Micron D9 GMH\GKX\JKH etc club

[kuebk]

chad - why don't you use 2:3 multiplier?

[CpmF]

I have 2*512Mb Micron PC4200 to test, mounted with D9GCT chips

[LennyRhys]

I just got another Cell Shock 8000 C4 kit and dang it's good. 1200MHz 5-5-5 at 1.98v. I'm sure it'll Do 1000MHz 4-4-4 at around 2v but have yet to test it. Not taking it over 2.1v for bench runs, and as for daily use I want it no higher than 2v.

[CryptiK]

Nice find Lenny, they're great kits.

[Shiranui Gen-An]

I wish I could find another CS2221440 kit

[LennyRhys]

It was on ebay for like a whole week before I decided to buy it. Based on previous "preowned" Micron D9 kits I was skeptical, however the previous owner of this particular kit ran it at 800MHz 4-4-4 @ 1.8v... I was like LOL you kidding? That kit will be virtually brand new when it goes in my system, and it jolly well is - straight into memtest at 1000MHz 5-5-5 1.9v and flawless pass.

Question regarding CAS and PL - because my CPU has such a low multi (8x) I have to run really high FSB for bench runs, and consequently my memory has to be 1:1.

I got 10.2 seconds 1m at 4.6GHz, but this was DDR1150 at PL12. Will PL10 and/or tighter primary timings make much difference to the bench?

here she is:

[Don_Dan]

You're running PL12? I never quite understood why people could run very low VDimm at such clocks, but this explains it I guess.
Performance product of this run isn't very good, you should be able to run PL10 or rather PL9 in any case to get your efficiency improved.

Nice kit nontheless, it's a shame Cellshock is gone.

[LennyRhys]

Yes but can I beat 10 seconds with PL9? I agree, Cell Shock were the best DDR2, no contest. LOVE the look of their modules, and the performance is outstanding.

Tomorrow I'll try PL10, maybe even PL9, and we'll see...

[Don_Dan]

I'm happy to have four of their kits as well. The blue DDR3-1866 is very beautiful in my opinion.

Unfortunately I don't have a run with similar settings, but I did 9.830s at 4650MHz with my E8600 ( 10x465MHz, 5:6, CL4, PL7 though. ).

Oh, and I ran the special F4-PI Bios for my UD3P which improved efficiency in 1M a lot.

[LennyRhys]

9.83 seconds at 4.65 is amazing, but cas 4 PL7 explains that lol.

I tried PL10 at 575fsb and got a BSOD loading desktop, but that was with 2.1v on the memory. Do I need more vdimm for DDR1150 1:1 PL10? I always thought that PL being a chipset latency would require more chipset/fsb voltage.

[CryptiK]

Don Dan not sure what you mean about the association between PL and vdimm for a given ram speed. In my understanding, PL only affects the MCH and has nothing at all to do with what vdimm is required fpr a given ram speed. At least that is the case with my extensive testing.

For example I have run DDR2-1200 | 5-5-5-15 | trfc 30 | PL8 at 1.97v with a Cellshock kit and the same timings/speed at 1.93v with my best Team kit, both totally stable. I have also passed SPI32M at 650 MHz at <2.25v with both kits, using PL8.

[LennyRhys]

For example I have run DDR2-1200 | 5-5-5-15 | trfc 30 | PL8 at 1.97v with a Cellshock kit and the same timings/speed at 1.93v with my best Team kit

That's awesome, especially tRFC 30 - was this at 500fsb, and are your Teams GMH or GKX? The general opinion on D9 seems to be that GMH is better than GKX for high MHz cas5, and since I've only ever had low binned GMH/high binned GKX I can't say from experience.

I've been at PL10 since getting these Cell Shocks, and now I'm trying PL8. I've upped the vMCH one increment and have taken vdimm down to 1.96v, still stable. Woot.

[Don_Dan]

Thanks for the input CryptiK, maybe I have been mistaken all the time.
I always was under the impression that running tighter PLs requires both an increase in vMCH and vDIMM. I never did undervolting tests, so I don't have any data about how my kits fare at ~2.00V. Something I know for sure is that I degraded one of my Anniversary Kits when benching during the holidays, it won't boot with less than 2.00V.

Now I have some more things to check when I get back home, it has been almost 9 months since I played with DDR2 D9 for the last time. Time to get back to good old D9.
I have a new and unused TeamGroup 1066CL5 kit that will be perfect for these tests.

btw, nice sig image Lenny

[JULO]

LennyRhys, CellShock is not the best DDR2 provider
GKXs used by them are great @ CAS5, but @ CAS4 there're a lot of mems better than CS
that's because of very different steppings used by cellshock, I've had 1000 C4, 1000 C5 and red devil - all the same: pretty CAS5, crappy C4
In my opinion Team Group rules However it's hard to decide, there's too much of good rams

[LennyRhys]

They are the best DDR2 kits I've ever had. The best kit I've seen was a Ballistix kit doing 1000MHz 4-4-4-12 below 2v ( ) and it was D9GMH.

I've not pushed this kit at cas 4 and will probably hold back on the vdimm for a while - don't want this to happen

[JULO]

people say that CellShocks die very often because of too high voltage. that's not true.
they die because of too high temperature. they just throw out tones of heat

if you'll supply good enough fan you can easly give them 2.45V

[LennyRhys]

I've got this on my memory. Cell Shock have nice looking heat spreaders but they do get very hot

[JULO]

With this fan installed your cellchocks can easly work with 2.25V, and 2.4V for testing I think

[Stuen4y]

I had A-Data Vitesta Extreme 1066+ with micron GMH chips which did the remarkably on my opinion 665MHz on AMD controller and 2.42V.

I think I have screens of the whole desktop but I'm not quite sure I'll find them. The problem with the AMD memory controller was that it could not be stabilizer above 1200MHz no matter the memory can do it. Now the chips are dead and are on resurrecting program in the fridge. I hope it works...
P.S. I got them out of the fridge and Oh, miracle, they work again. After the resurrection they managed to be semi stable over 1300MHz not only on Everest but on almost all benchmarks.

[CL3P20]

@ LennyRhys- how much MCHv/VTT you running currently? For 550+ @ PL9 or 10, I need 1.38v VTT and 1.46v MCHv on my MIIF... I can run 1.35VTT and 1.42MCH with PL11 and lower still for PL12..

[Bigred1]

Im not sure if I'[m allowed to be doing this here but I dont have access to the for sale section yet. Hopefully nobody we'll be mad. But I have a set of Corsair 2x1gb DDR2-800 with 3-4-3-9 timing that i know have at least GMH's on them. I have not been able to test them to the max because i dont have a 775 board anymore. I have had them at 1280MHz at 5-5-5-15 at only 2.2 volts though and they were stable. I dont want to part with them on ebay so I figured someone here might be interested. If you are please PM me. If i wasnt supposed to do this, im sorry.

Actually I found a pic of them

[CryptiK]

That's awesome, especially tRFC 30 - was this at 500fsb, and are your Teams GMH or GKX? The general opinion on D9 seems to be that GMH is better than GKX for high MHz cas5, and since I've only ever had low binned GMH/high binned GKX I can't say from experience.

I've been at PL10 since getting these Cell Shocks, and now I'm trying PL8. I've upped the vMCH one increment and have taken vdimm down to 1.96v, still stable. Woot.

Well I ran many settings just testing, but 24/7 was 500 FSB / DDR2-1200 at the above timings & vdimm. My Team Group kit (the one that does 1200 @ 1.93v) is unknown, as I've never removed the heatspreader.

Thanks for the input CryptiK, maybe I have been mistaken all the time.
I always was under the impression that running tighter PLs requires both an increase in vMCH and vDIMM. I never did undervolting tests, so I don't have any data about how my kits fare at ~2.00V. Something I know for sure is that I degraded one of my Anniversary Kits when benching during the holidays, it won't boot with less than 2.00V.

Now I have some more things to check when I get back home, it has been almost 9 months since I played with DDR2 D9 for the last time. Time to get back to good old D9.
I have a new and unused TeamGroup 1066CL5 kit that will be perfect for these tests.

btw, nice sig image Lenny

Well I can say for sure on the boards I've tested, including DDR3, I've never needed a vdimm increase to retain stability at a given ram speed when reducing PL (tRD). Sorry to hear about your anni's degrading.

LennyRhys, CellShock is not the best DDR2 provider
GKXs used by them are great @ CAS5, but @ CAS4 there're a lot of mems better than CS
that's because of very different steppings used by cellshock, I've had 1000 C4, 1000 C5 and red devil - all the same: pretty CAS5, crappy C4
In my opinion Team Group rules However it's hard to decide, there's too much of good rams

Both Team and Cellshock make great kits. Cellshock just stand out for their build quality, their heatspreaders are second to none. That's why they get so hot - they are actually working!

[zsamz_]

Both Team and Cellshock make great kits. Cellshock just stand out for their build quality, their heatspreaders are second to none. That's why they get so hot - they are actually working!

the only heatspreaders that work lol

[PizzaMan]

I beg to differ. The OCZ Flex II heatsinks are the best in my book. Work great on my Ballistix. I'm even building a little mini loop to use on them for benching runs.

http://i170.photobucket.com/albums/u...l/100_1152.jpg

[LennyRhys]

Flex II as air heatsinks? Never thought that would work, but hey.

I don't know of any brand other than Cellshock to use thermal paste in their memory modules - this explains at least partly why they are so dang hot!