Printable View
You know what is weird is that I am getting way more condensation with this card than with with 9800xt, even at 680/540 and 1.69v on the core. My r507 prommie is almost overkill for this card. It just seems like it doesn't put out anywhere near the ammount of heat my xt did @ 2.15v/3v....for obvious reasons.
The card is still very cold even after like 10 straight benches.
yeah because it generates so much less heat.
Have anybody news about the vmem mod? It looks so far like only Vgpu (and Vddq + Vddr perhaps) works.
bump for vmem mod....
k|ngp|n,
I read somewhere that you were able to do the vmem mod (at least I saw a pic of some mods close to the memory)...
Were you succesful with the vmem mod??
If so, please share the wealth:D
:toast:
Minnyboy
good pic for vgpu
i really like my Minolta Dimage A2 :D
good mod :D
http://sites.rapidus.net/maxbleau/x800%20vgpu.JPG
Nice ram but isnt 500Mhz at cas4 about the same speed as 250Mhz at cas2?....as far as total throughput?
As a curiousity, anyone ever written a video memory test along the lines of SiSoft sandra memory tests for CPUs? Just move blocks hither and fro... and time the action?
can you show me how to do the vmem mod please.
Vgpu, vmem and overvolt mods are the same for X800pro and X800XTPE?
Thanks
Yes, mods are the same on both X800 series cards. Can someone confirm that for the vmem mod, it's pin 5 to ground of the chip on the left from the picture above. What size VR also?
Found it somewhere else already...vmem mod is @ pin 5 to ground with a 20k VR. @ 20k it increases vmem .08 or so.
Thanks stasiu:)
You are right about the vmem
may you indicate, please, were is the measurement poin for vmem?
Thanks
Look at the last pc ice pic and on the left side it's where the green wire is soldered....
Thanks
well i did the vcore mod for the X800XT, the damn solder wouldnt stick to the resistor! Almost burnt off the resistor next to it, but its fine. Damn solder wouldnt stick but eventually it worked ok.
Crunchtime, use expensive solder! With flux already built in. :)
My Vmem mod works, but I am getting far worse results in Mem Overclock! WTF?
Card is a X800 VIVO modded to 16 pipes. Mod works - I can raise the voltage. But the higher I set Vmem, the carppier is my MemOC. :confused:
It's not an issue of temp. I have BGA Freezers put on and a 80mm fan blowing over them.
I'm getting worse mem. clocks as well since doing the vmem mod. Before I could do 625 no problem...now it artifacts like mad.
Quote:
Originally posted by pc ice
kinda wierd i could no way do 640 mem before on stock voltage but after doing the mod and removing it..i can
Strange :confused: .....but I hope the same happens with my card when I remove it. ;)
OK that's weird. Anyone did the Pin5 to Pin7 method for Vmem?
Maybe i can give a little hint. I cannot solder myself.
The Vddq must be raised together with the vdd (memmod).
Overclockersforum:
hxxp://www.ocforums.com/showthread.php?t=309528
Picture of vddq mod:
hxxp://www.ocforums.com/attachment.php?attachmentid=31428&stc=1
Picture of vdd and vddq mod, different view:
hxxp://www.ocforums.com/attachment.php?attachmentid=31429&stc=1
Important words i think:
"Partially right. With previous GDDR-I and GDDR-II memory running Vddq high and/or above Vdd
would ding the memory OC. With the exception of Samsung 8E's which like the Vddq a little hotter
than nominal spec (and right where ATI had it BTW) Vddq at or a bit below nominal spec worked
best in most cases.
This new Samsung GDDR-III is a different animal and likes it's Vddq run hot and above Vdd (again
exactly how ATI have the memory voltages setup stock) which is against decidedly against
Samsungs spec that Vddq should always be lower than Vdd. ATI knew what is was doing lol!
These new Sammy GDDR-III chips typically do not respond to Vdd increases without a
corresponding increase in Vddq to go along with it. Kraigs Vdd is running right at the high end
of samsungs nominal spec for it and Vddq is running .050 volts above it. Fairly soft really
considering Samsung DDR-I RA chips were often run with VDD a full 1/2 volt above nominal spec.
Note that the GDDR-III chips ran a bit cooler than say Samsung 8E chips at stock voltages but
heat dissapation went up rapidly and radically with increased memory voltages. TM's have no
place on Vmodded Samsung GDDR-III chips"
Can anyone post good pics of measuring points for the vgpu and vmem?
Is there any guide already for the vddq mod???
Is the value of the pot needed to the OVP mod the same that ppl use in 9800xt...20k?
I guess i read somewhere 50k...confused now hehe!
In vr-zone shamino says 500k.
Thks in advance!
I have done the pencil Vgpu mod and it worked a treat :) I have the card running 549 Core from about 500 before the mod.
However the memory on my card will only do 460 at stock voltages. I took a reading of the vmem voltage and found it to be at 1.93v :(
If I do the Vmem mod I can push it to 560 with about 2.1v, but I still can get graphical glitches in Soldiers-Heroes of world war II and a few other games so I know it is not 100% stable :(
I am thinking that Vddq mod might settle things down. Especially if the Vddq is low like the Vmem was :(
I find it very odd that I have seen quite a few posts where people can't get the 1.6n rated ram to run at anywhere near its rated speed, myself included. Maybe the ram has been crippled in some way? Or there was a really bad batch put on some cards?
Is the Vddq mod the same mod as Vmem ? I can't really find much info on it. Is it the same style pin 5 to ground mod that it looks like to me?
On a side note I can also confirm that the pencil vmem mod does not work :( At least not on my card ;)
Cheers
the CAS is a delay timings, which they are equal, but 500MHz gives more bandwidth ...
Quote:
Originally posted by uwackme
Nice ram but isnt 500Mhz at cas4 about the same speed as 250Mhz at cas2?....as far as total throughput?
As a curiousity, anyone ever written a video memory test along the lines of SiSoft sandra memory tests for CPUs? Just move blocks hither and fro... and time the action?
Thats a beautifully done mod. Hats off to the man with the iron skills :toast:Quote:
Originally posted by arrgh5
Maybe i can give a little hint. I cannot solder myself.
The Vddq must be raised together with the vdd (memmod).
Overclockersforum:
hxxp://www.ocforums.com/showthread.php?t=309528
Picture of vddq mod:
hxxp://www.ocforums.com/attachment.php?attachmentid=31428&stc=1
Picture of vdd and vddq mod, different view:
hxxp://www.ocforums.com/attachment.php?attachmentid=31429&stc=1
Important words i think:
"Partially right. With previous GDDR-I and GDDR-II memory running Vddq high and/or above Vdd
would ding the memory OC. With the exception of Samsung 8E's which like the Vddq a little hotter
than nominal spec (and right where ATI had it BTW) Vddq at or a bit below nominal spec worked
best in most cases.
This new Samsung GDDR-III is a different animal and likes it's Vddq run hot and above Vdd (again
exactly how ATI have the memory voltages setup stock) which is against decidedly against
Samsungs spec that Vddq should always be lower than Vdd. ATI knew what is was doing lol!
These new Sammy GDDR-III chips typically do not respond to Vdd increases without a
corresponding increase in Vddq to go along with it. Kraigs Vdd is running right at the high end
of samsungs nominal spec for it and Vddq is running .050 volts above it. Fairly soft really
considering Samsung DDR-I RA chips were often run with VDD a full 1/2 volt above nominal spec.
Note that the GDDR-III chips ran a bit cooler than say Samsung 8E chips at stock voltages but
heat dissapation went up rapidly and radically with increased memory voltages. TM's have no
place on Vmodded Samsung GDDR-III chips"
But a value for that VR would be nice ;) I read 770 ohms from pin 5 to ground, so would 20k be a good starting point?
I take it the Vddq measurement point is the bottom edge of C141? Or is it off C142 (which is directly connected to the FET drain)?
Vddq measurement...
http://img56.exs.cx/img56/2067/x800x...urement.th.jpg
No matter what capacitor, choose one.