Anarion

clone
I must be one of the lucky ones as my 1070FTW runs very cool but ive still asked for the thermal pads.

GPU might run very cool but the VRM are could be well past the boiling point.

This is rubbish, you have no documentation to prove your statement. Don't post something you don't have a clue about.

I hope evga is including one of these with every pad kit:

emsir

Anarion

clone
I must be one of the lucky ones as my 1070FTW runs very cool but ive still asked for the thermal pads.

GPU might run very cool but the VRM are could be well past the boiling point.

This is rubbish, you have no documentation to prove your statement. Don't post something you don't have a clue about.

What do you find to be inaccurate about their statement? The temperature sensor on graphics cards is for the GPU chip, and not the VRM or VRAM. Whatever temperature is read from the GPU chip does not reflect the thermal conditions of the VRAM and VRM.
 
I hope that a cool GPU means also cool VRM and VRAM, but without thermal pads to transfer the heat from them, there seems to be significant risk to those components.

emsir

Anarion

clone
I must be one of the lucky ones as my 1070FTW runs very cool but ive still asked for the thermal pads.

GPU might run very cool but the VRM are could be well past the boiling point.

This is rubbish, you have no documentation to prove your statement. Don't post something you don't have a clue about.

I don't think anyone could 'prove' anything without FLIR equipment measuring their particular example, but you can arrive at some informed hypothesis looking at approx. temp offsets of others measurements: http://imgur.com/a/D268D

shannonjpower
I replaced my vram pads today and the stock ones on the 1070ftw appear to be 1mm. I replaced them with 1.5mm and the new ones are ever so slightly thicker.

Thanks Shannon. That's very helpful. I suppose trying the 1.5mm first would be better (more pressure), but if anything bends, warps, whatever I'll then use the 1mm. I expect that 1.5 will be fine though, and probably stay with that. Doubt there will be warping lol. Cheers.

 

gahelm

GFAFS
I told you day one, EVGA, playing with your customers will get you cornered and in full panic mod...didn't i?
 
Now to add some food for the thoughts i'll let statistical numbers do the talking (one may investigate if necessary):
 
Onsemi Reliability Data: NCP81382HMNTXG/NCP81382MNTXG
Without any mods 'being optional' ( aka Stock cards/no additional OC): VRMs Temps measured at 107°c to 114°c max
0.7eV - 107°c - 90% confidence - MTBF/MTTF: 7321258hrs - 78.98 FITS
0.7eV - 114°c - 90% confidence - MTBF/MTTF: 8603197hrs - 116.24 FITS
 
An imaginary card with properly contained VRM heat at 78°c (close to the market average):
0.7eV - 78°c - 90% confidence - MTBF/MTTF: 64869441hrs - 13.52 FITS 
 
The same can be done with the VRAM (temps are usually between -40°c ~95°c max for industrial grade, far less for commercial grade 0°c ~85°c max)

Why do your "Mean Time Between Failure" numbers go up with the higher temp?  Something is wrong with your stats dude....  Higher temps should mean more frequent failures thus your MTBF number should go down.  Higher MTBF = better, Lower MTBF = worse.  Not sure where this data is coming from or its relevance, just pointing out a problem.  Am I looking at your data wrong? 

I wondered the same as you, and I just looked up the data, and it appears to me that GFAFS quoted the wrong number for MTBF/MTTF at 107C. 12,660,980 hours is the actual figure, so you're right, the MTBF goes down from 107C to 114C.
 
The hours for the 78C example are also off, according to my verification with the link BFAFS provided.
 
http://www.onsemi.com/Pow...do?part=NCP81382HMNTXG
http://www.onsemi.com/Pow....do?part=NCP81382MNTXG
 
 
The corrected figures are:
 
An imaginary card with properly contained VRM heat at 78°c (close to the market average):
0.7eV - 78°c   - 90% confidence - MTBF/MTTF: 73,950,633 hours - 13.52 FITS
0.7eV - 107°c - 90% confidence - MTBF/MTTF: 12,660,980 hours - 78.98 FITS
0.7eV - 114°c - 90% confidence - MTBF/MTTF: 8,603,197 hours - 116.24 FITS
 
Does these numbers mean that one of these particular components operating at constant 107°c should last, on average, for 12,660,980 hours before failure? And what is a FITS?

The status of thermal pads change, now says "pending shipment" but whats means? 1. The ship its coming 2. The ship is taking more time.

Omarlink
The status of thermal pads change, now says "pending shipment" but whats means? 1. The ship its coming 2. The ship is taking more time.

Where can I check to see the status of my thermal pads order?

delicieuxz
 

gahelm

GFAFS
I told you day one, EVGA, playing with your customers will get you cornered and in full panic mod...didn't i?
 
Now to add some food for the thoughts i'll let statistical numbers do the talking (one may investigate if necessary):
 
Onsemi Reliability Data: NCP81382HMNTXG/NCP81382MNTXG
Without any mods 'being optional' ( aka Stock cards/no additional OC): VRMs Temps measured at 107°c to 114°c max
0.7eV - 107°c - 90% confidence - MTBF/MTTF: 7321258hrs - 78.98 FITS
0.7eV - 114°c - 90% confidence - MTBF/MTTF: 8603197hrs - 116.24 FITS
 
An imaginary card with properly contained VRM heat at 78°c (close to the market average):
0.7eV - 78°c - 90% confidence - MTBF/MTTF: 64869441hrs - 13.52 FITS 
 
The same can be done with the VRAM (temps are usually between -40°c ~95°c max for industrial grade, far less for commercial grade 0°c ~85°c max)

Why do your "Mean Time Between Failure" numbers go up with the higher temp?  Something is wrong with your stats dude....  Higher temps should mean more frequent failures thus your MTBF number should go down.  Higher MTBF = better, Lower MTBF = worse.  Not sure where this data is coming from or its relevance, just pointing out a problem.  Am I looking at your data wrong? 

I wondered the same as you, and I just looked up the data, and it appears to me that GFAFS quoted the wrong number for MTBF/MTTF at 107C. 12,660,980 hours is the actual figure, so you're right, the MTBF goes down from 107C to 114C.
 
The hours for the 78C example are also off, according to my verification with the link BFAFS provided.
 
http://www.onsemi.com/Pow...do?part=NCP81382HMNTXG
http://www.onsemi.com/Pow....do?part=NCP81382MNTXG
 
 
The corrected figures are:
 
An imaginary card with properly contained VRM heat at 78°c (close to the market average):
0.7eV - 78°c   - 90% confidence - MTBF/MTTF: 73,950,633 hours - 13.52 FITS
0.7eV - 107°c - 90% confidence - MTBF/MTTF: 12,660,980 hours - 78.98 FITS
0.7eV - 114°c - 90% confidence - MTBF/MTTF: 8,603,197 hours - 116.24 FITS
 
Does these numbers mean that one of these particular components operating at constant 107°c should last, on average, for 12,660,980 hours before failure? And what is a FITS?

My numbers are correct, yours too, you may check the formula and see that some values could change due to their randomized nature and error margin (fitness), but the FITS result remain the same (Failure In Time). 

emsir

Anarion

owagpus

Anarion
This is somewhat annoying... The BIOS changed the fan speed so that now it's quite often between 1100 RPM and 1200 RPM. However... At that range fan motors makes somewhat annoying noise and it's much more audible than even at 1350 RPM.

So I didn't imagine it. My 1070SC has the same strange whirring noise at exactly 33% (1210rpm) in EVGA Precision. At 32% (1170rpm) and 34% (1240rpm) there's no noise. Very odd.
Thankfully on my card the fans remain below 1100rpm @64C, which is the max temp in the games I tested. Maybe because I have a sidemounted fan in my case pointed directly at the graphics card?

In my case it makes that annoying noise at 32% and 33%. So it looks like it's a "feature" of this particular fan.

Just update primary bios again and you will have 0rpm up to 55 degrees.

Huh? Firstly, non-FTW cards have only one BIOS. Secondly that's not the issue here...
 

emsir

Anarion

clone
I must be one of the lucky ones as my 1070FTW runs very cool but ive still asked for the thermal pads.

GPU might run very cool but the VRM are could be well past the boiling point.

This is rubbish, you have no documentation to prove your statement. Don't post something you don't have a clue about.

It sounds like you don't have a clue. VRM is, pretty much without exception (just check out any review that has  thermal camera images), the hottest part on any GPU. Often because AIBs do not cool that are properly. External thermal camera shots don't even tell the full story because backside of the PCB is definitely cooler than the actual VRMs on the other side.