So you pencil the yellow part in the NEW pictures and not the blue part circled in the OLD pictures that were posted here?
Just Tested the new Pencil Mod and it works great its easier to apply
Here is a image for anyone that is confused about the Pencil mods
The Mod is exactly the same for the 3850 the only difference is the
Resistance measuring point. Its the Right Blue Square and the Screw Above 'Q1252/Q1612/Q1602'
Top left Corner of the image
Last edited by Rumon; 12-28-2007 at 09:56 PM.
Gigabyte 965P-DS4
E6600 (494x7)@3.45GHz 1.36v
corsair CM2x1024-6400 CL5 2x1GB @988MHz 5-5-5-15 2.0v
Sapphire HD3870 @958.5/1251MHz 1.49v
Accelero S1+120/80mm fans
QX 9650 5ghz with 1.55v 4.8ghz with 1.5v 24/7 in a VAPOLI V-2000B+ Single stage phase cooling.
DFI LP LT X-38 T2R
2X HD4850's water cooled , volt modded
Thermaltake 1KW Psu
4x Seagate 250GB in RAID 0
8GB crucial ballistix ram
I have a 3850. Whats the benefit if this volt mod if Im using stock cooler? = None?
QX 9650 5ghz with 1.55v 4.8ghz with 1.5v 24/7 in a VAPOLI V-2000B+ Single stage phase cooling.
DFI LP LT X-38 T2R
2X HD4850's water cooled , volt modded
Thermaltake 1KW Psu
4x Seagate 250GB in RAID 0
8GB crucial ballistix ram
What Vcore is quite safe for 3850 ? now I have ~1.45V but I want more
I don't get it -- I can't get a pencil mod to work on my 3850. My analog multimeter is a little unreliable, but that still shouldn't stop a voltage increase. So, according to this mod, I'm supposed to be able to draw a tiny line of graphite across those two points and get a resistance decrease enough to dramatically increase voltage?? I don't really see how -- I've never done a pencil mod, but it doesn't seem a number 2 pencil is doing much when I'm attempting to put down some graphite on such a tiny spot.
Maybe someone can enlighten me with specific details of application technique of the actual mod as oppose to just showing pictures of the location. How do you apply the pencil, did you shave the tip to be a tiny flat point or did you leave it like a standard pencil??
Thanks.
And where the heck is the new Atitool, it would be nice to have new voltage control and monitoring..
pdiggs, not a tiny line, keep grinding on that resistor with a 2B, or just move the pencil a bit with a 4B, you should get good results.
I tried it with a 2B, but had to keep grinding a lot just to get to 1.4v, and if you grind too much and find a lot of small particles everywhere, you have to blow it away to get constant voltage, or it will fluctuate.
CPU: Intel Pentium D 925, 3.0GHz @3.82, 40/60c
RAM: Spectek 2048MB (512x4), DDR2-533 @510, 4-3-3-9-12
MB: Gigabyte GA-8I945P-G
GFX: Chaintech 8800GT 600/1500MHz (1.05v) @800/2000 (1.19v), 512MB 900MHz (1.9v), Zalman VF700-Cu, 45/80c
Monitor: IBM 21" P202 @1600/70Hz
PSU: Task International 450 Watt, 12v/26A
+1 Vote for a reading voltages & resistance(especially resistance) How-To on these cards. Basics. I have to send my card back...its now apparently fryed.
2500k stock (for now )
Biostar TP67XE
G.SKILL 2x4GB Ripjaw 1600
AMD 6950 --> 6970 2GB
Antec TPQ-1000
Example of the mod
and some results.
2005 need more cpu power to get a decent score
Couldn't you please try to test Crysis with/without the overclock? I'd like to know if it boosts Crysis's performance .
P.S.: Great clock, good job .
Last edited by Onecar; 12-31-2007 at 02:08 AM.
For some reason I usually measure about .2V difference between the 2 marked vGPU points.
For example I measure 1.6v at the Black point and 1.41 at the green point.
Which is the real vGPU? At stock without any mods the black point says 1.45v (!!!) and the green one says 1.27v
BTW I have a Powercolor 3850pro 512mb
Last edited by gurusan; 12-31-2007 at 02:18 AM.
|| 2500K @ 5GHz 1 thread, 4.8 2 threads, 4.7 3, 4.6 4 1.284V ||
|| P8P67-M Pro || 8GB @ 2133MHz ||
|| 5850 @ 1000/1225 || XFX 650W || Silverstone FT03B ||
|| 37" LCD TV || CM Hyper 212+ || Samsung 2.1 Soundbar ||
I would say the black would be closer to the truth. could you get a little further back so we could see more of the card, there may be a better place to check.
QX 9650 5ghz with 1.55v 4.8ghz with 1.5v 24/7 in a VAPOLI V-2000B+ Single stage phase cooling.
DFI LP LT X-38 T2R
2X HD4850's water cooled , volt modded
Thermaltake 1KW Psu
4x Seagate 250GB in RAID 0
8GB crucial ballistix ram
but 1.45v stock? also when it's at stock I have a 780mhz limit on my clock....that would be terrible for 1.45v
and ok I'll have upload a better pic in a bit....doing some 3dmark runs
Last edited by gurusan; 12-31-2007 at 02:47 AM.
|| 2500K @ 5GHz 1 thread, 4.8 2 threads, 4.7 3, 4.6 4 1.284V ||
|| P8P67-M Pro || 8GB @ 2133MHz ||
|| 5850 @ 1000/1225 || XFX 650W || Silverstone FT03B ||
|| 37" LCD TV || CM Hyper 212+ || Samsung 2.1 Soundbar ||
|| 2500K @ 5GHz 1 thread, 4.8 2 threads, 4.7 3, 4.6 4 1.284V ||
|| P8P67-M Pro || 8GB @ 2133MHz ||
|| 5850 @ 1000/1225 || XFX 650W || Silverstone FT03B ||
|| 37" LCD TV || CM Hyper 212+ || Samsung 2.1 Soundbar ||
What would be the safe temperature on the Voltage regulators on the 3850 ?? I changed the stock cooling ,put some decent radiators there ,and in FUR i got about 60-65 degrees on that spot(finger-measured ).Is it ok?
i7 920 | Asus P6T Deluxe V2
Pali GTX 460 768Gb DDR5
3*2Gb TeamXtreem 1600Mhz DDR3 cl8
Corsair HX 620W
E66 + Plantronics Voyager 520
[IMG][/IMG]
let us know what the readings are. one or all of the points should be a read point.
QX 9650 5ghz with 1.55v 4.8ghz with 1.5v 24/7 in a VAPOLI V-2000B+ Single stage phase cooling.
DFI LP LT X-38 T2R
2X HD4850's water cooled , volt modded
Thermaltake 1KW Psu
4x Seagate 250GB in RAID 0
8GB crucial ballistix ram
scores 20020 in 2005 and 12216 in 2006 3DMarks on 3850 512mb@877/1090 and E6300@3400 are ok ?
Cat 7.11 WinXP SP2
Thanks so much jimmyz!
I currently have it pencil-modded so it won't show 1.27v...but it was as I suspected.
The black point I was measuring first is the incorrect vGPU and displays a few 10ths of a volt higher than the actual vGPU. The green circled point and almost all of the other points you showed me display my vGPU.
1.54v idle and 1.57v full load. Have it 100% stable @ 958.5core/1098mem
Temps are around 39-40c load under water with a MCW60A.
thanks again!
Now, has anyone figured out a vGDDR mod? This mem already does 1098 which is great, but it would be even better if I could just squeeze a bit more out of it...something like 1150-1200
|| 2500K @ 5GHz 1 thread, 4.8 2 threads, 4.7 3, 4.6 4 1.284V ||
|| P8P67-M Pro || 8GB @ 2133MHz ||
|| 5850 @ 1000/1225 || XFX 650W || Silverstone FT03B ||
|| 37" LCD TV || CM Hyper 212+ || Samsung 2.1 Soundbar ||
I was trying to read resistance from the 'measurement point' next to the new pencil mod...machine restarted and I have blue lines. Beautiful. RMA time, huh?
2500k stock (for now )
Biostar TP67XE
G.SKILL 2x4GB Ripjaw 1600
AMD 6950 --> 6970 2GB
Antec TPQ-1000
When you decide to mod a card you assume responsibility. you killed the card, now be mature about it.
never measure resistance with power to the card.
QX 9650 5ghz with 1.55v 4.8ghz with 1.5v 24/7 in a VAPOLI V-2000B+ Single stage phase cooling.
DFI LP LT X-38 T2R
2X HD4850's water cooled , volt modded
Thermaltake 1KW Psu
4x Seagate 250GB in RAID 0
8GB crucial ballistix ram
Bookmarks