havent put it in yet :/
Printable View
havent put it in yet :/
his is a GT. on a GTS PCB.
Yeah how is that thing not a GTS? It's got a GTS PCB, GTS memory, and a "GT" core, except the GT core is just a downclocked GTS core. Same amount of shaders. Not trying to be a jerk but I've always been under the impression that a GT is just a downclocked GTS with slower ram? Just wondering what the ruling is on that one.
i give up
:rofl:
well if it makes you feel better I could use my back up card - but I don't think it's gonna help as it's nearly as fast!
http://www.itsjohnnybravo.com/Overcl...20(Medium).jpg
http://www.itsjohnnybravo.com/Overcl...ked%20temp.JPG
C'mon guys, if you buy a card, you need to look for the best out there. There has beeen many reviews in which people deal with GT cards on GTS pcb's. The card Johnny has is not the sole in it's category. Other manufacturers have released similar cards.
Im with massman on this one. I had the chance to get that card as well, but o opted against it. i believe the classification with a GTS is it must have a memory clock speed of 1100mhz???
anyways, i`ll get some results up soon, but these 775mhz core clocks you lot are getting on stock, im getting once i get to 1.6v, so there is no way this thing is going all the way. Thinking about sending it back, cause you have a large amount of electrical nosie even before you touch the clocks and voltage. memory isn't bad, but doubt such a weak core is going to get me anywhere.
- Paul
im ready for the ln2 session :)
ive found only pencil mods publicated, its working also, so dont have to solder the card, but will see after some testing...
http://fercsa.com/bala/8600gt.jpg
i was just kidding of course
there is nothing wrong with 8600GT with GTS PCB and RAM
it is what it is :up:
Look at your core.
G84-400 = 8600GTS
G84-300 = 8600GT
http://img187.imageshack.us/img187/2347/dscf1406yn4.jpg
I think I passed the test? :)
Let's all get back to the fun of volt mods and sold stuff shall we?
which one is the best
qimoda or hynix?
do you know what they are rated to?
I'm in, if I can still enter.
EDIT: I hope so, I just picked up a Asus EN8600GTTOP/HTDP/256.
question
i used grounding on the chokes where the RAM regulator is..........as i was getting accurate resistance readings
now one of those grounding points is actually reading 2.05v which is RAM voltage i think
should i change that grounding point or it doesn't matter as i'm gonna slap another wire on there to have RAM voltage readout
i've also measured two readins on the choke where the Core voltage regulation is and i am getting one reading saying 1.10v and other 1.25v[edit]NVM they are both reading the same >> that must be core voltage
which one should i go with..................
i'll keep playing i guess hehe
[edit] weird ... i am decreasing resistance and my vcore is dropping :S >> how do i fix that
ram is fine though
no problems there
Lol odd, sure you are decreasing resistance? Maybe turn the screw the other way? :D
lol i increase resistance all the way and you can hear that little clicking noise as you are screw the VR head meaning you've reached the end and as i am increasing resistance voltage increases and then stops at 1.28v (loaded) and will not increase any more as it's sitting on 50K
maybe i should slap a 100K resistor and increase resistance to get my mod happening :rofl:
this card is defying electrical engineering laws :rofl::rofl:
my core VR grounding point is also vRAM readout point
would that make a difference?
Dino that is all messed up :confused:
For the chokes (inductors) - neither is ground or at least should be, think of them as short lengths of wire, so if they went to ground they'd short. Always take your ground reference either from the correct capacitor leg or the ground plane.
For measuring voltages it's best to measure across the electrolytic capacitor that is closest to the component, that will give you an accurate measurement.
For vmods try and get the ground for the IC rather than from some far off reference. This will minimise the electrical noise it will pick up and also will provide better ground referencing, as some may be floating. In general try to keep vmods as close to the IC as possible as we are usually adjusting the feedback loop with means any noise picked up also gets amplified.
hi mate
so how do i find proper ground then
i just used a 100K VR and now my vcore is 1.32v lol
hang on I'll trawl through this thread and see if you've uploaded any pics ;)
Right!
http://img337.imageshack.us/img337/8...1582mi7.th.jpg
The vmod ground pin is number 7 which should be the top right and pin on both of the ICs in that pic - obviously test that before going gun ho with the solder iron ;)
http://img402.imageshack.us/img402/6...1553iu4.th.jpg
See the purple caps in the top right of this photo (NOT the silver one) those are the vgpu - measure the voltage VGPU same story for the memory lower down
http://img353.imageshack.us/img353/8...1543el8rh0.jpg
i have no idea how you did that
i can see what you are saying about pin#7 and that's a good start for me
how do you "test" if something is ground
also how the heck did you find out that you can mod some of those larger points for vGPU and vMEM
that's the kind of thing i want to learn so that i don't look like a retard every time i need a mod
for the capacitors - its pretty straight forward. See that strip of (-) signs down the sign. When you use an electrolytic capacitor that means that leg has to be the lower potential, so in this instance that refers to ground.
As for testing ground - easiest way is to use the blanking plate as it should be grounded. The problem is that sometimes (its rare but does happen) not all the grounds are "joined up" when the card isn't connected in the motherboard as the PSU will provide a common ground. On your multimeter there usually is a continuity tester which makes your multimeter buzz/beep when the probes short.Use this to check the grounds ;)
The inductor/choke will always be a good starting point for finding a voltage, but you have to be careful which leg you measure from - get more info here ;) From that you can usually see where the points all join up.
lol
went all gung ho and knocked it all up
i use blutac to secure the stuff the PCB
looks messy but it works for me hehe
http://img516.imageshack.us/img516/5563/img1664qu6.jpg
Dinos I'd recommend using the ground at the IC for the vmod but apart from that you should be good to go ;)
all works now
will have to slap a 50K back i think hehehe
:rofl::rofl::rofl::rofl::rofl:
been doing some testing
i can bench ~800MHz core linked with about 1.47v on that little stock cooler
absolute max RAM ~900Mhz
so LN2 this will easily hit 1GHz >> maybe 1.1GHz+
fingers crossed i can get my LN2 pot to sit properly on the core
here is a 3DMARK Vantage Performance run with 750/900 during early testing
http://service.futuremark.com/result...&resultType=19
P2429 on Skulltrail ROFL
NOOO!!!
LOL
Incorrect ground point for sure.
Put your multimeter on the continuity setting (where it makes the beep noise when you touch the tips of the probes together) then hold 1 to the backing bracket (where the VGA/DVI connectors are) and probe the card to find a point which "beeps"... usually a ground leg on a capacitor or the 4/6/8-pin power connector works fine...
EDIT
I think we should clear something else up here ;)
The point of a voltage mod is to create a divider in the circuit. The sensing pin (FB) sees X voltage and put placing the VR in parallel to the pin sensing voltage it draws away some of the power so it thinks the item getting X voltage (in this case the GPU) isnt getting enough and it overcorrects. If you do not use a point that is grounded then you have a lower potential to work with (or in your case a negative potential).
Got my new baby:up:
XFX 8600 GT Fatal1ty
http://img389.imageshack.us/img389/159/13092008ps7.jpg
is there a big difference benching 3dmark2k1 and 2k with 2*512 2*1gb or 2*2gb , considering they have the same specs(mhz, timmings)?
added some cap pr0n
http://img329.imageshack.us/img329/6409/capsqu0.jpg
i hope i did it right
reverse polarity as caps on the front of the card.....so these ones at the back were soldered + to -
does it matter that i used 1200 & 2200uF cap.........can it be too big??
here are the pics of front and back of the card for reference
http://img402.imageshack.us/img402/6046/img1553iu4.jpg
http://img353.imageshack.us/img353/8...1543el8rh0.jpg
Dinos those are soldered the wrong way round :D
really
youngpro told me to solder it in reverse :confused:
well you add capacitors in parallel, and you certainly don't wanna get the polarity wrong on electrolytic ones cus they can explode :down:
It should be negative to negative, positive to positive. I'm thinking YP probably meant that its the reverse from what you see at the front but who knows
Yeah you can actually have too much capacitance, I found this in the last comp with the memory. Best thing is to test with and without and see if you get any gains - the capacitors are there to provide instantaneous current in order to stiffen the voltage. Therefore there job is to cope with load changes, however they will not fix a voltage drop, just a monetary droop.
Mixing the capacitance values isn't that big a deal really :)
OT, know how to improve stability of the 5:8 divider on the RE to get memory running at 7-6-5-15 1T @ ~2GHz ? I keep getting a "checkVRAM" coder on the LCD readout :(
i'm only using it on vcore as i am going to feed monster vcore
i'll give it whatever it takes....so 2v+ here we come hahah
I think he meant to reverse (rotate) the capacitor so the leads match on the backside (+ to +, - to -).
went over 3.0v to mine ;)
EDIT - 730/1600/800 at full stock is my best run through aquamark 3 for testing so far.
Asus Silent edition, 512mb DDR3 qimonda 1.4ns
Fully stock card, two fans blowing on it (one on front, one on back) because the memory chips get retardedly hot...
CPU sucks, motherboard sucks (4x PCI-E only).. and absolutely not a single optimization or image quality change at all. No LOD, no mip, no nothing.
Manages AM3 @ 742/1836/800 giving ~ 107,500
what vore voltage do you have at stock settings
around 1.25v??
honestly, I dont know.... there's a bunch of caps that read as 1.33v and one that reads as 1.2v... memory is at 2.0v
i would say you're at 1.33 because of your clocks
Found it. 1.28 idle, 1.33 ish load ;) ... going through 3D03 now.....
that's even better than my core
what kind of cooler do you have on that card
Full stock.
Asus 8600 GT 512mb DDR3 silent.
http://images.google.ca/images?hl=en...-8&sa=N&tab=wi
~62c loaded at those clocks with ~25c ambient and a single low rpm 80mm pointed at the heatsink......
edit - 3d03 numbers.
ah ok
nice cooler
http://www.tech-hounds.com/review34/...USEN8600GT.jpg
better than my card for sure hehehe
It actually sucks pretty bad ;)
3D03 numbers added btw ;)
edit - and i'm running rivatuner temp monitoring int he background through another run just to see how 3d03 affects temps vs AM3.
edit - awesome, my CRT caught fire :D
Yes.
I guess 1024x768 @ 100hz was too much for it with a box on top :ROTF:
Its still working fine though. Took the box off and used a box fan to whisk the magical smoke out the door. Re-ran 3d03, +300 points.
64c through 3d03. Changed fans now though so -1 or 2 c maybe from 64 if I reran.
i usually read from caps after the inductor for whatever voltage...
my card is bytching out at 1.74v :(
single stage cooled
don't know if contact was good or not
this blows
MASSIVELY BOTTLENECKED by RAM speed
so the faster the ram the better the scores will be fellas
core speed really means very little once again
weird
i can bench nature with say 700MHz core linked 850MHz RAM with 1.9v to the core
yet if i bump it to 800MHz it crashed WTF
Sounds like the 8800GT's where GDDR3-2200 was possible with stock cores but 800MHz cores made GDDR3-2000 a fight.
yeah but you could still clock the core by lowering RAM
this is different
i have the same exact card.. stock sink and a 120mm fan @ 100cfm, 1.6v 972 mhz core and it does not hit 60c... going to remove the sink here soon and toss the chiller on her.... only thing is now i think i am at the power circuit limit.. since no matter what i do now i get the instant line artifacts when i change to anything above 972 mhz.. even with 1.8v...
and again the mem is :(
anyone know if all these cards are cold bugged?
happy modding!!!
how many volts are u running dino? and 945mhz the max u can do?
Possible that the core is exhibiting a coldbug like the 8800GT and the 9600GT? Perhaps try naked die without TIM and see if the warmer core helps.
Card arrived today. I'll see what I can do on air tonight on my e8600 rig.
maybe mount it on phase or something?
I'm getting around to doing my vmods sometime this week ;)
i was running 1.72v loaded there
i have a problem which i don't know the answer for...........i could not have too much or too little voltage at certain clocks...............it was very strange for me
i was in rivatuner/ati tool doing 1100MHz core and i think i even managed to do some '01 nature at 1050 or 1100 but i could not repeat it later as i didn't make notes on voltage used
i could run nature '01 at 1Ghz core but lobby will crash almost instantly
the setup is perfectly stable at 4.8GHz so that wasn't giving me issues. It was the card.......
i could for example start off with lower clocks and increase the volts a lot just for the sake of seeing if it could handle higher volts................... >> 1.9v loaded no probs...evap temp stays at around -48C and if there is a 20C delta i should be in a safe zone...i had a ~15C delta at stock clocks and volts
but if i try to use 1.9v volts with slightly higher clocks...say 800Mhz linked.....it bombs out instantly. i get a black screen most of the time....the pc seems to be running as i have FF codes and it doesn't reboot...just black screen....the volts on GPU change to idle volts (so 1.85v roughly)...........i would also get freezes occasionally too
Sounds like the video driver is dropping out from the clock changes. Does your keyboard still respond with the numlock and capslock buttons? That'll tell you right quick if it's still running video-less.
keyboard
it does man
sometimes it even comes back alive
So it's dropping out of 3D due to instability when linked. Guess your core has weak shaders?
First card I got was DOA, so have to get another.
posting a few pics of my card so far... :up::cool::D;)
http://www.buydurango.com/josh_files...t/DSCF7687.JPG
http://www.buydurango.com/josh_files...t/DSCF7684.JPG
http://www.buydurango.com/josh_files...t/DSCF7685.JPG
Nice :D
i luv this phred
Dinos - does your card have a power connector on it? If not, add one.
I might get time to work on my card tomorrow if we dont work late...
EDIT - Can you tell me what exact VR's you used there james? I'm laz... er working. :D
hey gomeler
shouldn't you be updating leaderboards so that we know what's up......it helps keep the competitive vibe going too i think
http://xs131.xs.to/xs131/08383/jamesbong962.jpg
How much did it help you with clocks? :lol:
oh man that's gold
hammer next to the card
love it
Scores updated :up: Need to pick up my roll of solder and get started I suppose, been super lazy as of late without LN2 :shakes:
i hope you guys have good luck with your cards
it willbe super scary seeing the scores with high RAM clocks
shame andy's not a part of the comp......he's such a funny bastard :D
Pretty certain I sent him a PM on XS inviting him a long with a slew of other people, maybe he'll turn up :)
LOLOLOLOLOLOLOL!!!!!
had to crack her a few times to get her to boot... j/k...
nah my roomate actually had JUST borrowed my hammer to put something up in his room and thought that was the best spot for it. LOL....
@STEVIL here you are... :up:
http://www.buydurango.com/josh_files...gt/8600mod.jpg
Man, I would have loved to take part in this competition ... damn. If days would have had 48 hours instead of 24, I'd find time, I guess. In any case, Johnny asked me if I had LOD scaling graphs of the 86GT card as I published an 86GT overclocking article on Madshrimps a while ago. Well, officially I have no graphs, but I dug out my data XLS file and I did found some LOD scaling data. I'm not sure if I'm helping someone with this, but here they are:
http://i154.photobucket.com/albums/s...lodscaling.png
http://i154.photobucket.com/albums/s...k03scaling.png
3Dmark03 might have been biased by the low cpu clock. GT1 turned out to be the test to tweak in order to get the highest boost.
http://i154.photobucket.com/albums/s...an-/XS/GT1.png
http://i154.photobucket.com/albums/s...234scaling.png
Thanks for the data massman, incredibly helpful and saves us a lot of time :up:
looks like i was using the WRONG driver too :D
have a look at your CH and then look at mine
i would have an easy 60-61K somewhere :)
LOD is for handicapped people! ;)
Hahaha STEvil! :D :up: