degraded d9 GTS, try more volts, sell or rma

[zanzabar]

i have a set of gskill HZ i have clocked them and done the 2ghz cas8 on a blackops and they used to do 1622 6-6-6-18-1t on my 720be but now they dont clock well and i have not taken them over the 1.9V that they are rated for (so i dont feel bad about rmaing them since they were never overheated or over volted and thats what i would consider miss use and or abuse)

so should i try more voltage (im on a 1090 now and heard that thulbans dont like high volt ram and degrade), sell them as they may get me an upgrade in density and some ok ram since u cannot buy new good IC's or should i rma and hope i get something good back (knowing full well i have a 99% chance of psc coming back and not up to my old rams specs and wont clock but i might get a density upgrade but cas8 or 9)

[CryptiK]

Gonna be hard to sell them when degraded, maybe for $5 or something. If the choice is between causing possible damage to your cpu running them at higher vdimm or rma and get whatever back, it's a no brainer. You can always sell what you get back and buy something else used or new.

[zanzabar]

Gonna be hard to sell them when degraded, maybe for $5 or something. If the choice is between causing possible damage to your cpu running them at higher vdimm or rma and get whatever back, it's a no brainer. You can always sell what you get back and buy something else used or new.

d9 like to degrade to need more volts (not clock or time lower when benching), they should still clock better than anything u can get new right now that is not really old holdout stock.


i also dont have anything 775 ddr3 anymore so i cannot test high clocks without ln2 or dice

[CryptiK]

Hypers out clock the best D9's, I have a GTS kit myself and got better timings/times using Hypers on my X48 REX. Running 900MHz+ at 6-5-4-X is no problem with hypers and no GTR/GTS kit I ever tried could manage that. You may notice a few MHz come off the top end when they degrade, especially running super tight timings which is what GTR/GTS did best until Hypers came on the scene.

[zanzabar]

since u cannot get hypers or even the non hyper elpidia anymore, good ic kits even if they like voltage are kinda rare.

hence the question, as i would hate to just send thees off to get some $40 crap back when u cannot buy anything of their quality even if they are degraded currently. then again it could be the cpu that sucks on the imc but the ram is still good.

[Leeghoofd]

DGTRs are still in plenty of stock at CSX company ( cross checked with them 1 month ago ) but with your Setup I would go powerchips now Zanzabar....Asus and Thuban love them...

[BeepBeep2]

Hypers out clock the best D9's, I have a GTS kit myself and got better timings/times using Hypers on my X48 REX. Running 900MHz+ at 6-5-4-X is no problem with hypers and no GTR/GTS kit I ever tried could manage that. You may notice a few MHz come off the top end when they degrade, especially running super tight timings which is what GTR/GTS did best until Hypers came on the scene.

6-6-6-18 at 1800 is possible with good bin Hypers, 6-5-5 you are looking at about 1450, 5-6-5 maybe 1480-1500 if you have a decent kit. 6-6-5 at 1800 ish is possible, 6-5-4 at 1800 though, that's impossible.
http://valid.canardpc.com/show_oc.php?id=1680718


Those aren't Hypers But stability at that speed is a different equation.

My Hypers manage 1840 6-6-6-18 stable at 1.74v, these need 2.3v for 1900 6-6-6-15 but it's only stable enough to run 1M.

[Hondacity]

6-6-6-18 at 1800 is possible with good bin Hypers, 6-5-5 you are looking at about 1450, 5-6-5 maybe 1480-1500 if you have a decent kit. 6-6-5 at 1800 ish is possible, 6-5-4 at 1800 though, that's impossible.
http://valid.canardpc.com/show_oc.php?id=1680718


Those aren't Hypers But stability at that speed is a different equation.

My Hypers manage 1840 6-6-6-18 stable at 1.74v, these need 2.3v for 1900 6-6-6-15 but it's only stable enough to run 1M.

you're showing your suicide runs against fully stable runs..

[BeepBeep2]

you're showing your suicide runs against fully stable runs..

What do you not understand about "Stability is enother equation"

I've been extremely time limited the last few days, I haven't gotten enough time to test stability, or even sit down and watch 32M.

[Erklat]

since u cannot get hypers or even the non hyper elpidia anymore

Are you sure? Check my sig

Are they degraded in a manner they won't run declared latencies and voltage set? If that is the case, then selling them is out of the question as they are broken.

[Leeghoofd]

Beep you are comparing to results on X48 and AMD 890FX... Cryptik is known as a ram addict and only does real stability runs to call it stable. 2.3Vdimm through a rated kit of 1.9-2.0 Vdimm is never a good idea for daily usage...

The OP is looking for a a 24/7 solution, he has got an Asus board and a Thuban : nothing like a powerchip based ram kit for that... eg GSkill Flare : 2000-2200Mhz CL7-9-7-21 1.65Vdimm for 24/7... (only if his IMC is up for that)

[BeepBeep2]

Beep you are comparing to results on X48 and AMD 890FX... Cryptik is known as a ram addict and only does real stability runs to call it stable. 2.3Vdimm through a rated kit of 1.9-2.0 Vdimm is never a good idea for daily usage...

The OP is looking for a a 24/7 solution, he has got an Asus board and a Thuban : nothing like a powerchip based ram kit for that... eg GSkill Flare : 2000-2200Mhz CL7-9-7-21 1.65Vdimm for 24/7... (only if his IMC is up for that)

I was comparing results from my platform to my platform, I was simply stating 6-5-4 at 1800 will not happen at all with any ddr3.
2.0 to 2.15 would be fine 24/7 imo on older Deneb, but yes 2.3 is really pushing things.

I'll shut up now as I'm not of any help to zanzabar. More volts isn't the answer on Thuban though, you just get lesser results

[CryptiK]

6-6-6-18 at 1800 is possible with good bin Hypers, 6-5-5 you are looking at about 1450, 5-6-5 maybe 1480-1500 if you have a decent kit. 6-6-5 at 1800 ish is possible, 6-5-4 at 1800 though, that's impossible.

Those aren't Hypers But stability at that speed is a different equation.

My Hypers manage 1840 6-6-6-18 stable at 1.74v, these need 2.3v for 1900 6-6-6-15 but it's only stable enough to run 1M.

Here's a 32M on X48 with Hypers @ 900 MHz 6-6-4-20 tRFC 60 @ 1.76v Link

The kit had more in it had I pushed harder on the vdimm. It did much better on X58 too, needing less vdimm for the same clocks vs the REX.

Did 32M @ what I remember to be 900MHz 6-5-4-X at >2.00v (possibly DICE assisted can't remember) with the same kit but cannot find the screen, think it was on a HDD I hosed during benching. I was hell bent on going sub 11min 32M at sub 4GHz with nothing but raw power and no OS optimization so the Hyper abuse was necessary.

Beep you are comparing to results on X48 and AMD 890FX... Cryptik is known as a ram addict and only does real stability runs to call it stable. 2.3Vdimm through a rated kit of 1.9-2.0 Vdimm is never a good idea for daily usage...

The OP is looking for a a 24/7 solution, he has got an Asus board and a Thuban : nothing like a powerchip based ram kit for that... eg GSkill Flare : 2000-2200Mhz CL7-9-7-21 1.65Vdimm for 24/7... (only if his IMC is up for that)

Yeah there's even some insane clocking samsung? IC based kits out there at the moment too maybe worth considering - doing crazy CL6 clocks but higher tRCD. Not sure how they play with Thuban though.

[Alex-Ro]

1800 with 6-5-4 in 32M is just simply not possible.

[Hondacity]

1800 with 6-5-4 in 32M is just simply not possible.

i believe cryptik

Did 32M @ what I remember to be 900MHz 6-5-4-X at >2.00v (possibly DICE assisted can't remember) with the same kit but cannot find the screen, think it was on a HDD I hosed during benching. I was hell bent on going sub 11min 32M at sub 4GHz with nothing but raw power and no OS optimization so the Hyper abuse was necessary.

[Alex-Ro]

Pics or you know what

[Hondacity]

read the statement again

[CryptiK]

I can't prove it with pics Alex, I can't find anything from those bench sessions, and I did the typical thing of thinking i'd saved them onto USB, pushed harder for stupid tight timings and corrupted the OS and couldn't retrieve it.

I will say though that those sticks (1866C7's) loved CL6, and did almost as well at CL7 as CL8, and tRDC was about as influential as CAS in determining max clocks at a given vdimm. My 2000C8 kit behaves very differently. I did 900 6-6-4-X on air at ~1.75v and at ~2.1v on DICE 900 6-5-4-X seems plausible on paper and I'm quite sure I got through 32M with those timings as I went sub 11min without changing anything else from the bench I linked. I'm not infallible though and I'm not claiming anything really, just saying what I am pretty damn sure I did.

[cdawall]

tried throwing them in the freezer yet? saved my D9JNL's after some 2.4v DICE runs

6-6-6-18 at 1800 is possible with good bin Hypers, 6-5-5 you are looking at about 1450, 5-6-5 maybe 1480-1500 if you have a decent kit. 6-6-5 at 1800 ish is possible, 6-5-4 at 1800 though, that's impossible.
http://valid.canardpc.com/show_oc.php?id=1680718


Those aren't Hypers But stability at that speed is a different equation.

My Hypers manage 1840 6-6-6-18 stable at 1.74v, these need 2.3v for 1900 6-6-6-15 but it's only stable enough to run 1M.

and my D9JNL could run 1840 6-6-6-18 as well



thats not stable but its neither is yours