[Various] NVIDIA GeForce GTX 1080 Review Thread

[96Firebird]

And how do you know review sites don't warm up their cards before testing?

Who knows, the performance they show could be slower due to being at base clocks instead of boost clocks, and aftermarket cards will be even faster than FE results.

 

[sirmo]

And how do you know review sites don't warm up their cards before testing?

Who knows, the performance they show could be slower due to being at base clocks instead of boost clocks, and aftermarket cards will be even faster than FE results.

By looking at the performance delta observed. Jayztwocents modified the fan curve in order to get good results in his 1080 review (without disclosing it). He owned up to it a few weeks later once the reports of thermal throttling came out and he got called on it.

 

[tential]

By looking at the performance delta observed. Jayztwocents modified the fan curve in order to get good results in his 1080 review (without disclosing it). He owned up to it a few weeks later once the reports of thermal throttling came out and he got called on it.

https://www.youtube.com/watch?v=OXUo1S55ZUM

Looks like it thermal throttles and it is apparent because he was CALLED OUT on it.

I mean how much more evidence do you need...

The performance will be even slower than most reviews have shown once you taken into account cases. Quite a shame actually that Nvidia is continuing to show less and less care about the core things that made them a go to choice for so many consumers in the past.

 

[Keysplayr]

Your guessing Tential. Need cold hard numbers. Sorry. And again, we arent really hearing any end users reporting being pinned at base clocks, let alone not reaching advertised boost clocks. Have you?

 

[sirmo]

Your guessing Tential. Need cold hard numbers. Sorry. And again, we arent really hearing any end users reporting being pinned at base clocks, let alone not reaching advertised boost clocks. Have you?

The issue is corroborated by tons of sources.. Gamer's Nexus was probably the first one to break it.

We've also had a discussion about it here: http://forums.anandtech.com/showthread.php?t=2473396 Link to [H]'s article: http://hardocp.com/article/2016/05/17/nvidia_geforce_gtx_1080_founders_edition_review/5#.V1W8576DFBc

You can chose to ignore it though. But it's very much there on FE cards.

 

[sontin]

https://www.youtube.com/watch?v=OXUo1S55ZUM

Looks like it thermal throttles and it is apparent because he was CALLED OUT on it.

I mean how much more evidence do you need...

The performance will be even slower than most reviews have shown once you taken into account cases. Quite a shame actually that Nvidia is continuing to show less and less care about the core things that made them a go to choice for so many consumers in the past.

There is no thermal throttling. There is a reason why nVidia is calling it a boost. :thumbsup:

 

[tential]

Your guessing Tential. Need cold hard numbers. Sorry. And again, we arent really hearing any end users reporting being pinned at base clocks, let alone not reaching advertised boost clocks. Have you?

We are really hearing reports of throttling hence why review sites are testing this. Which is a shame because like I really think providing a proper cooling solution should be standard, especially from Nvidia.

It's not all bad, at least AIB cards will remedy the problem. The issue is Nvidia can't continue like this. How can this cooler be used on the 1080Ti? Or the new Titan without throttling?

It's a shame that the vast majority of reviews out there of the 1080 are not indicative of real world performance.

 

[Bacon1]

Your guessing Tential. Need cold hard numbers. Sorry. And again, we arent really hearing any end users reporting being pinned at base clocks, let alone not reaching advertised boost clocks. Have you?

https://hardforum.com/threads/gtx-1...read-post-yours-here.1901047/#post-1042332692

Not to mention in this thread the only one trying to "refute" throttling issues was not only being throttled but not even running @ 100% usage.

 

[Keysplayr]

https://www.youtube.com/watch?v=OXUo1S55ZUM

Looks like it thermal throttles and it is apparent because he was CALLED OUT on it.

I mean how much more evidence do you need...

The performance will be even slower than most reviews have shown once you taken into account cases. Quite a shame actually that Nvidia is continuing to show less and less care about the core things that made them a go to choice for so many consumers in the past.

There are newer drivers an ive heard corrected fan profiles in the retail release. There were reports of erratic fan behavior continually oscillating its speed from low to high. This would most certainly explain any excessive throttling in reviews and why we dont hear about it from the owners of retail cards. What more do YOU require to see that you really jumped the gun on this one and now have to stick to your guns based on pre release drivers.
If we start to hear end users complain that there not even reaching advertised boost clocks, then Id say you have a leg to stand on. As of now, your on stumps.

 

[Keysplayr]

We are really hearing reports of throttling hence why review sites are testing this. Which is a shame because like I really think providing a proper cooling solution should be standard, especially from Nvidia.

It's not all bad, at least AIB cards will remedy the problem. The issue is Nvidia can't continue like this. How can this cooler be used on the 1080Ti? Or the new Titan without throttling?

It's a shame that the vast majority of reviews out there of the 1080 are not indicative of real world performance.

Sources please.

 

[Bacon1]

There are newer drivers an ive heard corrected fan profiles in the retail release. There were reports of erratic fan behavior continually oscillating its speed from low to high. This would most certainly explain any excessive throttling in reviews and why we dont hear about it from the owners of retail cards. What more do YOU require to see that you really jumped the gun on this one and now have to stick to your guns based on pre release drivers.
If we start to hear end users complain that there not even reaching advertised boost clocks, then Id say you have a leg to stand on. As of now, your on stumps.

Sources please

 

[Olecki]

The framing of the 1080 has been very misleading.
Testing the 1080 in benchmarks where the bench lasts a short time and it's in an open bench setup doesn't reflect real world cases where the performance on the 1080 can be 10-15% higher than in a real world case.

As described on PurePC (portal quoted by you) about testing procedure for GTX 1080 Gaming X. As far as I know they are also using ot for all cards, including FE:

Temperatura otoczenia podczas testów wynosiła 24 stopnie Celsjusza (+/- 1). Pomiary przeprowadzane były w obudowie, która otrzymała komplet 120 mm wentylatorów (2x front i góra, 1x ty&#322. Taka konfiguracja zapewnia środowisko pomiarów zbliżone do warunków w jakich karty pracują w typowych komputerach. Na wykresach dłuższy pasek oznacza temperaturę w Wiedźmin 3 po dwudziestu minutach maksymalnego obciążenia

You can use google translator to check it, but I can quickly brief it: all test are done in closed case (with 3 fans) after 20 minutes of max load. And no, they don't observe real throtling below declared boost clocks both in GTX 1080 Gaming X AND 1080 FE.

 

[tential]

There are newer drivers an ive heard corrected fan profiles in the retail release. There were reports of erratic fan behavior continually oscillating its speed from low to high. This would most certainly explain any excessive throttling in reviews and why we dont hear about it from the owners of retail cards. What more do YOU require to see that you really jumped the gun on this one and now have to stick to your guns based on pre release drivers.
If we start to hear end users complain that there not even reaching advertised boost clocks, then Id say you have a leg to stand on. As of now, your on stumps.

Except we do and have had owners of retail cards complain.
This is undoubtedly a worse cooled card than the GTX 980. Is there even an improvement in cooling since the GTX 980?
How was this a more premium solution?

The power consumption of both cards is similar yet the GTX 1080 has throttling issues and the GTX 980 doesn't have those same issues which is very odd.

 

[tential]

Since boost clocks are not ensured and the 1080's cooler prevents people from seeing them except in the most ideal situations, we should take the 1080's base speed and compare it to cards. And separate the OC results out.

 

[Keysplayr]

Except we do and have had owners of retail cards complain.
This is undoubtedly a worse cooled card than the GTX 980. Is there even an improvement in cooling since the GTX 980?
How was this a more premium solution?

The power consumption of both cards is similar yet the GTX 1080 has throttling issues and the GTX 980 doesn't have those same issues which is very odd.

Ok. Sources please.

 

[96Firebird]

Since boost clocks are not ensured and the 1080's cooler prevents people from seeing them except in the most ideal situations, we should take the 1080's base speed and compare it to cards. And separate the OC results out.

Since you're assuming the cards "throttle" to base clocks, we can all assume the review results are at base clocks. Perfect, more performance to come!

 

[tential]

Since you're assuming the cards "throttle" to base clocks, we can all assume the review results are at base clocks. Perfect, more performance to come!

Since the reviews were not done in cases the reviews are indicative of what happens in an open bench setup. So there is less performance once the end user gets this as evidenced by what Jay's video I posted earliere.

Within a case, even more throttling would occur.

 

[Keysplayr]

Since the reviews were not done in cases the reviews are indicative of what happens in an open bench setup. So there is less performance once the end user gets this as evidenced by what Jay's video I posted earliere.

Within a case, even more throttling would occur.

You assume way too much. A properly cooled case could have greater cooling than an open test bed. More controlled airflow over and through components.

Listen. Go get us some sources of users with 1080FE gpus with the latest retail drivers that have proof of heavy base clock pinned and not reaching advertised boost clocks as you claim.

 

[96Firebird]

Within a case, even more throttling would occur.

And with this, I'll leave you to your assumptions.

 

[Erenhardt]

As described on PurePC (portal quoted by you) about testing procedure for GTX 1080 Gaming X. As far as I know they are also using ot for all cards, including FE:

You can use google translator to check it, but I can quickly brief it: all test are done in closed case (with 3 fans) after 20 minutes of max load. And no, they don't observe real throtling below declared boost clocks both in GTX 1080 Gaming X AND 1080 FE.

Nope, you are wrong. Only temperature and noise tests are done in a case. Gaming performance test are done on the open bench test bed. Even more, they do not heat up the cards before running benchmark.

 

[tential]

You assume way too much. A properly cooled case could have greater cooling than an open test bed. More controlled airflow over and through components.

Listen. Go get us some sources of users with 1080FE gpus with the latest retail drivers that have proof of heavy base clock pinned and not reaching advertised boost clocks as you claim.

You assume way too much. Not everyone has properly cooled cases with great airflow that is controlled to provide better cooling than an open test bed.

Listen, the point is, for $100 extra, Nvidia provided a substandard cooler that throttled.

Last generation, in the same price class, AMD provided a watercooler, free of charge, on the Fury X, that did not throttle, and had great temps/noise.

Nvidia is getting complacent by offering WORSE cooling on the GTX 1080, than the GTX 980, while charging more money than the GTX 980, and while offering less overclocking gains with the GTX 1080.

To let the GTX 1080 truly shine, Nvidia needed to provide a proper cooler, a proper BIOS (Since high OCs are only obtainable on different overclocking BIOS that the LN2 crowd are using), and proper power delivery. Without all of that, the GTX 1080 got released with not only an inadequate cooler, but also other issues which amounted to just OK stock performance in REAL WORLD CASES, and poor OC capabilities.

That's a shame for Nvidia's current flagship product.

A flagship product should be a premium product. Not this, and not this especially in the cooling department. Throttling is not acceptable.

 

[sirmo]

+1

 

[Headfoot]

And how do you know review sites don't warm up their cards before testing?

Who knows, the performance they show could be slower due to being at base clocks instead of boost clocks, and aftermarket cards will be even faster than FE results.

Some sites actually do warm up the reviewed item before tests. IIRC Anandtech does this for phones after all of the benchmark debacles using very high but very temporary boost states in the ill-fated 20nm gen SoCs.

It will usually say in the "testing methodology" section of the review

 

[LTC8K6]

You assume way too much. Not everyone has properly cooled cases with great airflow that is controlled to provide better cooling than an open test bed.

Listen, the point is, for $100 extra, Nvidia provided a substandard cooler that throttled.

Last generation, in the same price class, AMD provided a watercooler, free of charge, on the Fury X, that did not throttle, and had great temps/noise.

Nvidia is getting complacent by offering WORSE cooling on the GTX 1080, than the GTX 980, while charging more money than the GTX 980, and while offering less overclocking gains with the GTX 1080.

To let the GTX 1080 truly shine, Nvidia needed to provide a proper cooler, a proper BIOS (Since high OCs are only obtainable on different overclocking BIOS that the LN2 crowd are using), and proper power delivery. Without all of that, the GTX 1080 got released with not only an inadequate cooler, but also other issues which amounted to just OK stock performance in REAL WORLD CASES, and poor OC capabilities.

That's a shame for Nvidia's current flagship product.

A flagship product should be a premium product. Not this, and not this especially in the cooling department. Throttling is not acceptable.

The AMD water cooler that was often a pain in the butt? That water cooler?

I seem to remember the intro of that water cooler a little differently than you do.

 

[Bacon1]

Ok. Sources please.

Still waiting for your sources claiming newer drivers have fixed the issues or that the issues are fixed in any way. Testers proved there are throttling issues when actually used in a case for an extended period of time (actual gaming, not just running benchmark).