GSkill F2-8500CL5D-4GBTD - Trident - Low voltage

[FrAN_oC]

Hi guys!!

This is my first "review" post at Xtremesystems, so be patient

I have just received 2 kits of brand new Trident DDR2 modules. These are the 8500's ones, so I hope to get 1200mhz out of them...

I got them, because I received a refund for my previous D9 sticks. Two of them dead, and they sent me a refund to buy another ones. I was between the PC8800PIs and these brand new trident.



There is a mistake on the HS. You can read "DDR3 Trident", but these are DDR2 sticks... First, I thought... "they sent me another kits... WTF ", but when I took the box, I saw the P/N... and it was correct...

Another pics of the sticks:







It seems very difficult to take off the heatspreader, so I will find out what ICs do they have only by testing them.

Specifications from GSkill site:

CAS_Latency: 5-5-5-15
Capacity: 4GB (2GB x 2)
Speed: DDR2-1066 (PC2-8500)
Test_Voltage: 1.8 Volts

Registered/Unbuffered: Unbuffered
Error_Checking: Non-ECC
Type: 240-pin DIMM
Warranty: Lifetime

Test setup:

ASUS Rampage Formula x48
Q6600 G0
4 x 2Gb GSkill PC8500 Trident LV
2 x ATI Radeon 4870 512Mb CF
2 x 500 Gb RAID 0 WD
Windows Vista x64 SP1

I have only one kit installed at this moment, and I will start with testing right now.

After installing them, I just booted and configured default timmings, speed & voltage. Also, I pulled down the CPU to start with testing.

SPD information:





Wait until I finish the tests.

(EDITING POST... WAIT FOR A WHILE)

RGDS!

[FrAN_oC]

2 x 2Gb 5-5-5-15 1.80v (First run without adjusting subtimmings and without burning sticks)

[Jor3lBR]

I will be following results, mostly interested to see how much more then 1200Mhz they can clock (if they can cause they can be binned to run 8500 only). They are in stock at the Egg.

[radaja]

as will i,i really like those heatspreaders
and if they are anywhere as good as the g.skill 9600's
this might be a great kit to get

[LagunaX]

I bet they are Elpida like the OCZ Blade Low Voltage

[gymenii]

The pics on Newegg also show DDR3.



http://www.newegg.com/Product/Produc...ddr2%20trident

Someone at G.Skill is either blind or this is a brilliant marketing scheme?

[FrAN_oC]

I took the pictures when I opened the package... I think that GSkill wanted to use this HS even being from DDR3 kits...

Now I am testing the first kit @ 1.80v. I am using Prime95 Blend test. It seems difficult to take them above 550mhz. I have no idea of what subtimmings do I have to set for these kits. I used to "tweak" my previous D9's. These are quite different...

Also I did a couple of memtest with all sticks (8Gb) and everything went fine @ 1066 1.80v.

Can anyone help me?? (Sorry... I am a "noob" testing this kind of memory)

[gymenii]

I took the pictures when I opened the package... I think that GSkill wanted to use this HS even being from DDR3 kits...

Or they decided to "get these modules out to market now" instead of waiting for the correct "DDR2" HS to be made. Anyway, looking forward to your results.

[purecain]

what performance lvl are you using....?

[FrAN_oC]

I used PL 9 for first tests @ 5-5-5-15. You can see in the screenie.

I did a fast test @ 4-4-4-12 1.80v and they didn't go above 430mhz...

I think that PC8500 tridents don't have the "famous" Elpida chips. I will keep on testing tomorrow (here in Spain is 3:00 AM...) I will leave memtest running overnight to burn the sticks.

Bye!

[18 is # 1]

WAKE UP!!!!
Sleep is highly over-rated. Tell us what's next!!!

[FrAN_oC]

EDITED: I gave them a try, and I was only able to put them @ 575 5-5-5-15 with 1.9v. I think that these are bad Elpida ICs, or PSC/ProMOS... I will leave them burning, and tomorrow I will keep updating this post with new tests.

[zsamz_]

i guess they all binning tight
but the gskills price is better n looks nicer

[Don_Dan]

I bet they are Elpida like the OCZ Blade Low Voltage

I agree with you, they must be binning them pretty well like OCZ, I wouldn't be surprised if we saw 1150 and 1200 Trident kits in the not so distant future.

i guess they all binning tight
but the gskills price is better n looks nicer

Think so too, I guess they are building up stock for higher-binned versions.

I used PL 9 for first tests @ 5-5-5-15. You can see in the screenie.

I did a fast test @ 4-4-4-12 1.80v and they didn't go above 430mhz...

I think that PC8500 tridents don't have the "famous" Elpida chips. I will keep on testing tomorrow (here in Spain is 3:00 AM...) I will leave memtest running overnight to burn the sticks.

Bye!

Well, as mentioned above, I'm quite sure they are using Elpida ICs, no other chip can reach those clockspeed at such low voltages, and the SPD settings are similar to my sticks.

For higher CL4 results the sticks need a bit voltage, they are scaling quite well up to 1.90V, maybe even a bit higher, but you should raise MCH voltage a tad to improve 4GB clocking. Up to now your results are in line with the results of OCZ's binned 8500 Blade LV.

[zsamz_]

I agree with you, they must be binning them pretty well like OCZ, I wouldn't be surprised if we saw 1150 and 1200 Trident kits in the not so distant future.



Think so too, I guess they are building up stock for higher-binned versions.



Well, as mentioned above, I'm quite sure they are using Elpida ICs, no other chip can reach those clockspeed at such low voltages, and the SPD settings are similar to my sticks.

For higher CL4 results the sticks need a bit voltage, they are scaling quite well up to 1.90V, maybe even a bit higher, but you should raise MCH voltage a tad to improve 4GB clocking. Up to now your results are in line with the results of OCZ's binned 8500 Blade LV.

@ least ic secret is out so ocz cant gouge people like they did with first batches
in canada 8500s went as high as 104$
theGSkill F2-8500CL5D-4GBTD are 20$ cheaper

[FrAN_oC]

I have been testing some 4 modules setups, and the result has been the same.

I can run 4 sticks altogether (8Gb) @ 550mhz 5-5-5-15 PL-6 at only 1.80v 1 complete memtest and 1 hour prime95 blend (I stopped it... no error). They doesn't seem to scale with voltage. This result is the same as yesterday tests, so... perhaps my NB is holding them down? Is strange to get the same results with 2 modules or with 4...??

I am currently running 1.45vNB at this moment... I tested with 1.55v, and I was able to reach 565Mhz with 4 sticks.

How much vNB would be safe on a x48 board?? I will give them a try on my other rig (P5K premium).

I think that are binned Elpidas. But not bad for it's price!! And they didn't cost me anything, because I RMA'ed my D9's, and with the refund I bought these sticks, a mouse-mat, some fans and AS5...

I will leave them burning today while I am working.

[-n7-]

Thanx for sharing.

Too bad these don't appear to be monster OCers in disguise...

These really could be a number of different ICs.

PSC can actually get pretty high at lower vDIMM, at least if it's better binned.
G.Skill's 1100 Pi's are low vDIMM also & use ProMOS (lower binned stuff).
It's definitely possible it's Elpida, but frankly, the results basically say nothing...they could be PSC, ProMOS, Hynix, Micron, or Elpida, at least based on what i've seen.

[18 is # 1]

the results basically say nothing...they could be PSC, ProMOS, Hynix, Micron, or Elpida, at least based on what i've seen.

It's definitely maybe possibly inconclusive.

[FrAN_oC]

And how can I find out? I don't want to remove HS... Perhaps comparing some setups... or voltages... with other known ICs...

ProMOS scale with voltage up to 2.1v
PSC don't like voltage
Elpida... tests on this ICs are inclocluent... (depends on binning)

So, the veredict says that these sticks aren't brilliant clockers, and we won't know what ICs do they have, because GSkill remarks their chips with an internal P/N, except D9s.

Now, the questions:

- How much voltage is safe for ProMOS chips? And for PSC? Elpida? Do you have information about how these ICs scale with voltage?

At this moment I am at work, and I am burning the sticks. Tonight... time to OC.

The strange thing is that I can run the sticks @ the same speed, no matter how many sticks I install on the mobo (1 stick, 2 sticks , 4sticks...) 550Mhz 5-5-5-15 seems to be the limit of my setup. I don't know what is holding my OC back. No matter what voltage I apply to the sticks (1.8, 1.9... ), they run at the same max speed. No matter what vNB... I am going crazy... VTT? divider? GTL? ... I will read some posts around to find out what is failing...

[BeepBeep2]

FrAN_oC, I hope I can add a little (just a little) to answering your questions above.

My Super*Talent sticks below are ProMOS, and they are stock at 2.2v. I've run them up to 2.25v for benches, and these are binned for low timings.
These are the highest I've had them CAS 4-4-4-10, validation only @ 2.225v: DDR2-983
http://valid.canardpc.com/show_oc.php?id=576496

Highest CAS 5-4-4-6 @ 2.2v: DDR2-1066 (limited by my CPU...)
http://valid.canardpc.com/show_oc.php?id=495783

I've booted 1100 5-5-5-15 @ 2.2v, but once again not stable. Again, thats probably my CPU.

Highest memtest86+ stable CAS 4 is DDR2-955 4-4-4-10.

Again, that's ProMOS...I've had these sticks for 2 years now. And I'm 13, turning 14 on the 3rd of July...

[18 is # 1]

Powerchips generally won't do cas4 past 900. The usually max out 1100-1140 5-5-5-15 2.0v
More voltage usually won't help them go higher although you can put 2.2v into them.

[LagunaX]

Good work Fran OC.

1100 5-5-5-15 PL 6 @ 1.8v - not bad, I might have to pick up a pair - $20 cheaper than the OCZ LV Blade 1066's...

[Don_Dan]

The strange thing is that I can run the sticks @ the same speed, no matter how many sticks I install on the mobo (1 stick, 2 sticks , 4sticks...) 550Mhz 5-5-5-15 seems to be the limit of my setup. I don't know what is holding my OC back. No matter what voltage I apply to the sticks (1.8, 1.9... ), they run at the same max speed. No matter what vNB... I am going crazy... VTT? divider? GTL? ... I will read some posts around to find out what is failing...

My guess is you're finding the limit because they are well-binned and the better chips were used in the other kits. Could you please post a screen of 8GB running at your max overclock with MemSet open? I would like to see the timings and compare them.

My Super*Talent sticks below are ProMOS, and they are stock at 2.2v. I've run them up to 2.25v for benches, and these are binned for low timings.
I've booted 1100 5-5-5-15 @ 2.2v, but once again not stable. Again, thats probably my CPU.
Highest memtest86+ stable CAS 4 is DDR2-955 4-4-4-10.

What you posted are 1GB ProMOS-sticks, I have two different revisions of those which are already a little different from their characteristics, but you can't compare them to the 2GB-sticks, as those are most likely totally different chips and you'll encounter more problems when overclocking the high-capacity stuff. Even if they use the same ProMOS chips nowadays for 2GB and 4GB kits, the 4GB kits will hit their limit way earlier most of the time.
Well binned ProMOS can reach up to 600MHz CL5 in 4GB flavour though, but that is very very rare ( ask G.Skill ).

Powerchips generally won't do cas4 past 900. The usually max out 1100-1140 5-5-5-15 2.0v
More voltage usually won't help them go higher although you can put 2.2v into them.

This is not entirely true too, but is only correct for lower-binned 4GB PSC kits. PSC can reach very high frequencies too, but of course only if it's heavily binned. In 2GB kits PSC can reach up to 600MHz CL4.


I don't want to offend you two, I just want to show that binning is a very important factor to look at when comparing 4GB kits, but the problem is that we don't know how well binned the memory is that we want to compare.

[FrAN_oC]

My guess is you're finding the limit because they are well-binned and the better chips were used in the other kits. Could you please post a screen of 8GB running at your max overclock with MemSet open? I would like to see the timings and compare them.

OK, i will set the ram @ 550 1.80v PL6. In a while I will upload the screenies.

[FrAN_oC]

4 Sticks (8Gb) 550Mhz 5-5-5-15 PL-6 1.80v

I didn't pass any test, because I passed them yesterday. I am preparing some tests with different latencies, so in about 1 - 2 days I will finish and publish them.