GEIL Ultra PC-3200 Golden Dragon

[Hans]

hmm... interesting looking packaging, "Advanced WLCSP package".

[Hans]

The chip sure is small, even smaller than Kingmax's BGA chips.

[Hans]

I can tell you one thing for you,

the Golden Dragon hates advanced timings, e.g. 2-2-2-5 or 2-2-1-5. It failed to POST @ 195 MHz @ 2-2-2-5 timing @ 2.9V.

If a more "relaxed" timing is used, 2.5-3-3-5, then it would go over to our current max FSB of 218MHz.

Question:

Do you think we should test the module according to SPD value?
If that method is used, then we can't directly compare modules, as different timing is used.

Do you think the ability to run advanced timing is important, e.g. 2-2-2-5, etc?

The responses to these questions will be used for our DDR Guide III.

[dustyslapper]

Well Hans, interesting point. I have the Abit NF7 2.0 which has some handy preset memory settings in bios. What I discovered last night when I bothered to check is that at different fsb these settings are different.

Forinstance at 133 the 'turbo' timing is OC correct: 2-2-2-5. At 220 fsb this balloons out to 2.5-3-3-8. Obviously this is user adjustable, but this is the most aggressive preset option in bios remember. Clearly the Abit boards are pointing towards faster timings at slow speeds, with the corollary being that slower timings at insane bus may not be such a bad idea. However obviously a poll accross other DDR platforms would be the right thing to do. What is the canterwood up to forinstance?

The roundup should also reflect on what the high end and low end ram manufacturers are doing. If people like Corsair are still making memory that can do 2-2-2-5 at 250fsb then clearly your roundup should be able to reflect this. Maybe you should include a section of overclocking at the top SPD speeds, as well has fast and slow settings.

A comparable Sandra bench at the SPD settings would be a good thing to compare at the end of the roundup.

Goodluck mate, the roundups are damn fine resource. Quick question - what mobo are you planning to use on this? i hope you got bus to burn... :p If you need help give me a yell ;-) I'd love to have a go with some nice stuff at 500ddr. Brmmmm brmmmmm!

[Stove]

Similar thing happens with the board I have. So much so, that aggressive overclockers have found that there is a LARGE difference between setting FSB in bios, and overclocking the clock generator using SetFSB software in overclocking. Setting a fast FSB in Bios causes the BIOS to relax ram timings, whereas using SETFSB kept hard timings, and hence was faster.
both are important, and so therefore maybe the absolute overclocking figure should not be the only quoted one :)
so quote 2 figures: Overclock at most aggressive timings, and overclock at rated timings

[TooMuchCoffeeMan]

I meant to reply earlier but have been a bit busy.

My experience with golden dragon has been interesting, it seems that when I go to relax the timings in order to achieve higher FSB that it has the opposite effect! It became unbootable and was in no way helpful.
This goes against current theory, we are all used to using less agressive memory timing in order to get a little more speed out of it.

The sticker on the modules states cas2, 6 3 3 and that seems to be how they are best operated in my experience.

I noted this in my reader review of PC3200 Golden Dragon that I was able to achieve high FSB of 234Mhz at those timings.

I am not sure what to think, was it just me? I would be interested to volt mod the motherboard and see how far I can really push it because I do not think the ram was holding me back at that point (many intel users are getting higher FSB and memory scores on the right motherboards)