Nvidia Pascal Reviews [1080XP, 1080ti, 1080, 1070ti, 1070, 1060, 1050, and 1030]

pharma said:
EGVA Classified .... 2.2 Ghz Clock out-of-box


https://twitter.com/dimitry49/status/740517917326970880
Click to expand...
I think some EVGA staff posted a correction apologising and mentioning that it should be possible OC, rather than out-of-box.
I would need to try and find the original context mentioning this, until then taking as just "I think".
Cheers

Edit:
Found it: http://forums.evga.com/Evga-1080-GTX-Classified-22Ghz-m2491836.aspx#2492040
Sadly definitely not out of the box, but fingers crossed it is achievable as a further OC.
 
Last edited:
Dr Evil said:
guru3d's tests never show big differences in noise levels, which is very odd...

This table under load is a joke.

http://www.guru3d.com/articles_pages/asus_rog_strix_geforce_gtx_1080_review,11.html
Click to expand...
Well 3db difference would be roughly 23% volume louder and double acoustic sound intensity, so I would not expect them to vary too much more in general as this is pretty noticable.

I cross checked TPU that also use a meter, and they have same 3db difference between the Nvidia 1080FE and AMD 290X.
They have different base measurements but the relative values correlate.
Cheers
 
CSI PC said:
Well 3db difference would be roughly 23% volume louder and double acoustic sound intensity, so I would not expect them to vary too much more in general as this is pretty noticable.

I cross checked TPU that also use a meter, and they have same 3db difference between the Nvidia 1080FE and AMD 290X.
They have different base measurements but the relative values correlate.
Cheers
Click to expand...

On that table they have Gigabyte's custom 1080, Fury X and original Titan 2X SLI putting the same amount of noise, I guarantee you, that is not correct.
 
Dr Evil said:
On that table they have Gigabyte's custom 1080, Fury X and original Titan 2X SLI putting the same amount of noise, I guarantee you, that is not correct.
Click to expand...
I cannot comment about the SLI, but their original Titan single GPU matches TomsHardware.
And we know the relative difference for 1080FE and 290x matches with another publication in TPU.

All 3 sites use meters, but yeah first glance some of those measurements do look grouped and maybe a few are anomalies.
One IMO could be the Fury X as it should be lower with load, but depends how bad the whine was I guess.

That said I would expect to see some kind of clustering because 3db is a big difference and the more cards you put on the chart the more you see grouped, made worst by not providing decimal point.
And I think most of the older measurements may not be as accurate as the more recent ones; depends when they all purchased a decent calibrated meter.

Cheers
 
Last edited:
I.S.T. said:
How is that possible?! It makes no fucking sense!
Click to expand...

They have some issues with their testing methodology or at the very least it does a very poor job of serving the reader.

CSI PC said:
I cannot comment about the SLI, but their original Titan single GPU matches TomsHardware.
And we know the relative difference for 1080FE and 290x matches with another publication in TPU.

All 3 sites use meters, but yeah first glance some of those measurements do look grouped and maybe a few are anomalies.
One IMO could be the Fury X as it should be lower with load, but depends how bad the whine was I guess.

That said I would expect to see some kind of clustering because 3db is a big difference and the more cards you put on the chart the more you see grouped, made worst by not providing decimal point.
And I think most of the older measurements may not be as accurate as the more recent ones; depends when they all purchased a decent calibrated meter.

Cheers
Click to expand...

Guru also has the 1080 Founders edition quieter than the Gigabyte custom model, whereas Techpowerup has it clearly reversed and that is just one example. I have basically zero trust in Guru's noise tests, whereas Techpowerup has seemed alright. Other sites put out results that have far more difference between cards. Guru3d is the anomaly here.
 
I.S.T. said:
How is that possible?! It makes no fucking sense!
Click to expand...
Cards are not consistent so each reviewer may have a slightly different loudness/whine/etc, position of meter and distance, open bench or case, use of decimal point or not.
But yeah I would want it verified myself.

Anyway as an example TPU results look unusual in one range if I remove the decimal point.
at 40dBa they have; GTX970, TitanX, 290X, 295X2,

To be honest, any loudness measurements should only be used as a rough guideline IMO, and see if relative loudness correlates to any other publication measurements.
Cheers
 
CSI PC said:
Anyway as an example TPU results look unusual in one range if I remove the decimal point.
at 40dBa they have; GTX970, TitanX, 290X, 295X2,
Click to expand...

The 970 is probably some type of error. They never reviewed the stock cooler model and suddenly 970's number rose from 37 to 40Db quite recently, they have had some other errors on their reviews also. Recently they had some really weird results with AMD 380X and now that card has been removed from their charts completely. The figure for 290X is taken with the "Quiet mode" Uber is 10Db more, but they don't bother to mention that anymore...
 
Guru3D testing technique:
For each dBA test we close the PC/chassis and move the dBA gun 75 cm away from the PC. Roughly the same proximity you'll have from a PC in a real-world situation.
Click to expand...
 
Malo said:
Well that's not very good practices. Probably more widespread than we know.
Click to expand...
Yeah I think you are right, considering how tight the custom AIB market is between them.
2% can be enough for it to stand out compared to a competitor, but then voltage-clockspeed can be monitored with independent utilities and publications should be able to catch anything going on outside scope of advertised product.
I guess though this is made more complex by how dynamic Boost3 is these days and its sensitivity with individual games.
Cheers
 
Pascal Memory Bandwidth issue linked to driver version v368.39.
The 2-year old mobile GTX 980M in Windows 8.1 is over 35 times faster than the desktop GTX 1080 for random memory reads for an allocation size of 2048 MB, which is a fairly common size for our applications.

Also the L2 cache size for the GTX 980M is not larger than the GTX 1080 so that is not the issue.

Any ideas why the GTX 1080 memory bandwidth is this poor?
Click to expand...

....
Quick update from the bug report (from NVIDIA);
Thank you for the reporting!

This issue could be reproduced locally when we tested on "Windows7sp1_x64/GTX 1080" system with v368.39 driver. However, it has been fixed now in our development driver builds, and the fix would be available for you in the next driver release from R367 family(v368.xx). Sorry for any inconvenience!

So sounds like good news and the problem will be fixed. Glad they were able to reproduce the issue..
Click to expand...


https://devtalk.nvidia.com/default/...mory-bandwidth-gtx-1080/post/4903672/#4903672
 
You must log in or register to reply here.

Similar threads

Rys
GPUDB integration FAQ
Replies
8
Views
3K
kaiser77_1982
K
BRiT
PSA: EVGA Cards using Nvidia 1080 and 1070 catching on fire
Replies
4
Views
3K
DuckThor Evil
DuckThor Evil
S
Pascal vs. Maxwell - a Generational Performance Comparison & Analysis
Replies
19
Views
13K
Kaotik
Kaotik
A
GP102/GP100 FP16 support vs. Performance in old DirectX 8 and DirectX 9
Replies
12
Views
5K
Razor1
R
R
Building a PC with XB1X specs *spawn
Replies
10
Views
2K
Blazkowicz
B
Back
Top