[Various] NVIDIA GeForce GTX 1080 Review Thread

[Bacon1]

Nvidia can't adjust the fan speeds or anything to do with hardware you need a Bios update not a driver update because there will be many different fans also profiles soon with AIB's partners.

Thats not true, there have been many drivers that have altered how fan profiles work, including making them not work at all.

Not to mention:

http://www.guru3d.com/news-story/geforce-gtx-1080-fan-issues-to-be-solved-with-driver-update.html

 

[ShintaiDK]

Great since your 1080 doesn't throttle you can make a YouTube video showing how it doesn't throttle. Just show your fan curve then let a game warm the system up for 20 minutes at load then record the Temps/clockspeed at 100% gpu usage so we can see how well the 1080 holds up and whether they fixed the throttling issue.

Please don't make this personal let's stay on topic. This is about the 1080 and not about my own personal preferences. It's an unfair attack to say I have no interest in Nvidia products.

Either way though, like I said, 1080s may throttle based on the substandard cooler. If you have low enough ambient Temps or a golden chip of course you may be one of the lucky few with a non throttling chip. Just like there were people with fury x with out pump noise. Doesn't change the fact many people had a substandard experience.

It's a shame that people are willing to ignore Nvidia's substandard cooling situation this time around. We need to have high expectations of Nvidia if we want them to continue operating at a high level.

Why? So you can reject that as well like the previous test I made?

Lets be honest here, you dont have any interest in if the retail card throttle or not. You already decided.

 

[kraatus77]

so you agree your card throttles on 100% load ? regardless you are the last person on earth after jhh who will admit something negative about nvidia. lol


Infraction issued for Trolling.

-Rvenger

 

[ShintaiDK]

so you agree your card throttles on 100% load ? regardless you are the last person on earth after jhh who will admit something negative about nvidia. lol

Below advertised boost clock? No.

 

[antihelten]

Why? So you can reject that as well like the previous test I made?

Problem is that the test you made is more of a best case scenario and not worst case scenario like you claimed.

• You used a Total War game to test, which is primarily a CPU limited game, not a GPU limited game (unless you run it at 4K). This is also evident from the fact that your GPU wasn't running at full load.
• You used an SG08 case where the GPU can draw fresh air directly from the outside, which is about as close to an open air scenario that you can get whilst still using a case.
• You turned of the case fan which doesn't affect the GPU (since as mentioned it draws air from the outside), but does potentially affect the CPU, thus causing it to throttle and making the test even more CPU bottlenecked.

If you actually run the 1080 in a case where the GPU can't draw air directly from the outside (which would be the vast majority of cases), and test with a GPU heavy game instead, then there's clear evidence that the 1080 FE does indeed throttle.

 

[ShintaiDK]

The case is packed in a corner. The "free air" gap to the wall is 3.5cm. And the wall itself got quite warm since it more or less ends up circulating from the rear output that ends behind the monitor. In other words, even if it was the case, the GPU would draw in air in that is way above ambient.

Seems you decided to ignore that from the post you else seem to remember so well:
http://forums.anandtech.com/showpost.php?p=38269034&postcount=216

So no, its certainly not a best case or equal to open air.

Warhammer wasn't CPU limited under the test.

 

[Qwertilot]

Just to repeat this - that Tom's chart definitely showed that the cooler can happily cope with holding an overclock around 2000-2050, let alone stock levels.

For pities sake, its a cooler from a ~240/50W chip so its really very hard to see it not coping on a priori basis. The noise level when doing so will be another matter of course.

As an aside on noise, that isn't just subjective, but also relative to how quiet the rest of your computer is.

 

[iiiankiii]

Below advertised boost clock? No.

As long as you're willing to sacrifice some potential performance, I really think the blower style cooler is ok. Nvidia knows what they're doing. At "stock" speed, there shouldn't be any problem at all. I just can't find myself ever using a blower style cooler again because the noise level is unbearable once you start to overclock.

 

[showb1z]

Don't lose sight of the real argument.

http://forums.anandtech.com/showpost.php?p=38242302&postcount=87

1) Crazy overclocking nope
2) Craftsmanship nope
3) $100 premium yep

If Nvidia hadn't made such an obnoxiously big deal out of FE nobody would care about this.
Then again, 1 out of 3 isn't bad. And at least you get the chance to pad the margins of your favorite company so much gloating can be done next quarterly report.

 

[antihelten]

The case is packed in a corner. The "free air" gap to the wall is 3.5cm. And the wall itself got quite warm since it more or less ends up circulating from the rear output that ends behind the monitor. In other words, even if it was the case, the GPU would draw in air in that is way above ambient.

Seems you decided to ignore that from the post you else seem to remember so well:
http://forums.anandtech.com/showpost.php?p=38269034&postcount=216

So no, its certainly not a best case or equal to open air.

Warhammer wasn't CPU limited under the test.

So the card is drawing in air that is 28-37C (mixture of ambient and wall temp), that is still significantly cooler than the air inside the case (47.6C). Thanks for proving my point I guess.

And as far as Warhammer not being CPU limited, how do you know this? You didn't include any numbers for the CPU as far as I can tell, plus your GPU was running at less than full load (only about 90% load).

I didn't say it was the absolute best case scenario nor that it was equal to an open air bench, but simply that is was closer to a best case scenario rather than a worst case scenario like you claimed.

 

[Silverforce11]

Just to repeat this - that Tom's chart definitely showed that the cooler can happily cope with holding an overclock around 2000-2050, let alone stock levels.

For pities sake, its a cooler from a ~240/50W chip so its really very hard to see it not coping on a priori basis. The noise level when doing so will be another matter of course.

As an aside on noise, that isn't just subjective, but also relative to how quiet the rest of your computer is.

At 100% fanspeed.

Throttling shouldn't occur at such airflow.

Heck, the crap blower on the 290/X, if you crank it up to 70%, you can mine coins (a much heavier load than gaming) with it with zero throttling. I did it for a bunch of them.

Doesn't mean the cooler is capable. Because it's operating way beyond acceptable noise levels.

Either way, lessons have been learnt and I hope you guys demand better of Founder's Edition "premium" GPUs the next time it happens again, ie. 1080Ti..

 

[Qwertilot]

Well yes, but that OC is well past standard boost clocks, so it should be able to cope at sane enough fan speeds. AIB stuff will be much quieter/slightly cheaper of course.

I'm not defending the FE as value for money (unless a huge part of value is getting hold of it now), just trying to temper some of the hate.

It does look very much like ~180W is getting near to the sane limit of this cooler though, and if it gets used unchanged on the 1080ti/Titan it might well be approaching actively offensive.

 

[ShintaiDK]

So the card is drawing in air that is 28-37C (mixture of ambient and wall temp), that is still significantly cooler than the air inside the case (47.6C). Thanks for proving my point I guess.

And as far as Warhammer not being CPU limited, how do you know this? You didn't include any numbers for the CPU as far as I can tell, plus your GPU was running at less than full load (only about 90% load).

I didn't say it was the absolute best case scenario nor that it was equal to an open air bench, but simply that is was closer to a best case scenario rather than a worst case scenario like you claimed.

Do you really think the air was 37C or below if a reinforced 180mm concrete wall 3½cm away will reach 36.7C in ~20mins?

I can monitor CPU load on my own thank you

 

[LTC8K6]

Perhaps some of the cards have worse thermal characteristics than normal?

It seems pretty clear that we have some claims of throttling, and some claims of not throttling.

Why would it be unusual to have some cards with a problem, and some cards okay?

 

[antihelten]

Do you really think the air was 37C or below if a reinforced 180mm concrete wall 3½cm away will reach 36.7C in ~20mins?

I can monitor CPU load on my own thank you

Yes I absolutely believe the air was below 37C, otherwise you wouldn't have been able to stay in the room (at least not comfortably so). Also I don't see any evidence that the entire wall was 37 C after 20 mins (that would be fairly close to impossible given the heat capacity of a wall that size, and the total heat being generated by your system), only that a spot on the wall near the exhaust of the case was 37 C.

And the fact that you can monitor your CPU doesn't really tell the rest of us anything. I have already shown you evidence of TW: Warhammer being CPU bottlenecked, you haven't shown anything to the contrary.

 

[Erenhardt]

Do you really think the air was 37C or below if a reinforced 180mm concrete wall 3½cm away will reach 36.7C in ~20mins?

I can monitor CPU load on my own thank you

Yes. Do you know how the heat transfer mechanisms? Heat radiation could make that wall 60'C yet the intake air could almost at the room temp level.

Funny you didn't show Core clocks graph., it only shows max/min and current clock. We don't know how low it dropped. Or we can assume the 139Mhz is the throttling point of your GPU at which point the discussion is pointless.

 

[Qwertilot]

Perhaps some of the cards have worse thermal characteristics than normal?

It seems pretty clear that we have some claims of throttling, and some claims of not throttling.

Why would it be unusual to have some cards with a problem, and some cards okay?

It seems moderately established that the review cards shipped with a overly passive fan curve enabled. A bit unusually careless for NV perhaps.

The noise while not throttling is the potential worry (and AIB stuff will be >> at this.).

 

[Bacon1]

Do you really think the air was 37C or below if a reinforced 180mm concrete wall 3½cm away will reach 36.7C in ~20mins?

I can monitor CPU load on my own thank you

Yes and we can all see that your GPU never hit 100%. Max was 97% and average seemed to be around 90%.

So we know it was CPU limited.

 

[Baasha]

Any update on when the 4-Way SLI thing will be unlocked?

 

[Raising]

Asus GTX1080 Strix Review

Video

Amazing default factory overclock, up to 2.05ghz core, he only managed to oc to 2.1ghz though.. wondering if this will be the limit for these early chips..

If I end up selling my last 980ti I'll probably grab that strix

 

[tential]

Asus GTX1080 Strix Review

Video

Amazing default factory overclock, up to 2.05ghz core, he only managed to oc to 2.1ghz though.. wondering if this will be the limit for these early chips..

If I end up selling my last 980ti I'll probably grab that strix

People broke that with custom bios from Nvidia. Just wait for it to be figured out imo although it's not a big upgrade
Side grade get a 1080ti

 

[brandonmatic]

Asus GTX1080 Strix Review

Video

Amazing default factory overclock, up to 2.05ghz core, he only managed to oc to 2.1ghz though.. wondering if this will be the limit for these early chips..

If I end up selling my last 980ti I'll probably grab that strix

He only tried automatic OC - not manual tuning, according to the article.

Rather than spend days tweaking and testing to only achieve an additional 50MHz, and force you to take up more of your busy lives reading carbon-copy results, we're solely utilising the automatic overclocking for the foreseeable future.

This has three main benefits. Firstly it gives you results that you'll achieve, freed from the worry that your model wont achieve the same overclock. Secondly, it frees up our time to bring you a wider variety of product reviews, and lastly it makes our graphs easier to read by simplifying them.