Gskill 2133 8-9-8 testing

[chew*]

Lets take a quick look at the sticks first.

Packaging is simple, a fan is included albeit a molex connector which in the past I have noted is sort of a sore thumb other than that nothing to gripe about.

Excuse my need to make multiple posts forum only allows 5 attachments and currently my ftp is not setup yet.

[chew*]

Testing methodology is rather simple, I prefer 3 hour minimum prime blend to stress both the memory and IMC, unstable will fail, stable won't.

This results in more realistic speeds in terms of what you would end up with for 24/7 use.

Example can found in first pic.

First up Cas 8 testing, in a day or so I will round up Cas 7 testing but I have a 12 hour trip interfering in between then.

Settings used were preety much primaries and I let the auto timings roll dirty basically testing as avg joe would test to keep things simple.

[chew*]

Moving a little further along and one of the reasons why I delayed reviewing these. To truly find a memories potential you must have a cpu that can take the ram to the limits. Luckily we have such a CPU.

This I did not include in the above post as to be quite honest it took quite some time to get stable and is most likely not realistic for most 24/7 users however interesting none the less.

This post also concludes c8 testing/scaling.

[chew*]

C7 testing

Ok interesting to note even though we could complete prime 95 prior with trcd 9 @ 1133 when running Cas 7 and TRP 7 Trcd 9 is just not possible at reasonable voltages. It needs to be dropped to 10 for this combo of timings somewhere between 1110-1333.

The last screenshot in this post Is most likely not very realistic for 24/7 users so take it with a grain of salt, only put it up as it's interesting.

[chew*]

Cas 6 testing, I messed up voltages on first shot however just reference shot 2 for voltages........

As you can see even dropping to trcd 9 will not allow my set headroom above 1000 in shot 4 however from previous testing we know trcd 9 is stable up to 2220 leaving the conclusion that my particular set is cas 6 limited.

[zanzabar]

looks nice, are those psc or elpeda. also any chance of seeing what they do on amd.

[CryptiK]

They'd have to be PSC or Samsung with timings and clocks like that.

[Johnny87au]

psc chips hehe, not good enough for bbse or elpida! Chew can you run cas7 or is cas8 the best, im guessing trfc is impossible to lower..?

[FlanK3r]

Im thinking, if limitation will be "IMC" of SB or limit RAM modules here . Looks good!

[chew*]

psc chips hehe, not good enough for bbse or elpida! Chew can you run cas7 or is cas8 the best, im guessing trfc is impossible to lower..?

I'm not even pushing them yet big guy I ran out of bclk.

Read the text, give me a few days and i'll finish up.

[acos]

i see VCCsa above 1.0v.
Is this the "magic secret" for stability and not VCCio?
At least for my rig if i use FSB>104, VCCsa should be increased!

[chew*]

i see VCCsa above 1.0v.
Is this the "magic secret" for stability and not VCCio?
At least for my rig if i use FSB>104, VCCsa should be increased!

Only magic secret is knowing your own particular hardware from meticulous testing not just using someone elses settings because it worked for there chip

If you look close at the dates time of prime 95 you will see those results were done over a period of 2 days.

[l0ud_sil3nc3]

Only magic secret is knowing your own particular hardware from meticulous testing not just using someone elses settings because it worked for there chip

If you look close at the dates time of prime 95 you will see those results were done over a period of 2 days.

excellent post as usual I have been amused by seeing people just asking for templates thinking that it is the magical solution to all of their stability woes

[C-N]

Subscribed... I have some of these & was wondering where the tRCD 9 limit was, will be nice to see what you make of them Chew

[runmc]

110 bclk looks great Brian. GSkills continues to amaze.

[stasio]

This is not Gskill 2133 8-9-8 kits(only first post).
Benching 3 months ago?

[cstkl1]

isnt that like a world record for dram speed on p67??

[PolRoger]

Nice work chew*! ... post #3 I'm looking forward to you Cas 7 results...

[owikh84]

Sorry quite confused here... isn't it

Post #1 is 2133 8-9-8 kit and
Post #2 and #3 is 1600 9-9-9 kit

maxxmem 2/1/2005 hehe

[cstkl1]

hmm ure right different kit

[PolRoger]

Sorry quite confused here... isn't it

Post #1 is 2133 8-9-8 kit and
Post #2 and #3 is 1600 9-9-9 kit

maxxmem 2/1/2005 hehe

hmm ure right different kit

http://www.xtremesystems.org/forums/...5&postcount=16

[chew*]

This is not Gskill 2133 8-9-8 kits(only first post).
Benching 3 months ago?

My sample I received early, for whatever reason the spd is not correct.

Date and time in os is set backwards for a version of afterburner to work

As far as WR heh I have no clue if you want to call it a WR then go ahead I will refrain from doing so however.

[Don_Dan]

Heh, you get all the good stuff...
Looking forward to the max clock on 6-9-6!

Guys, it's probably the wrong SPD on these sticks, most likely it's the SPD of the basic sticks that these higher-clocked sticks were binned from.

[Dumo]

Nice ram

Its higher binned 17000 C8-D probly need less Vdimm for 2133 cas 6

[stasio]

My sample I received early, for whatever reason the spd is not correct.

Date and time in os is set backwards for a version of afterburner to work

As far as WR heh I have no clue if you want to call it a WR then go ahead I will refrain from doing so however.

CPU-Z and AIDA outdated also.