Broken Link HDCHOPPER.. here is fixed link ;) http://www.nordichardware.com/skrive...16/ToolLib.zipQuote:
Originally Posted by HDCHOPPER
edit - I tried the patch, you dont need it with the current version. Also AMD GPU Clock Tool doesnt work with XP-64 so i'm still up **** creek without a paddle. :(
Hey guys, hoping you could help me out.
Try not to flame me too hard for being a bit noobish.
Every time I try to run the commands in the first step, I get the below errors. I know I'm doing something wrong, what is it?
http://img264.imageshack.us/img264/284/cmdfj2.jpg
Thanks for any help.
You have to navigate to your RivaTuner directory buddy.
(Whoa, I just restated myself?) EDIT BUTTON IS YOUR FRIEND
cd C:\Program Files\<Rivatuner directory>
(cd == Change Directory, if you're wondering.)(Because Rivatuner.exe isn't in your C:\Users\Daniel folder now is it? xD)
then do Rivatuner.exe <whatever>
:up:
LOL oh God, I can't believe I thought right past that.
Thank you.
Alright, for the first step, RivaTuner.exe /ri1,70,1A gives me the correct return. It gives me "I2C 01 70 1a :0a"
Step two gives me
01 70 15 :41
01 70 16 :41
01 70 17 :41
01 70 18 :41
Is this right? Thanks again. :)
thanks STEvil :up:
knew eye should of checked that link :coffee:
ATi Tray Tools has support for adjusting voltage. I dont know if it works beyond the stock 1.25v though because I uninstalled it after I found that it wouldnt overclock the second core of my X2.. like everything else out there.
Hey guys how would one do the 2nd x2 in crossfirex?
nevermind just found it out...
If you want to do the second hd4870x2 in crossfire you must use
RivaTuner.exe /sd3 /wi1,70,18,4C /sd4 /wi1,70,18,4C
of course that is the values for 1.4v
assuming your working directory is c:\program files\rivatuner v2.22 you can use this batch code that I slightly modified from the other guys post to do both gpu's. I take no credit in this other than adding the strings for the 2nd hd4870x2 in crossfire!
Anyone knows a good tutorial for Windows PowerShell? I think all this reading and setting volts with RT could be automated in a little script without much problems. Or would a CMD script be a little more universal? (dunno how many ppl have powershell installed)
Edit: NVM, found something.
OK someone explain this.Quote:
Originally Posted by Gir92
Why are there four different registers all returning the exact same voltage?
41=65 decimal Vid=1.263v
Seems like the people with 4870 x2's were changing the "18" register, while on my "old" model visiontek 4870, changing the 18 register does not change the readout in gpu-z or Riva; But changing the 17 register does.
But like everyone else has found out, no improvement in 3dmark stability (firefly forest always locks up randomly after 830 core, and -sometimes- even canyon has done it).
But notice this posters findings:
FOUR different registers have 1.263v.
What exactly does this mean?
What happens if you change all four (15-18) to 1.35v instead of just the only one that shows the change in Rivatuner or GPUZ?
Anyone know if that will make a difference or will 3dmark still crash? (this goes for X2 owners too).
Maybe those different registers are for power states, so this card you have isn't supposed to be switching between states, which would mean all states supplied by the controller would be the same. :shrug:
You can "ask" VT11xx to report OCP/OVP conditions by reading its register 13. Bits 0 and 1 are OCP/OVP states. Take a note that these bits are read-only status bits, you cannot change them.Quote:
Originally Posted by japamd
Surprized to meet you here :)Quote:
Originally Posted by Falkentyne
It means just one thing: separate 2D/3D voltages are disabled on your system (either in BIOS or via software, e.g. overclocking 4800 series via RT/ATT/AT or AMDGPUClockTool disables PowerPlay and 2D/3D voltages switching). Other words, VRM has 4 different entries allowing the GPU to select one of them via GPU GPIO pins connected to VRM VID pins. Only one entry is in use on your system.
Hi unwinder.
Yes I lurk around quite a bit. Didn't know this thread was here till someone mentioned it in another section though.
Thanks for the explanation about the power states.
BTW do you have any ideas why higher voltage isn't helping firefly forest at all but seems to help Vantage, folding, etc?
What's so special about firefly forest that would make it not respond to voltage increase?
I can't pass firefly forest at higher than 830 core, at any voltage...830 core is fine, but higher...like 835+.fails at 1.26v, 1.3, 1.35, 1.4...
Is it the shader core? Shaders aren't affected by the vcore change?
I guess I should also ask:
Do people who do the -hardware- voltage mod also have this problem with firefiy forest? This might shed some answers if only the softmod is failing in firefly....
Thanks for explaining it, Unwinder :up:Quote:
Originally Posted by Unwinder
Too bad my rig is locking up after the freeze, can't do anything but restart.
arggggg it didnt work on my visiontek 3870X2
volts are still at 3.125 both cores
must not be a referance board
Dito, exactly the same here (see my above post for details).Quote:
Originally Posted by Falkentyne
It seems the shaders are maxing out, not the core speed. And it sure seems the vgpu voltage doesn't affect shader voltage as with NV cards :/
So who here has seen improvements in OC with the 4870X2 and softmod? It seems that I haven't read about anybody experiencing better clocks.
Does anyone here know the safe zone of operation volt and temp wise for those GTX295 VRM's? I'm currently running 1.14v for the first and 1.13 for the second GPU, which leads to VRM temps ranging from 64 to 71°C. I mean Unwinder already said the controller will support much higher voltage, but what about the VRM's?
Check the other voltage, thats the supply voltage.Quote:
Originally Posted by HDCHOPPER
Critical overheating temperature of VRM phases is 150C. So i would start worrying seeing it higher than 125-130C.Quote:
Originally Posted by p2501
That's good to hear, thank you. I was a bit worried that, while staying in the safe zone for the GPU itself with below 1.15 volts, I would fry my VRM's if I kept those setting long term. :p:Quote:
Originally Posted by Unwinder
Quote:
Originally Posted by scottc19
enlighten me
thats the voltage that gpuz 0.3.1 shows me .. what other voltage ?
btw was using the batch file (just micro flashed by) and thier never was a window allowing me to pick a voltage :shrug:
that 3.125v is where it stays idle or 100% loaded seen this sucker pull 37+ amps ! clocks are 918/1134
any help please :D
I can run vantage higher with the softmod on my 4870 X2. I crash on the first test at 880 without it, but pass it without issues with the voltage soft modded to 1.45v. I fail later on the same test at 860 with stock voltage, but am able to pass it with 1.36v. I crash on the second test with 880 and 1.36v.Quote:
Originally Posted by CedricFP
--Matt
the second step just returns "41" for all the values for me??? on a sapphire 4870. Is there anything i can do with that??
Try reading the thread?Quote:
Originally Posted by jarman
Already answered. a few posts ago -.-
Do people read anything these days?
yes, i read it...but i dont quite understand, sorry.
i see u had no better clocks increasing voltage when the program returned 41 on the 4 commands?
is 65 hexadecimal for 41??? and what should be entered into the formula??
sorry if the questions seem a bit noobish....my knowledge lies more with cpu oc/watercooling than this sort of thing
Hehe, just thought the exact same thing. :up:Quote:
Originally Posted by Falkentyne
The answer, in most cases, is unfortunately NO.
Most people today are comparable to RAM. Random Access Memory, which loses all its information if shut down for a second or two. An the bad thing is they (that is their brains, if there is something worth calling a brain) shut down repeatedly after very short periods of time. :ROTF:
On a side note and again on topic:
If you don't understand hex or binary and those "things", then don't even bother to try to modify your card, unless you want to brick it. ;)
First you learn and then you use, not the other way round. It's the same thing with talking: First you think and then you talk. At least that's how it should be done.
First you think and then you talk. At least that's how it should be done.
i'll retract my previous comment and wont say any more than that.
Wow, you do have the ability to read and comprehend.Quote:
Originally Posted by jarman
Already understood how to vmod your card? :shrug:
Quote:
Originally Posted by jarman
firstly jarman do not retract anything.you are entitled to say what you want.Quote:
Originally Posted by celemine1Gig
the replies to you are one of the underlying issues in this forum,:banana::banana::banana::banana:ing knowitalls who like to pontificate to the greatness of their knowledge and thought processes,and coming down hard on those perceived to be less than them.
go to this forum where there
is a nice discussion about this process without any condescension
http://www.ocxtreme.org/forumenus/showthread.php?t=4427
thanks jtf2, much appreciated. I will register on that forum this evening :)
Well, I updated to 9.1 Cat's yesterday and decided to try this again. Seems now that 840 core is stable for me in FEAR with 1.3125v. Before I'd get tons of VPU crashes for no reason within seconds, while it worked fine in Furmark for extended periods of time.
Now that this is working like it should for me, I'm curious. How high should I safely be able to take the volts on this with just a stock cooler? Anyone have a rough value?
Also, for jarman; For me 15 is the idle voltage, 17 is the 3dload voltage, and 18 is the voltage it uses for 3dload after a VPU crash. Adjust accordingly.
if i recall correctly on that card it is 1.37 volts and then ovp kicks inQuote:
Originally Posted by Soultaker52
is there a way to bypass OCP? anyone knows?
Wow, that's only like .1v increase over stock. Sorta low compared to when the GTX 200's OVP kicks in. Oh well, I guess I'll just see what I can get on the stock cooler then, play with it and see what heat it pumps out on the vregs. I don't really see much of a reason to get some aftermarket on there if OVP kicks in at 1.37v, unless the vregs are going over 100C with normal usage.Quote:
Originally Posted by jtf2
Thanks. Now off to play.
From what I know, it'll require a hardmod. Seeing as this is a softmod, that sorta defeats the purpose. If you want to go higher, you might as well just do the hardmod for the volts as well.Quote:
Originally Posted by drifter
Guys I just upped the volts and furmark even crashes at stock! =o)~ This is even with 100 percent fan speed on the jet turbines!
+Quote:
Originally Posted by jtf2
OVP...Quote:
Originally Posted by Soultaker52
Is never a problem with this mod because OVP trigger voltage is always higher than VID. Such a software voltage controll feature wouldn't make any sense if this wasn't the case, now would it?
For what card? And what makes you think OCP is a problem?Quote:
Originally Posted by drifter
I hope the "purpose" is not what I think it is...Quote:
Originally Posted by Soultaker52
:shakes:
For the spanish speaking people:
http://foro.noticias3d.com/vbulletin...postcount=2994
Does anyone know if those Xpertvision "sonic" cards use the same controller?
gurusan,
It's quite safe to say they don't.
PALiT (aka. Xpertvision, Gainward) Sonic cards are all in-house custom boards.
Ok thanks for the info.
This softmod works for GTX 260 55nm ?
for 280 gtx. The reason i think there is a problem with ocp is because current does not go up after a certain V setting (like some users have reported).Quote:
Originally Posted by largon
so any1 got a linky to ocp hardmod?
Current not going up is not a sign of OCP. It means the VID didn't get changed.
When OCP triggers the only sign you will get is an instant reboot.
Any idea what going on here then?
I know for sure is the vcore measured from the back of the card never exceeds 1.35v volts no matter what value I use. This is a trace from 'in game' loads compared to resting loads at different voltages. It doesn't pull any more current in 3d 1.52v than 3d 1.35v. And were it moves to 1.52v in resting mode the current is well under even the 1.26v stock load current. So no evidence OCP. And OVP should always be above the set VID.
http://img218.imageshack.us/img218/1585/55424494zm5.jpg
1.06v (2d,3d low registers) to 1.55v gives ~1.35v. Upto 1.35v its very close to the DMM values in 0.0125v steps. But something is limiting it, voltage increase is linear till it hits the wall.
http://img292.imageshack.us/img292/533/87865169yv3.jpg
^^^^^ i was talking about the very same thing. so what is it if not ocp?
I don't see current limiting, just the VID won't go over 48(h), 49(1.36v) no increase. And I've polled what I think is register 13 to indicate overvoltage and that doesn't change bits either. Could do with a data sheet for the registers.
Quote:
Originally Posted by Unwinder
No use going higher than 1,36V anyway if I you can't get even 1 Mhz more out of your card than at stock voltage, now is it... :(
Mine does fine, need some more for 860+
So you're saying you can pass 3dmark06 at 1.36v at 860?Quote:
Originally Posted by fornowagain
--Matt
i can pass 3dmark06 at with 860 on the cores, but any higher and it just locks up or reboots, or i get a driver recovery (if i'm lucky) on the firefly test
No, I'm saying I can pass 3dmark06 at 1.35v at 860Quote:
Originally Posted by mattkosem
Good for you, I can't even run 840 at 1,35V through firefly forest. 832 working fine at eitehr stock 1,26V or 1,35V.. this sucks.
You have any pics? Hwbot score? You must be pretty lucky, since it doesn't seem to work that well for most other folks.Quote:
Originally Posted by fornowagain
--Matt
Lol, you don't believe me?
It does seem pretty unbelievable given the luck that others have had. Does your card pass 3dmark06 at higher than 837 without the mod? Perhaps you're just blessed with a good OCer? My GPUs never pass 31C with the mod set to 1.35v, but I sure can't run 3dmark06 at 860mhz at that spec.Quote:
Originally Posted by fornowagain
--Matt
I believe you now. I haven't tried this with the Asus bios yet. I wonder if it might have something to do with your success. Or, if it's a genuine Aus TOP card, maybe it's just a better than average card.Quote:
Originally Posted by fornowagain
--Matt
Gee thanks, how gracious of you. :clap:Quote:
Originally Posted by mattkosem
fornowagain:
I think what people are asking you is to overclock your card to 860, WITHOUT the vcore soft mod, and then see if you can pass firefly at 860.
And if you dont pass, to tell us what your highest gpu mhz you can get at the stock volts without firefly forest crashing.
That will help us at least have some idea what's going on. Do you see?
Because, I, like many others, arent getting 1 higher mhz with the vcore mod in firefly forest, and people think it's the shaders not getting higher volts (but no way to know for sure?).
Pyr0:
This goes for you too.
Please test highest clocks without the vmod.
I see pyr0 has an asus card, but what card do you have?
Hey I don't mind questions, I do resent being called a liar and printing dumb pictures. Its not for hwbot points, word should be good enough.
Don't you guys volt mod? There is one voltage control for the core and that includes the shaders. Get a multimeter, there are no more voltages that change with VID. It won't pass consistently at 862 with much less vcore (stock volts first run is 50/50), drop 25mhz is solid again. It doesn't crash in one spot, but tends to freeze in Canyon Flight or firefly. Just wish I could figure out why I can't get anymore vcore.
Oh, its a Sapphire, Asus bios. TOP bios helps the ram timings is all.
i can't pass 3dmark06 over 850 without the softmod, and see some artefacting with stock voltage at that speed
[edit] incase it matters, my card is just a standard asus 4870x2, but i've flashed it to the top bios
If this work with 9800 GX2, can someone post the command?
Thank's!
It does work yes, you'll need to run through the OP to get the numbers you need,Quote:
Originally Posted by nickless
or here > http://www.ocxtreme.org/forumenus/showthread.php?t=4427
The problem is, my card passes firefly forest at 830, with stock voltage, but wont do it any higher, regardless of voltage.
And the temps are fine too, around 62C ....
And this same problem is happening to others too....
yet the people who are having the firefly problem can fold at higher speeds than before or run vantage more stable....
Have you verified that the vcore has actually increased? Because I've proven to myself at least that you can't believe the VID value displayed by Riva. Does the load current scale, does it stop at some point? I know mine stops at 1.35v, do you know yours does to? Maybe its lower, maybe they're not all the same?Quote:
Originally Posted by Falkentyne
Clutching at straws here, some bios limit? I can't select more than 1.35v using RBE either.
Hey now, I wasn't calling you a liar. I simply requested proof. Heck, I even gave you the benefit of the doubt and suggested that you might have gotten lucky with your card. Your results are, as you can see, substantially better than the results that everyone else (Pyr0 excluded) is seeing. I don't feel that it's unreasonable to request some data given the circumstances, and would have gladly done so without harsh words if I were in your situation.Quote:
Originally Posted by fornowagain
--Matt
Quote:
Originally Posted by fornowagain
Thanks for the reply,
I looked at the load scale at idle but it didn't change.
I haven't checked when running a game though.
Should it change at idle?
In this instance I judged that entirely the same thing and said so. Had I not produced a screen shot what then would have been response eh? You said my clocks were unbelievable, I can but take that one way. Choose your words more carefully or take people at face value. Had I been in you position I would have taken your word without ridiculous screen shots. There were no points to score, I merely stated I needed more voltage for a certain clock. Do try and have some faith in your fellow man.Quote:
Originally Posted by mattkosem
Go back to the image I posted of the load current. IF the VID has increased then the core MUST consume more current at a given frequency. Load a game that doesn't mind dropping to/from desktop a few times or folding etc. Watch the load current for the stock VID, increase the VID a few steps. See if the current increases, mine does to start with then stops at a point.Quote:
Originally Posted by Falkentyne
Hi,
Just tested it with gpu-z. it changed with each setting correctly.
I just ran "AMD GPU tool" to overclock, and when I changed the voltages, the current DID NOT change.
(neither did the vcore, but I do remember it changing before >_>).
WTH is AMD GPU clock doing? locking out the registers?...or are the sensors just failing to respond?
Bah whatever the AMD GPUtool is doing, doesn't matter; I used ATT to overclock and current/vcore changed correctly also, but i still can't get higher than 830 mhz on firefly forest. I tried 850, 1.35 vcore and 65% fan and it locked in 10 seconds.
840 mhz died after a few passes. 835 mhz did 5 passes before crashing. Vcore didn't matter...
I guess it's back to the drawing board then, as to why more vcore isnt helping firefly forest....
The only thing I have NOT done is lowered the RAM speed. (I always set it to 1125, which is artifact free).
Well, therein lies the problem. You read my post with a negative connotation. You understood the word unbelievable as "too improbable for belief" instead of as "of such a superlative degree as to be hard to believe" as I intended to communicate.Quote:
Originally Posted by fornowagain
I'll work on more careful wording, but you should probably work on meeting curiosity with less hostility. Anyways, this has gone OT enough. Thanks for providing the image, nice work.
--Matt
The VDDC load current stepped up inline with the increased VID? Does it stop increasing or stop. Rivatuner is probably better as you can expand the scale to see more.Quote:
Originally Posted by Falkentyne
I kept editing my post several times through reboots and testing (different computer).
Yes the monitoring worked fine, as long as I didnt run AMD GPU tool. as soon I ran it, (without even changing settings) the sensors 'vanished' or something or if already active, froze. Using ATT to overclock was fine.
The current went up with the vid (at idle) and apparently alot more at load.
anyhow, same result..can't go any higher in firefly forest, regardless of the vcore/current. My max is 830 mhz regardless of voltage. (835 mhz works for about 3-4 runs then lockup, etc).
Thank's man and @3oh6 too. I've tried this and now my card works at 846MHz GPU, 1.44v from the first try :up:Quote:
Originally Posted by fornowagain
Now I'll burn it :hehe:
AMD GPU Tool does weird things.. it shuts down CCC and breaks fan speed control on 4870X2 cards..Quote:
Originally Posted by Falkentyne
Quote:
Originally Posted by fornowagain
OK, now, besides the fact that I don't really understand why you would really want to run your card at 1.55V, here's what I'm thinking.
You are simply hitting the power circuits physical limits. ;) If the monitored data is correct, what you see is that the current actually decreases while the voltage is set to 1.55V, in comparison to 1.35V. And at the time the voltage falls back to 1.35V the current increases again.
It would seem as if the circuit just can't deliever any more current than what is available at 1.35V. That's why the current decreases when the voltage rises. You hit a point where the power output is limited to a certain amount. It's like Intel's VDroop function. Once the current draw gets too high, you either have to drop the voltage to deliever the current, or drop the current to get the voltage. Both won't happen, as it's a physical limit. ;)
Perhaps this could be modified in some way, but I doubt that it would be very healthy for the card.
Oh and BTW, what is the actual DMM reading of the voltage, when VID is set to 1.55V? I didn't fully understand that just from reading your post. It's a bit confusing. But maybe that's just because English is not my native language. In one passage it reads like it was 1.35V and then you talk about 1.52V. What is the correct value? :shrug:
i have 4870 1gb giagbyte, tried this on cmd but nothing, everthing is not recognized, maybe this is because 64bit vista sp1??
i run cmd as adminitrator.
yepp i did it(i hope so) in riva tuner i got new voltages but havent tested yet.
here is pic, how can i fix this that gpu dont change voltage every second(it went from 2d voltages to 3d voltages every second) i have chanhged both 2d and 3d voltage card has also before doing this chnged this voltges every secon(before was 1.2-1.26v)
http://img518.imageshack.us/img518/9...odedgpulo5.jpg
how can i get it in 2d that it is always 1.1v??
I don't. I want to know why 1.35v is the maximum. A hardmod would take voltages higher, this method should as well.Quote:
Originally Posted by celemine1Gig
Its not hitting any limits, OCP causes instant shutdowns. There are two load states. One is as the game is running pulling the most load and the other as it idles at the desktop. If there was any current limit actuating at the higher VID-1.52v, whilst at idle there is still more current headroom. When switched to heavy 'in game' loads at the higher VID-1.52v then it could reach higher currents and crash, it doesn't. In fact the heavy load current is the same as 1.35v. Indication that 1.35v is the maximum voltage.Quote:
Originally Posted by celemine1Gig
Its in the picture as well, the actual voltage never exceeds 1.35v regardless of the VID set.Quote:
Originally Posted by celemine1Gig
Another, you see a little variance in load depending on what the game is doing. But loads currents 1.55v=1.35v. Finally default load current @1.26v is a lot lower.
http://img129.imageshack.us/img129/921/43730291lt6.jpg
Gigabyte HD4850 here , no Voltage reading nothing at all :D:
Running GTX 260 I can change voltage using this, but for some reason in RT Monitor my voltage regulator output never changes even when going to full load. Current increases but not output.???
Quote:
Originally Posted by fornowagain
I think you didn't quite understand what I meant to say.
There are circuits that just keep on supplying higher currents to drive higher voltages, until they reach the OCP and/or OVP trip point and shut down to prevent damage.
And there are circuits that have another way to deal with this problem.
They will only supply a certain amount of current. To get a higher voltage you will need the right amount of current to drive the voltage (higher voltage means higher current). In your case it seems that at about 1.35V you have reached the circuits' current limit. Now you can set any higher voltage via VID. There'll never be enough current to drive the higher voltage, i.e. it'll always remain at ~1.35V. Of course the current draw at 1.26V is lower (see above).
BUT, I'm not quite sure if this is the case with your card, because normally in idle state the higher voltage should show. It would just drop down in load state. Maybe it's because the card switches to a lower VID in idle anyway, or maybe on your card there is a different implementation that doesn't even allow VIDs higher than 1.35V. I have no datasheets here so I can't tell you for sure. You need to ask the poeple that have the datasheets. But unfortunately these datasheets seem to be under NDA anyway.
You could test and set all register value for all states to 1.55V (0x58h = 88). Then in idle you should theoretically get the higher voltage.
Besides if and how OCP gets implemented is a question that the developers need to answer. It's not always the same and you can't say that there will always be a OCP limit that will cause the controller to shut down.
4870 512, works here @ 1.3v going to try 1.35
nice got to 860 on the core :)
going for 870 now :)
I understand the phased regulation fairly well, I have volt modded many cards and seen OCP in action. Its not changing VIDs and the voltage is the target value which holds solid at 1.346v. Its a regulated value and quite obvious when you observe it in real time via DMM. Voltage increases in discrete 0.0125v steps until 1.346v.Quote:
Originally Posted by celemine1Gig
Again, the reason I included idle load current is to show that OCP is not occurring. Also the OCP/OVP register 13 is not changing state. And yes indeed you can activate the 2d VID to target values via register and that is also limited to 1.346v regardless of current (see previous posts). Unwinder (RT designer) has already stated the much higher voltages are supported with the 1165. OVP shouldn't factor here as its nominal VID + 200mV offset.
There is a definite lock here from 0x48h to 0x49h.
Set all 4 register entries to 58 and see, as I said.
First thing I tried (back in the thread)Quote:
Originally Posted by celemine1Gig
/wi1,70,15,58 /wi1,70,16,58 /wi1,70,17,58 /wi1,70,18,58
=1.346v
So that's what I wanted to know. If idle voltage is set to 1.55V, you still get 1.35V, right? Your pics didn't show this config, so...
As I said without the exact datasheets and the card itself on hand, this is a hard task.
I hope someone with the right data on hand will be able to assist you further.
Yep.Quote:
Originally Posted by celemine1Gig
Quote:
Originally Posted by fornowagain
Why is the title even saying "ANY" when it only works on a very limited number of cards?
I have tried this soft mod on both of my cards 4870x2 one its a Sapphire the other one its a Powercolor running on high end H2O EK Blocks both , not one single MHz gain as Jcool stated , it ran the same as before the soft mod , max clocks on the best card (powercolor) on stock voltages its 865/1000Mhz stable on 3D06 chilled water. The Sapphire does 855Max , tried both and again not a single mhz gain , yes i did it correctly ,i have been volt moding and using Riva for a very long time. Not worth the hassle if there is no gain :p:, i dont understand how people are seeing improvements when in my case i did not see any , like many guys already found out it does not help oc further 1Mhz on 4870x2cards , that conclusion its base on my extensive testing already done, so dont know whats up with this soft mod :shrug:.
Nothing wrong with the mod, it's just that RV770 has never been much of a voltage scaler. ~850MHz seems to be the max irregardless of voltage - until sub-zero cooling is involved. But even when supercooled, RV770 doesn't really go crazy on clocks.
That may be right, but the fact that the voltage seems to be somehow limited to 1.35V on the X2s is still strange. It's either a hardware constellation, like a voltage limit through diodes, but then again the current would increase with the higher voltage if it was. Or the registers are used a bit differently on these X2s. :shrug:Quote:
Originally Posted by largon
I asked this earlier but no one answered.
Were people doing the previous soldering resistor hardware mod also seeing issues with firefly forest in 3dmark not being happy with higher GPU speed (past what you could do at stock voltage) regardless of the hardware changed volts?
I may not aid with 4870x2s, but with straight out 4870s, it helps massively, bumping up from 1.26 to 1.4 (via software, not sure on exact, don't feel like crawling under my desk right now, just had knee surgery) took me from a max stable of 800 to 825, so it helps, just not a massive amount.Quote:
Originally Posted by chispy
I have too Gtx260 216 and in rivatuner show the vtt regulator on 1.06v , never changes with the mod :s :shrug:Quote:
Originally Posted by Hardass
any help?
darn and i was hoping for some improvements with my hd4870x2 heh .. so anyone actually report gains with hd4870x2 with this method ??Quote:
Originally Posted by chispy
I passed the link for this thread to some friends and one is having great results with 4870X2 after the mod, 880/960 on 3dmark06 and Vantage, card is WC.
I'll see if I can get him to post here.
My bad, I had to add a note and explicitly say that provided VRM status register address was in decimal format. You were checking wrong register, correct command for checking VRM status is /ri<bus>,70,dQuote:
Originally Posted by fornowagain
The card I have is hard modded 4870x2 and its stable at 900, Ive ran through vantage at 940 on water and I think I could make it to 950+. I guess I have a gem :P Though if it could have been done softmodded it would have saved a lot of trouble....
Keep up the good work with the softmod :)