G.SKILL Triple Channel X58 series F3-16000CL6T-6GBPIS 6-9-6-24 (2GBx3)

[hiwa]

Sweet. Hyper is too hard to find nowadays besides Dom GT's. These PSC's don't seem too bad. Looking foward to 6GB air runs when you got time.

you are right, hypers are damn expensive and fragile , i tried a lot of PSC IC under cold and hyperpi/memtest never die a kit new ICs seems cool to push and tight tRCD we weren't able to do before

[M.Beier]

you are right, hypers are damn expensive and fragile , i tried a lot of PSC IC under cold and hyperpi/memtest never die a kit new ICs seems cool to push and tight tRCD we weren't able to do before

True, its getting there - when is a question the individual will have to answer

One thing is certain; PSC under cold is interesting!

Keep on the testing buddy

[hiwa]

12GB 1880 6-8-6-24-88 1t 1.64

[eva2000]

nice with low uncore volts but frequency

[hiwa]

nice with low uncore volts but frequency

low uncore =low vtt = +200 runs on my a0

[hiwa]

12GB---1900 6-8-6-24-88 1t 1.64 3dmark05


switch to 6GB

[chuchnit]

Looking good hiwa. Is there any chance of talking you into doing a little super pi efficiency testing with these? I'm interested in how far you can push these before going cold.

[hiwa]

sure but u mean with server 2k3 or XP ? , i am testing some 12GB 7-7-7-24 with lower possible voltage first

[chuchnit]

sure but u mean with server 2k3 or XP ? , i am testing some 12GB 7-7-7-24 with lower possible voltage first

You pick man. I'm an XP kinda guy myself, but with the right tweaks they are basically the same. No hurries at all. I know that my request is low low on the priority list. So tell me more about this 7-7-7-24 testing

[cstkl1]

hiwa arent those Cl6 rtl at 2k seems a bit loose

shouldnt it be hmm around 48/49,51,53,

[hiwa]

You pick man. I'm an XP kinda guy myself, but with the right tweaks they are basically the same. No hurries at all. I know that my request is low low on the priority list. So tell me more about this 7-7-7-24 testing

sure i'll try but need to found 64bit if we want to see all memory usage
sorry for delay i'll try 7-7-7-24 this evening

hiwa arent those Cl6 rtl at 2k seems a bit loose

shouldnt it be hmm around 48/49,51,53,

you are right but with +3400/3500 Uncore i have to set higher RTL on channels,will try tight them for sure

[eva2000]

hiwa arent those Cl6 rtl at 2k seems a bit loose

shouldnt it be hmm around 48/49,51,53,

probably won't scale as high unless shove more cpu vtt volts which probably ain't a good thing with 980X/gulftowns heh

[hiwa]

12GB--1600 7-7-7-24-88 1.60

[chuchnit]

Definitely an improvement in their new IC's huh? That's darn good for so many GB!

[hiwa]

12GB---2164 6-9-6-24-88 1t 1.65



2200 6.9.6.24.88 1t 1.65



http://valid.canardpc.com/show_oc.php?id=1258614

[cstkl1]

hiwa arent those Cl6 rtl at 2k seems a bit loose

shouldnt it be hmm around 48/49,51,53,

sure i'll try but need to found 64bit if we want to see all memory usage
sorry for delay i'll try 7-7-7-24 this evening



you are right but with +3400/3500 Uncore i have to set higher RTL on channels,will try tight them for sure

probably won't scale as high unless shove more cpu vtt volts which probably ain't a good thing with 980X/gulftowns heh

oops forgot the uncore was lowered

eva.. u mean 980x high vtt aint a good thing?? so like 1.6v?? on vtt??

[hiwa]

oops forgot the uncore was lowered

eva.. u mean 980x high vtt aint a good thing?? so like 1.6v?? on vtt??

if u have good IMC 1.50 is good with +5.0GHz uncore and +980 mem
the good point on gulfy is u can set lower uncore to push max your memories so IMC count but not like nehalem.


sorry george

[Johnny87au]

if u have good IMC 1.50 is good with +5.0GHz uncore and +980 mem
the good point on gulfy is u can set lower uncore to push max your memories so IMC count but not like nehalem.


sorry george

So the 980's arent like the nehalems where u need to double your uncore? Can you try running those @ lower timings @ 1600? Also whats the highest OC you can do with these dimms @ the lowest timings?

[hiwa]

So the 980's arent like the nehalems where u need to double your uncore? Can you try running those @ lower timings @ 1600? Also whats the highest OC you can do with these dimms @ the lowest timings?

Exactly you can use Unlink uncore sure when my cpu back Alive (this chip need holidays for 2,3 days) i am waiting new chips, already have a retail but really don't want kill +6.2 06 cpu

[PatRaceTin]

how far with only 6GB ????

[Johnny87au]

Mate i want some of these,12GB---2164 6-9-6-24-88 1t 1.65 im guessing thats on air yeah? INSANE!

[Kentaro]

Very nice kit indeed. Thanks for sharing. Do you know how is the perofrmance of 6-8-6 comparing with 7-7-7 at the same frequency??

[hiwa]

Mate i want some of these,12GB---2164 6-9-6-24-88 1t 1.65 im guessing thats on air yeah? INSANE!

yep air

Very nice kit indeed. Thanks for sharing. Do you know how is the perofrmance of 6-8-6 comparing with 7-7-7 at the same frequency??

as i know massman will prepare a good giude for that
is a easy to calculate but need a formula

i try to explain you with a simple formula with FIX mem freq and fix uncore (if u change uncore you have to caculate IMC latency with RTL )

tCL = (CAS *2000)/ memory frequency in ns

2200 6-9-6-24 VS 2200 8-8-8-24


SO 2200 6-9-6-24 calculation :

6 X 2000 / 2200 = 5.45XX
9 X 2000 / 2200 = 8.18XX
6 X 2000 / 2200 = 5.45XX
24 X 2000 / 2200 = 21.18
then you calculate the addition of them = 5.45+ 8.18+ 5.45+ 21.18 = 40.89ns


2200 8-8-8-24 :

8 X 2000 / 2200 = 7.27
8 X 2000 / 2200 = 7.27
8 X 2000 / 2200 = 7.27
24 X 2000 / 2200 = 21.18

7.27+ 7.27+ 7.27+ 21.18 = 42.99ns

so with same uncore the 2200 6-9-6-24 is faster than 2200 8-8-8-24

[George_o/c]

yep air



as i know massman will prepare a good giude for that
is a easy to calculate but need a formula

i try to explain you with a simple formula with FIX mem freq and fix uncore (if u change uncore you have to caculate IMC latency with RTL )

tCL = (CAS *2000)/ memory frequency in ns

2200 6-9-6-24 VS 2200 8-8-8-24


SO 2200 6-9-6-24 calculation :

6 X 2000 / 2200 = 5.45XX
9 X 2000 / 2200 = 8.18XX
6 X 2000 / 2200 = 5.45XX
24 X 2000 / 2200 = 21.18
then you calculate the addition of them = 5.45+ 8.18+ 5.45+ 21.18 = 40.89ns


2200 8-8-8-24 :

8 X 2000 / 2200 = 7.27
8 X 2000 / 2200 = 7.27
8 X 2000 / 2200 = 7.27
24 X 2000 / 2200 = 21.18

7.27+ 7.27+ 7.27+ 21.18 = 42.99ns

so with same uncore the 2200 6-9-6-24 is faster than 2200 8-8-8-24

Great info Hiwa , thanks dude!

[Johnny87au]

SO 2200 6-9-6-24 calculation :

6 X 2000 / 2200 = 5.45XX
9 X 2000 / 2200 = 8.18XX
6 X 2000 / 2200 = 5.45XX
24 X 2000 / 2200 = 21.18
then you calculate the addition of them = 5.45+ 8.18+ 5.45+ 21.18 = 40.89ns

So your running a 2000 uncore right? I think i get it,, So for example the one i quoted you did

6x2000 uncore divide 2200 =5.45?