EVGA 7800GTX throttles back when replacing standard cooler with waterblock

I'm wondering if there's a way to just create a short between two of the pins that sense the speed and I would think if no fan means the card clocks way down, then the card seeing 100% fan speed would mean highest clock speed. Altough, since a fan does create electrical resistence, I'm thinking a resistor of some sort would need to be put in place just for same measure.

 
hmmm... flash the bios to another mfg without that fan monitor.
 
I wonder if eVGA uses a reference design in which case they would most likely all have that monitor. don't worry, i'm faithful that someone will find a way to bypass it.

 
Yes, this happens with every card, or at least it sorta should.

The card is design to throttle itself back when it does not detect a fan and if its not running at the correct speed. You need to modify the BIOS in order to fix it.
 
LunchboX3904 said:
I'm wondering if there's a way to just create a short between two of the pins that sense the speed and I would think if no fan means the card clocks way down, then the card seeing 100% fan speed would mean highest clock speed. Altough, since a fan does create electrical resistence, I'm thinking a resistor of some sort would need to be put in place just for same measure.

Click to expand...
The first link has a diagram for a "pulse generator," which is probably the most reliable way short of a new BIOS. Just shorting out the pins for "100%" wouldn't work, because it doesn't measure how much resistance is- or isn't- going through that wire. I believe they actually use a PWM or something similar, and actually can count the number of pulses per second/minute.

I'd just sit back and wait for nVidia's new BIOS to leak or be released, just because anything else- being a homemade solution- subjects your $600+ hardware to insane amounts of risk. I know I wouldn't trust my own solder-work under those circumstances.
 
UPDATE FROM E-VGA

It might actually be a fan problem....

Here is the direct response from EVGA....

"The bios/driver monitor the FAN speed as part of its thermal protection system. If the fan is disconnected it reads it as a fan failure and throttles back the GPU in order to protect it from damage (which in a NON-water cooled situation would be very good).

xxxxx is working to get us a new BIOS that does not pay attention to the fan but instead moves the thermal throttle monitoring over only to the temperature sensor on the GPU.

We should expect a new BIOS from NVIDIA either late today or sometime tomorrow"
This is good news....When I receive the bios I will report back...

-M
 
Hey managerman, I'm really glad you started this discussion over here. The more people are made aware of this and the more input we receive, the better. I'm just going to copy and paste my latest findings from the other thread:
Well, I just finished re-installing the waterblocks on my videocards while keeping the stock fans plugged into the headers to see if indeed they are sensing for rotation. Even though the fans are plugged in and spinning, the video cards still throttle back while running applications such as any of the 3D Mark programs. The current temperature readings in the NV control panel are 37-38 degrees for gpu #1 and 36-37 degrees for gpu #2. Obviously, the video cards are not sensing for fan rotation and instead are throttling back based upon the temperature readings. There is indeed a "cold bug" which kicks in at relatively modest temperatures.
Looks like we'll have to wait for a new BIOS that disables throttling or kicks in at far lower temperatures.
 
New Update!! Direct Quote from EVGA..

"This problem will require an update to the display driver – NVIDIA is in the process of writing that and validating it right now – they hope that they can have something this week (key word is hopefully)

The fix needs to be made in the driver instead of the BIOS because if they edit the BIOS for the fix, then the card will have no way to monitor the temperature anymore and that is very dangerous to do.

If anyone has direct information about this (setups and temperatures being hit, etc.) please be accurate and very detailed and post them here in this thread, as this would be very helpful to NVIDIA – they are aware of this thread."

Good news....

-M
 
I will have info in a day or so about the bios edit. My card is with VJ right now.
 
ummm sounds kinda stupid but what if you just leave the fan hooked up and stick it off to the side for now?
 
Beazalbob said:
ummm sounds kinda stupid but what if you just leave the fan hooked up and stick it off to the side for now?
Click to expand...
We've tried this and it doesn't have any effect. Although initially we thought that the videocard was sensing for fan rpms, this is not the case. Throttling kicks in at a specific temperature threshold still unknown at this time, but believed to be between 37-46 degrees celsius.
 
Well that makes no sense, since the the stock evga card I have runs at 50'C out of the box.

The throttle back temp must be higher than that.
 
They're saying it has not only a high temp throttle back point, which is normal, but also a low temp throttle back point which is kicking in below a certain temperature and slowing the cards down.
 
Well all of this is wierd. I have never heard of a too cold scaling.

My evga card is doing unexpected scaling also,

I have it overclocked to 500/1300mhz for 3d, and the default 275/whatever for 2d. When the card jumps from 270 to what I expected to 500 when starting up a 3d game, Both Riva Tuner, and the Nvidia tool report auctual mhz to be 540-549mhz. 3dmark also see's this, so what gives? Is it really scaling to 540+ mhz?
 
Gadfly, have a look at the thread over on the nvnews forums titled "EVGA 7800GTX throttles back when replacing standard cooler with waterblock". It explains the cold bug problem pretty well. Managerman has a link to it in the post above.
 
Over at extremesystems, a BIOS is available that disables thermal monitoring altogether for those brave enough to try it :eek:
 
More experimental BIOS are now available at extremesystems.org. Nay word on an official fix from nvidia?
 
managerman said:
Have not received the new drivers from Nvidia (evga) yet...

-M
Click to expand...
There is a beta 80.40 release available on the net now. I wonder if these fix the "Cold bug".
 
There is an article over on the inquirer where a Gainward pair of watercooled 7800s are tested and overclocked upto 500mhz core, and 1400mhz for the memory. THe tester states that the cards can go higher, but that they are awaiting a new BIOS from nVIDIA that disables throttling. Makes me wonder whether we can expect a new BIOS or a new driver that eliminates the throttling issue with these cards. Perhaps it will be a combination of both that solves this problem.
 
I want to watercool my 2 BFG 7800 GTX's. Do they throttle also with a waterblock on them???
 
Homer, from what I know, the BFG cards are also affected since they follow the same basic nVIDIA reference design. At the beginning of this thread, managerman provides a link to a thread over on the nvnews forum where someone confirms that the BFG cards throttle back when watercooling. You may wish to read over that thread.
 
Homer said:
Damn.... I hope theres some kind of fix eventually for it :(
Click to expand...
BIOS fixes work, a new driver will work even better....

Or you could keep the damn fan plugged in (or plug in a different one if you don't like that one!).
 
Ether said:
BIOS fixes work, a new driver will work even better....

Or you could keep the damn fan plugged in (or plug in a different one if you don't like that one!).
Click to expand...

It is not the fan. I have been working with 2 water-cooled 7800s. The problem is tempature related. System runs at 36C....crappy frame rate. Temp above 40C....100%. I have tested this with the easy. Cover the rad intake fan trick. I forced the cooling loop to run hotter and the system ran fine. There are modded bioses out there to fix the problem. They disable temp monitoring and throttling. (BTW I hooked the fan up to see if this fixed the problem....it didn't. Just to be sure.)
 
Will a patch or update ever fix this from one of the companies (nvidia, bfg, xfx, etc?)
 
I will contact eVGA tomorrow and see if there is any driver update....Has anyone tried the 80.40 drivers to see if the cold bug is still present...??

-M
 
Wait wait wait, your going to call evga, and say "Hi, I'm installing a waterblock on my 7800 and it throttles when it gets too cold, will you still even accept this call because I'm not using stock cooling?" :p
 
No..Since I work for a major computer retailer..I have contacts within eVGA through our buyers..

-M
 
managerman said:
No..Since I work for a major computer retailer..I have contacts within eVGA through our buyers..

-M
Click to expand...

I own a custom computer system company. I have personally talked with XFX about the same issue. (BTW XFX really impressed me with this card and support.) I got the same answer. (Fix coming this week. <----NVIDIA) From what was said it will be a driver fix. For now we have used a hacked bios. When the driver fix comes out we will switch back to the factory bios and retest.
 
AACDIRECT said:
I own a custom computer system company. I have personally talked with XFX about the same issue. (BTW XFX really impressed me with this card and support.) I got the same answer. (Fix coming this week. <----NVIDIA) From what was said it will be a driver fix. For now we have used a hacked bios. When the driver fix comes out we will switch back to the factory bios and retest.
Click to expand...


w000t!!! ty for the nfo!
 
hope it gets here soon. i wanna see what this thing will do with some real cooling!
 
The same problem occurs with ASUS 6800 Ultras. Mine have water blocks and they did the same thing. I cut a section out the side of the fan shroud and mounted it around the water block. It works great now. Do not let them you it is a GPU issue, it is a RAM sink issue.
 
Just a quick update. I talked with NVIDIA yesterday afternoon. The fix is most likely NOT coming this week. They are still doing QA ( quality assurance) on the driver. I hope we see it next week.
 
Gadfly said:
My evga card is doing unexpected scaling also,

I have it overclocked to 500/1300mhz for 3d, and the default 275/whatever for 2d. When the card jumps from 270 to what I expected to 500 when starting up a 3d game, Both Riva Tuner, and the Nvidia tool report auctual mhz to be 540-549mhz. 3dmark also see's this, so what gives? Is it really scaling to 540+ mhz?
Click to expand...
NVidia has addressed this issue also. As part of the power/heat/performance management features of the G70 GPU, some parts of the chip run at different clocks than others. Specifically, about the 40Mhz difference you are seeing.

One OEM is actually doing a little dirty marketing on this. Even though the clock used as the overall speed reference runs at 430Mhz stock, that leaves one section of the chip that runs at 470Mhz, so they are advertising the card as running at 470Mhz, like they've OC'd it from the factory when they haven't. Cute, eh?
 
You must log in or register to reply here.
Back
Top