a certain amount, like 2.25V with the 1000 C4 kit, memtest just locks up.

you gotta try runnin some kits @ cas4 i have 2 sets of team that shine @ cas3 my ballistix love cas5
i had a kit doin 4-4-4-8 @1080 2.24v but they fell victim to my old gigabut board
its as good as doin 1200+ cas5
just run lowest volts to highest bandwidth to lowest latency for 24/7 benchin is another story anythin goes

[Shiranui Gen-An]

Or if your ram is D9GKX, then it may not play as nice on DFI heh

The Cellshock indeed is D9GKX, and acts pretty strangely on all my boards. My Team 1200 kit however seems to act more like my D9GMH even though it is GKX; i.e. if I go too high or whatever I just get errors in memtest, not lockups.

[CryptiK]

I haven't got any lockups with my GKX kits, but I'm still experimenting with them to see what they can do. Do yours lock up just from voltage greater than 2.25v or also from going too far on a given voltage? I have run my GKX at 2.4v in memtest and they were fine. When I push them too hard, they just give errors in memtest or not exact in round in SPI32M, same as with my GMH.

[Shiranui Gen-An]

Well I was trying to push them higher than 625MHz so I gave them more voltage, but it seemed to make things worse rather than better. My Dominator 10000C5 kit acts the same way, making me suspect there is GKX under the hood on those as well.