The X1900XT/XTX Thread

[deathman20]

Well, I said Voltages, but I wes really interested in the Amp readings...

Although, I could do the math...

I'm really curious about this PSU, with its dual 18A 12v rails... I'm concerned it isn't going to hold out with the CPU OC, all the fans and now the GPU pulling 30A...
I really need a single rail 30A+ deal, right?

Its not pulling 30A from anything. Thats a rough reading of 30A being pulled from the GPU or Regulators at a certain voltage.

It only needs 10-11Amps on a 12V line to run really un-OCed. Out of the box its around 120W for the whole card. Add in some voltage and clock increases they of course it will go up but theres no way you'll need 30A on a line. Many run it off of a single 18A line, so surely you can.

Have you tried to un-oc your CPU and unclock the other card to see if thats the issue?

 

[digerati]

Its not pulling 30A from anything. Thats a rough reading of 30A being pulled from the GPU or Regulators at a certain voltage.

It only needs 10-11Amps on a 12V line to run really un-OCed. Out of the box its around 120W for the whole card. Add in some voltage and clock increases they of course it will go up but theres no way you'll need 30A on a line. Many run it off of a single 18A line, so surely you can.

Have you tried to un-oc your CPU and unclock the other card to see if thats the issue?

Yeah, now that I really think about it, dual or not - its still a true 36A PSU.
I should be fine, there's really no issues I've run into.. (that weren't just newb related )
I'm gettin the hang of it though...

Now I've got to figure out why my E6300 on a Scythe Ninja is still hitting 65C+ on Orthos... :

 

[deathman20]

Yeah, now that I really think about it, dual or not - its still a true 36A PSU.
I should be fine, there's really no issues I've run into.. (that weren't just newb related )
I'm gettin the hang of it though...

Now I've got to figure out why my E6300 on a Scythe Ninja is still hitting 65C+ on Orthos... :

I mean on just my OCZ 520W people have run crossfire on this setup with high end P4's and X2 systems as well.

 

[digerati]

I mean on just my OCZ 520W people have run crossfire on this setup with high end P4's and X2 systems as well.

Gotcha...
So, is there any good software to monitor GPU / CPU Amp usage?

 

[deathman20]

Gotcha...
So, is there any good software to monitor GPU / CPU Amp usage?

There is none. The one in ATI tool was a first if not mistaken but it doesn't read properly and even W1zzard said so.

 

[Neuromancer]

Stock sapphire x1950xt 256MB no Voltage adjustments fan barely kicks on and temps hit 70C

Max OC so far 688/945. Once I get it pulled out of my HTPC and setup ati tool to run the fan on high I will try and go for 700/1000

EDIT: Just tried upping the voltage and it appears that Catalyst is limiting the OC to 688/945 as it lists maximum OCs of 690/950 cause the numbers didnt change
Will have to look into Ocing in Vista I guess

 

[DemonEyez_NL]

Thanx for all the replies. I'm not convinced it's a power problem. How could can one explain the following:

When I run with my Sapphire X1900XT slave card @ 1.5v i'm not having any problems.
But when I run with the Connect3D X1900XT master card @ 1.41v it's giving me problems.
Running the Sapphire X1900XT @ 1.5v proves that the system has enough juice, heck, I can even do 1.55v

Ati Tool reports the current drawn from the voltages regulators, not the the draw from the PSU.
The Thermaltake 250W PSU has two PCIe 6-pin connectors, each on a separate 25A/12v rail.
It's Crossfire certified on the ATI site in conjunction with a 450W PSU:

Thermaltake Purepower Power Express 250W* (W0130RU)
*In conjunction with a standard 450W or above Power Supply

http://ati.amd.com/technology/crossfire/buildyourown2.html

It's a really annoying problem. I even had to change the vgpu in the card's bios itself, because default it's set to 1.425v for 3D Overdrive mode


I was wondering, what's the average OC with 1.2ns Samsung chips. ?

 

[Mr. Cornell]

I've actually had this X1900XTX for more than a year now. But I've never really tried hard to push it, because the cooling was for some reason worse in my old box even though I'm using the same case and fans. It makes no sense, but hey, whatever. Right now if I force the fan to spin at 77% I am able to hold the temperatures at 77C continuously under load with the stock cooler. Damn, does that thing whine. This is at Vcore 1.475V, VDD and VDDI at 2.089V.

Anyways I've been using ATITool 0.27 Beta 1 Artifact Tester with the slider all the way to the left (max detection, least OCing) and I've passed 5 minutes of the tester at 708/864. So far so good. Going any higher on the memory results in visible dots appearing and disappearing. Going any higher on the core, interestingly enough, results in dropping to a blank screen after a few minutes (no video signal) for a few seconds, then VPU Recover takes over and resets my GPU and I have to reboot my computer immediately or it will hard lock.

 

[ECH]

Ever since I got my brandnew Connect3D X1900CF (1.2ns) out of the box, i’ve been having a strange problem.
I get a flickering /black screen now and then

I only get the flickering/black screen when vgpu is set to 1.41v or higher!
With black screen, i mean loss of signal; my monitor goes black for a couple of secs and then is viewable agian, only to be followed by another black screen again.., and soforth and soforth...

But.... When I set vgpu to 1.40 or lower, the flickering/black screen flashing stops immediately!
Setting vgpu to 1.45v-1.50v results in an direct black screen, with a sporadic viewable desktop in between.
I've been testing this with AtiTool.

The problem occurs in 3d mode, but also in 2d mode (if vgpu is set to 1.41v or higher.)
Running crossfire enabled or disabled doesn't matter
Running with only one card in my systeem doesn't help either

At first, I thought it was VPU Recover kicking in, but I get it with vpu recover on and of.
It's almost like the GPU core of the master card itself cannot handle a vgpu of 1.41v or more.

My temps are good, the card is cooled with a full-cover EK waterblock , giving me 42 degrees Celcius stressed.

I'm still hoping to find a answer to this problem, because I'm limited to 1.4v on the GPU and I need more juice to take it over the 720MHz when overclocking.
My slave card (Sapphire X1900 XT 1.1ns) does 750MHz with 1.485v, so it would be nice to get those two cards in sync while overclocking.

I had that problem with my old x1900xtx that I received from ATI's RMA department. I have isolated the problem down to frequent voltage fluctuation when it should be rock solid (using Rivatuner). It appears that a mosfet, vreg or some other component is not working properly or, there is bad trace somewhere.

 

[DemonEyez_NL]

I had that problem with my old x1900xtx that I received from ATI's RMA department. I have isolated the problem down to frequent voltage fluctuation when it should be rock solid (using Rivatuner). It appears that a mosfet, vreg or some other component is not working properly or, there is bad trace somewhere.

Thanx! I've been searching for two weeks on Google to find someone who has experienced the same problem

I'll check with Rivatuner if the voltage is fluctuating when VGPU is set to 1.425 compared to 1.40v

So what should I do? I can RMA the card right? It's six weeks old.

 

[ECH]

^^ Like I said, the one I got back from RMA had this problem. RMA would be the ticket though as you really have no other choice in the matter. Make sure you are very detailed in your explanation of why you are sending it back. Also explain if you can/cannot go into 3rd mode (assuming you cannot as I couldn't) whenever you enabled a game. If so, explain to them what the 2D clocks are and don't hint to anything about over clocking beyond CCC. I've owned many ATi video cards in the past and, I have to admit the x1900xtx was the most bothersome out of the whole bunch for me. In total I replaced 2 x1900xtx which is a freaking record for me. I never had to RMA or exchange any ATI card in the past. Trust me, you are not the only one experience abnormal problems with this video card. Specially when it comes to driver updates, etc. I've gotten to the point where I don't recommended them at all to be honest, Although I am judging on my experience and a few posts here and there it's a bit odd to experience these many problems (at least for me). I was hoping that the HD series would fair better. However, from the looks of things there will be no ATI version only vendor versions. Also from what I read the R600 will require whopping 250 watts vs. the 8800 which only requires a 177 watts! Freaking OUCH! That will require a new PSU if you want to OC as the current standards for PSUs with PCIe is cable something like 75 watts and PCIe slot is something like 75 watts totalling 150 watts. If you add another PCIe cable that's 225 watts right there (someone help me if I am wrong here). If this is true I am sure reaching the limits of your PSU is not an option when using this card at standard clocks. ARGH!

Do me a favor, post a screen shot using RivaTuner's monitor. I want to see your voltage regulation. You can upload the photo to a place called photobuck. Wait 5 minutes before you take the screenshot (in other words let the monitor run for a few minutes).

 

[DemonEyez_NL]

I have to say that I've been using ATi card for five years now (R9700, R9800, X1800, X1900XT and X1900CF) and never had any problems whatsoever. (I'm not taking in acccount the software problems with Cat 7.1 7.2 because those where utter crap)

I going to RMA the card with the shop I bought it from. I spoke with Connect3D and they said I'll have to return the card to the shop from which th card was purchased, then they will RMA the card with Connect3D.
I'll tell them that the card does not function properly in 3D mode on stock voltage and speeds

I made a screenshot of Rivatuner. I highlighted two sections of the graph,
the first part is with VGPU set to 1.40v in Ati Tool, which Rivatuner reads as 1.42v The current draw from the vreg is 27.3A , that's with the 3D image open with Ati Tool. This runs fine.

The second part of the graph is with VGPU set to 1.42v in AtiTool, which Rivtuner reads as 1.45v. Again the current draw is at 27.3A and seems to be stable, but now I'm getting the signall loss.

So in either way, the voltage/current draw doesn't seem to fluctuate.
But raising VGP to 1.425v gives me signal loss

At the time of making this graph the card ran at stock 3D speeds of 620/720

 

[ECH]

the voltage fluctuation I am talking about is to the right of your arrow. I have seen it go from 2.7a to 5.5a when the computer temps are higher then normal but that's at idle no load temps. However, when the temps dip when under load is where I started noticing problems. It would dip down to a lower voltage then go back to 27a then back down, then back up...creating a fluctuation. After a few minutes I would notice problems. As long as my voltage read stable I would get no problems. As soon as they fluctuated under load (going up and down, up and down, up and down) is when problems started. Granted I am only talking of my own experiences and the research I did to try to isolate the problem (which I believe to be some how related to the ring bus).

This is what I mean when I say stable. The only reason you see any dips is because I moved RTHDRBL which pauses it.

As you can see, there are some slight increases but never any dips (which caused problems with my other x1900xtx). Try RTHDRBL and see if the voltage is stable or not.

 

[DemonEyez_NL]

The large dips to the right side of second arrow are caused by Ati Tool.
That was the moment I manually was switching profiles from 3D to 2D and closing the 3D image. That's the reason I darkened that part of the graph.

I'll upload a new grapf this afternoon, now with RTHBRBL running, as in your picture.
Did a quick run just now, and I saw that the current was fluctuating between 22A en 25A. This happens with 1.4v as well as 1.425v

 

[ECH]

The large dips to the right side of second arrow are caused by Ati Tool.
That was the moment I manually was switching profiles from 3D to 2D and closing the 3D image. That's the reason I darkened that part of the graph.

I'll upload a new grapf this afternoon, now with RTHBRBL running, as in your picture.
Did a quick run just now, and I saw that the current was fluctuating between 22A en 25A. This happens with 1.4v as well as 1.425v

I wouldn't worry about what voltage you have it at, if you had the same problem as me it won't matter. You will see it at 2d as well as 3d clocks. It may take a bit longer for it to happen in 2d mode.

 

[DemonEyez_NL]

I wouldn't worry about what voltage you have it at, if you had the same problem as me it won't matter. You will see it at 2d as well as 3d clocks. It may take a bit longer for it to happen in 2d mode.

Okay, I made another screenshot:

This is with RTHDR running. When running in 2D and doing nothing all is well.
But when I start RTHDR and 3Dmode is switched you can see how the current draw from the vreg's is fluctuating between 18A and 23A.
This happens with 1.40v and 1.425v

With 1.40v I can run stable all day without any visible problems (even though the vreg's are fluctuating), but 1.425v and higher will cause signal loss after 5 or 7 seconds .

 

[ViperJohn]

Okay, I made another screenshot:

This is with RTHDR running. When running in 2D and doing nothing all is well.
But when I start RTHDR and 3Dmode is switched you can see how the current draw from the vreg's is fluctuating between 18A and 23A.
This happens with 1.40v and 1.425v

With 1.40v I can run stable all day without any visible problems (even though the vreg's are fluctuating), but 1.425v and higher will cause signal loss after 5 or 7 seconds .

I can't stand it anymore lol.

It is NORMAL for the current (BTW that is the current through only one phase of the Vcore PS) to fluctuate
in 3D application. The load on the GPU is not static but is constantly changing as a moving 3D scene is
processed and rendered to the display. That changing GPU load is reflected in the GPU current draw at any
given moment in time.

Viper

 

[DemonEyez_NL]

Aha, okay thanks for clearing that up.

So what's your opinion of my card's behaviour when I set the vgpu to 1.425v and higher which causes signal loss?

 

[ViperJohn]

Aha, okay thanks for clearing that up.

So what's your opinion of my card's behaviour when I set the vgpu to 1.425v and higher which causes signal loss?

Since you can run several seconds before the screen goes blank the Vcore PS is most likely shutting down
or the Vcore is being seriously clamped downward by OCP (Over Current Protection).

When you increase the Vcore the current drawn by the GPU off the Vcore PS will also increase at the same
ratio that you have increased the Vcore by. While the point where OCP kicks in is normally higher, tolerances
in the Vcore regulator and the components supporting the OCP function produce max current draw range
where that will occur on every given card. Luck of the draw probably landed you a card where those tolerances
stacked up and pushed the trigger point of a Vcore PS shutdown or Vcore voltage clamp on OCP very low in
that range.

It is easy to check for. Use a good DVM and meter the Vcore on card while running. When the screen goes
blank take a look at the Vcore indicated on the DVM. You will probably find it very low or even at zero volts.

Viper

 

[DemonEyez_NL]

Since you can run several seconds before the screen goes blank the Vcore PS is most likely shutting down
or the Vcore is being seriously clamped downward by OCP (Over Current Protection).

When you increase the Vcore the current drawn by the GPU off the Vcore PS will also increase at the same
ratio that you have increased the Vcore by. While the point where OCP kicks in is normally higher, tolerances
in the Vcore regulator and the components supporting the OCP function produce max current draw range
where that will occur on every given card. Luck of the draw probably landed you a card where those tolerances
stacked up and pushed the trigger point of a Vcore PS shutdown or Vcore voltage clamp on OCP very low in
that range.

It is easy to check for. Use a good DVM and meter the Vcore on card while running. When the screen goes
blank take a look at the Vcore indicated on the DVM. You will probably find it very low or even at zero volts.

Viper

Oke thanx for all that info.
I'll measure the vcore with my DVM this afternoon, before I'm packaging the card to send to the RMA department. I've been having trouble with this card since it came out of the box. I had to edit the bios of the card to 1.40v instead of the default 1.425v to get it working.