what are these rams?
older then FD3 and FD5?
maybe Fatbody D9 :)
if so ,youll need more voltage
Printable View
what are these rams?
older then FD3 and FD5?
maybe Fatbody D9 :)
if so ,youll need more voltage
cpuz spd says serial number:840034c6
jedec #1 400mhz 4.0-4-4-18 trc 24 1.8v
please make a screen from spd table :)
my last post contains everything of interest in cpuz spd, all the other fields are greyed out
however, the stick in stick in slot # 3 has serial number 8a003d7d
only other stuff in cpuz spd that are not greyed out are the obvious, like crucial technology, 1024mb module size , pc26400, correction none, ddr2
spd timings table only has jedec #1 400mhz 4.0-4-4-18-24-1.8v
Yeah, as I said already, serial number is different on all sticks.
I'm pretty certain they use fat body D9, maybe some of the other old member that have such a kit will post and confirm my guess.
You can use MemSet to get a glance at the SPD with all timings and subtimings.
to save time can you tell me how to post screenshots in forums?
spd.bmp (471 KB)
When you post, there is a second part after the box where you write your message, it says "Additional Option", there you can find the button "Manage Attachments". If your screenshot is bigger than 200kB you'll have to upload to an imagehost service.
only persons i know of that have done extensive fatbody tweaking on intel systems are fugger and hipro5 i think, there maybe more that i dont know of
check post #32, it has link to entire spd table in memset, apparently there is cas2 in jedec#2, i thought tighter than cas3 is impossible with ddr2 lol
That SPD is funny indeed! :D
Subtimings at stock speed are quite tight, those must be using fat body chips, I can't think of anything else.
It might be worth to check out their old threads, here is another one I remembered that was posted not soooo long ago.
Just get some paper, a pen and Memtest and you'll figure your memory out! Have fun! :up:
so these respond to voltage, not loose timings?
the memory in your link are 5400ul 2x512, wont they behave differently to crucial ballistix 2x1024 even if they both use the same type of ics?
all d9 dont mind tight subtimings
and your fatties should do 1000mhz cas 4 easy just need volts
you gonna be pushin 2.3 v
my fatties actually prefered cas3
i did pump upto 2,7 into them and they still alive lol
its quite possible tou have d9gct and they need serious volts to scale
anything over 2.2v 24/7 for d9 is a risk;)
If it's the same chip, the bigger capacity sets will just stop scaling sooner than the smaller kits.
Yes, they do react to voltage, lots of voltage, if they scale any further at the same timings but with a notch more VDIMM, they should be some sort of D9, fat bodies are highly probable. Just remember to put a fan over them when pushing up voltage.
I'm off to do some work now, if you need more help I'll be online again tomorrow.
Or just listen to zsamz_, he's a good guy, his only problem is that he doesn't like Gigabyte boards... :p:
Don't you think those are fatties?
heres a link to what my mobo actually defaults my memory to, it might be of some use
memset.bmp (515 KB)
i guess later on i will try setting 2.25v in bios and hope my board doesnt overvolt it much past 2.3v, i already know 2.15 = 2.18 and 2.2 = 2.26
if the chips are big (19x13mm, take the most of pcb) then those are fatbody for sure
if you can't visually detect them - test them, fatbody hits roughly same clocks with 5-5-5 as with 5-3-3
450-500 with 4-3-3 (~2.3v) or with 5-3-3 (~2.2v) is the best daily fatbody setting
could it be my mobo just doesnt like my memory? i heard d9s especially fatties dont like gigabyte boards
i find it annoying to no end that cas latency can only be changed in bios :/
edit: i set ddr2 voltage to 2.2v, which overvolts to 2.26v and increased cas latency to 5 in bios. i set main timings to 5-3-3-12 with memset 4.0 didnt touch any subtimings and it runs just fine at 400mhz, now i just need to start raising the fsb.
so if it scales anywhere near 450 its definitaly fatbody?
check post #188 page 8
http://www.xtremesystems.org/forums/...t=77442&page=8
according to that .16fa is d9dqw thats fatbody right?
http://ramlist.i4memory.com/ddr2/
according to this d9dqw is one of the 4 possibilities
i bought them mid 2006 off ebay, cant remember for how much but it did cost me an arm and a leg at the time. i didnt think much of them as they didnt scale much at the rated voltage and timings, so i switched to some ocz reaperX with powerchip ics as soon as they became available but i kept the crucial ballistix around. recently i sold off the ocz reaperX and went back to the crucial ballistix
I picked up some 16FD5 that was in a HUGE batch of 8FE5, and it clocks horribly. It clocks like the single sided stuff +20mhz or so. Doesn't respond well to voltage either... My other set does 1200 easy w/2.25v
good news! my memory actually scaled higher with 5-3-3-12 and default subtimings 2.26v 450.5mhz qx9650 3378mhz super pi 1m 13.750s
super_pi_1m_13_750s.bmp (2.7 MB)
if it actually clocks higher at 5-3-3-12 and a slight voltage bump its got to be d9dqw right?
it could possibly have more in it, i didnt push the fsb until it bluescreened this time because i was so glad to achieve 450mhz at tight timings
this memory has been to this voltage in the past and never reached this speed, perhaps it just likes high volts, loose cas and everything else tight
after close inspection without taking off the heatspreaders the memory chips appear to take up about 2 thirds of the pcb. that and they seem to clock just as well with x-3-3-x timings im positive they are d9dqt/d9dqw aka fatbodies.
btw it has a black pcb making it hard to see, luckily i had a flashlight
edit: heres something from anandtech same memory from early 2006 they are d9dqt, if you go to page 2 they have .16fa on the heatspreaders same as mine
http://www.anandtech.com/memory/showdoc.aspx?i=2700&p=1
edit2: heres a review of .16fa sticks doing 3-2-2-8 timings but at low clocks
http://www.xbitlabs.com/articles/mem...dr2-800_3.html
Those pretty much confirms they are fat body chips, have fun with them and pump up the volts! :up:
With more volts you might be able to get them to 500MHz CL4-3-3-10, which is still a rather nice speed.
super_pi_sub_14_sec.bmp (2.7 MB)
4-3-3-10 is already a reality with tight subtimings (the same ones in the link you showed me)
http://www.xtremesystems.org/forums/...d.php?t=161861
only reason i didnt go further was because my cpu was getting too hot, i guess i will have to either lower the multiplier and say goodbye to any chance of beating my current super pi 1m time or change the fsb/memory ratio and keep the memory voltage up 24/7. or get a better cpu cooler. anyway i had my voltage set to 2.25v but the bios and everest were both reading 2.19v, same as if i set it to 2.15v. since it never scaled like that in the past i dont believe it for a second.
edit: just how hard is it to get fatbodies to work with 4-2-2-10 timings at 450+(900mhz effective) ? does it require excessive voltage?
I didn't notice you edited your post until I got your PM, sorry. :(
1M is not a good test for memory, it fits into the CPU's cache and your score will not change much from running looser timings. If you want to see the memory performance, run 32M or Everest bandwidth tests.
For 24/7 I would go for some more CPU speed to be honest, 3.4GHz is not really much. And getting into FSB or temp problems when testing memory is often the case when you run a quad, that's why I use a E8600. ;)
its summer over here and im running the qx9650 on the stock cooler for now, at 3.4ghz after the super pi 1m bench i tried running orthos, it never became unstable however i had realtemp running at the same time, after 2 minutes i stopped orthos and rebooted, setting everything except memory voltage back too stock, which i then set from 2.25 to 2.15 in bios. i just freaked out when i saw the qx9650's hottest core hit 85c!
i use a quad because i also like benching 3dmark06