First GTX1070 review

[tential]

Two different cards using the same gp104 chip. Secondly Producing more heat and consuming more power is virtually the same thing in this case (being an EECE major has its' perks lol). As for the rest of your post, refer back to my earlier post in which I addressed your concerns in a somewhat concise manner.

The discussion isn't about Vega though it's about the 980 ti and the 1070.

Obviously bringing AMD into the mix brings a whole other set of variables such as vendor bias and what brand people prefer to own. When it comes to 1070 vs 980Ti though it's pretty obvious the 1070 is the better card but:

Two different cards, but lets go ahead and assume the same thing applies. It boils down to this... If all you're interested in is marginally better performance in what will soon be legacy games, and don't mind paying more for an EOL card that produces more heat, consumes more power and will be slower in multiple other areas like DX12, VR, potentially other API's like Vulkan, and anything that could benefit from an extra 2GB VRAM, and don't mind the very real possibility that driver support will be a 2nd priority compared to the current gen cards, then the 980Ti is the card for you.

 

[stormkroe]

The problem is, this type of performance used to be expected, if not even more for less, from node shrinks and new card releases. Its the over pricing FE and over pricing in general of the "mid range" into high end that has people upset.

Anyone who's paid attention to GPUs over the last decade can tell you the are making a killing off of them.

If all I cared about was share prices I would be excited, but I care about price / performance.

Doesn't it make sense to attribute the lower-than-usual increase from this node shrink to the fact that we had 28nm for so long? That process was put through its paces and probably had every trick used to extract maximum performance out of it. So maybe we can shift some of the blame to the outgoing node's maturity.

 

[Bacon1]

Doesn't it make sense to attribute the lower-than-usual increase from this node shrink to the fact that we had 28nm for so long? That process was put through its paces and probably had every trick used to extract maximum performance out of it. So maybe we can shift some of the blame to the outgoing node's maturity.

No, because we should have had the leap @ 20nm which was skipped already. So this is 2x node shrink. I mean you can compare the price/perf against AMD's 300 series which is the 200 series which came out in 2013.

 

[jabroni619]

Two different cards using the same gp104 chip. Secondly Producing more heat and consuming more power is virtually the same thing in this case (being an EECE major has its' perks lol). As for the rest of your post, refer back to my earlier post in which I addressed your concerns in a somewhat concise manner.

I've referred back to your post, and doesn't address the rest of my post, it just adds another variable which we weren't even talking about. I think I illustrated quite clearly in my last post, who should be buying a 980Ti over a 1070 at this point.

As to the 4GB argument. DOOM won't even allow you to select the highest possible settings with a 4GB card.

 

[The Alias]

I've referred back to your post, and doesn't address the rest of my post, it just adds another variable which we weren't even talking about. I think I illustrated quite clearly in my last post, who should be buying a 980Ti over a 1070 at this point.

As to the 4GB argument. DOOM won't even allow you to select the highest possible settings with a 4GB card.

Dude if you're bringing up performance degradation and dx12 deficiencies, you have to bring up gcn as that is the reason you're bringing it up in the first place. Also your definition of marginal and legacy is questionable. and what your basing it on are ref 980ti vs 1070fe one of which is known to throttle even when oc'ed

 

[jabroni619]

Dude if you're bringing up performance degradation and dx12 deficiencies, you have to bring up gcn as that is the reason you're bringing it up in the first place. Also your definition of marginal and legacy is questionable. and what your basing it on are ref 980ti vs 1070fe one of which is known to throttle even when oc'ed

Dude, no you absolutely do not. Why do I need to bring up GCN when we are talking about 1070 vs 980Ti? What if one has a gsync monitor? Your argument is falling apart so you're trying to shift goal posts to something that isn't even part of the discussion lol.

If by throttle you mean maintain higher then advertised boost clocks even while throttling, then it throttles. I'll take a higher then advertised performance any day.

What else have you got? Shall we talk about AGP cards next?

 

[The Alias]

Dude, no you absolutely do not. Why do I need to bring up GCN when we are talking about 1070 vs 980Ti? What if one has a gsync monitor? Your argument is falling apart so you're trying to shift goal posts to something that isn't even part of the discussion lol.

If by throttle you mean maintain higher then advertised boost clocks even while throttling, then it throttles. I'll take a higher then advertised performance any day.

What else have you got? Shall we talk about AGP cards next?

I was talking about the 980Ti reference card guru3d review used. If the 1070 was still throttling back when being ran at 100% boost, that'd be another issue entirely. All I'm trying to do is make sense of the information available to us a d properly assign asterisks where they need to be assigned. I'm trying to assemble a new pc before the school year to play games on during my free time (never, but it's still cool to look at in the dorm haha). you must understand $400 is a BUNCH of money and spending it is a big decision.

 

[VulgarDisplay]

Dude, no you absolutely do not. Why do I need to bring up GCN when we are talking about 1070 vs 980Ti? What if one has a gsync monitor? Your argument is falling apart so you're trying to shift goal posts to something that isn't even part of the discussion lol.

If by throttle you mean maintain higher then advertised boost clocks even while throttling, then it throttles. I'll take a higher then advertised performance any day.

What else have you got? Shall we talk about AGP cards next?

The problem with this is that the advertised performance from review sites doesn't account for throttling unless they take the time to get the card good and hot prior to the benchmark run.

So you are actually being fooled into paying for a certain benchmark number, but in reality your gpu isn't as fast as advertised.

Sent from my XT1575 using Tapatalk

 

[OatisCampbell]

The problem with this is that the advertised performance from review sites doesn't account for throttling unless they take the time to get the card good and hot prior to the benchmark run.

So you are actually being fooled into paying for a certain benchmark number, but in reality your gpu isn't as fast as advertised.

Sent from my XT1575 using Tapatalk

The problem with your idea is this:

Where throttling will occur (or not) varies for every computer and every room.

Things like whether or not a person has a water loop dumping cpu heat out of the case, case ventilation, and room temperature vary widely.

 

[Keysplayr]

The problem with this is that the advertised performance from review sites doesn't account for throttling unless they take the time to get the card good and hot prior to the benchmark run.

So you are actually being fooled into paying for a certain benchmark number, but in reality your gpu isn't as fast as advertised.

Sent from my XT1575 using Tapatalk

Pretty sure all dynamically clocked cards today do the very same thing. So its all relative.

They get hot, they throttle back their boost/turbo/nitro methane burning clocks to match temp targets.

Still, pascal seems to stay above advertised boost clocks. And it performs quite well.

 

[littleg]

The problem with your idea is this:

Where throttling will occur (or not) varies for every computer and every room.

Things like whether or not a person has a water loop dumping cpu heat out of the case, case ventilation, and room temperature vary widely.

Yes but the vast majority of cards will be run inside a case, not an open test bench and will be loaded for longer than a 2 minute benchmark.

 

[Keysplayr]

Yes but the vast majority of cards will be run inside a case, not an open test bench and will be loaded for longer than a 2 minute benchmark.

What is the average percentage of performance lost when throttling?

 

[VulgarDisplay]

The problem with your idea is this:

Where throttling will occur (or not) varies for every computer and every room.

Things like whether or not a person has a water loop dumping cpu heat out of the case, case ventilation, and room temperature vary widely.

Exactly.

It's not a problem with my idea. I just illustrates why everyone should wait for non founders edition cards. Why everyone feels the need to defend overpriced cards with garbage coolers is beyond me.

Yes it's new shiny tech which is always exciting, but these are not good products and should not be defended as such. Especially for the inflated price.

 

[tential]

Exactly.

It's not a problem with my idea. I just illustrates why everyone should wait for non founders edition cards. Why everyone feels the need to defend overpriced cards with garbage coolers is beyond me.

Yes it's new shiny tech which is always exciting, but these are not good products and should not be defended as such. Especially for the inflated price.

I don't know why you would NOT wait for AIB cards and it's crazy that people are accepting the fact that their card throttles and doesn't sustain clocks now.

Anything to justify your GPU choice.

 

[ShintaiDK]

I don't know why you would NOT wait for AIB cards and it's crazy that people are accepting the fact that their card throttles and doesn't sustain clocks now.

Anything to justify your GPU choice.

Retail cards dont seem to throttle. Sorry to bust that one

 

[Hitman928]

Retail cards dont seem to throttle. Sorry to bust that one

Interesting. I guess these users must have purchased reviewer cards then?

Jerzeeloon

Got my founders edition today. Ran some benchmarks. This thing hits 83° and throttles like nobody's business. Had to make a custom fan profile for 90% fan @ 80° just to keep from throttling.

https://forums.geforce.com/default/topic/938624/geforce-1000-series/user-feedback-thread-gtx-1080-/post/4893152/#4893152

Totally agree with you. This card throttles really bad, had I known this before, then I would have skipped my purchase as well.

https://forums.geforce.com/default/topic/938624/geforce-1000-series/user-feedback-thread-gtx-1080-/post/4893988/#4893988

 

[tential]

Not to mention having stupidly high fancurve's just to prevent the card from throttling....

I thought AIB 1080s were supposed to be available shortly afterwards, where are they?

 

[Hitman928]

Not to mention having stupidly high fancurve's just to prevent the card from throttling....

I thought AIB 1080s were supposed to be available shortly afterwards, where are they?

It's only been a week since launch. Seems longer since nda lifted on the 17th. I think custom cards are expected in 2-3 weeks.

 

[ShintaiDK]

Not to mention having stupidly high fancurve's just to prevent the card from throttling....

And again you are wrong

My GTX1080 is more silent than my GTX980.

 

[ShintaiDK]

Interesting. I guess these users must have purchased reviewer cards then?


https://forums.geforce.com/default/...edback-thread-gtx-1080-/post/4893152/#4893152


https://forums.geforce.com/default/...edback-thread-gtx-1080-/post/4893988/#4893988

No numbers are listed. Sure my card also throttles...down from 1900Mhz+ to 1800Mhz something. Way above expected boost clock.

 

[Sweepr]

Thanks for sharing your real end user experience Shintai. Looking forward to more opinions/impressions.

 

[Hitman928]

No numbers are listed. Sure my card also throttles...down from 1900Mhz+ to 1800Mhz something. Way above expected boost clock.

So because in your situation you don't see throttling, no one does? Hard to believe one user experience with no actual evidence or context against a large group of professional reviewers all experiencing throttling and now users also reporting the exact same behavior even if they don't include the actual frequencies. Just for fun though, here's another:

At stock settings the fan was capping around 50-55% which, while quiet, seemed to hit the 80-82 degree ceiling relatively quickly when playing GTA V maxed out at 4K resolutions (minus AA, since 4K on a 27" does a pretty good job at masking aliasing). This would cause the boost clock to vary quite a bit, constantly going up and down and sometime dipping below the max stock boost clock (1733 MHz) but I don't recall ever seeing it dip below stock clock (1607 MHz). By changing the fan profile to run the fan at around 50% at around 60 degrees, 75% at 80 degrees, and 100% for everything else above my boost clock has done a much better job at settling to a fairly fixed rate and stays pretty consistently above the stock boost clock with no other changes. My boost clock during extended plays in GTA V so far appear to stay over 1800 MHz, settling around 70-100 MHz over stock boost clock for the duration of game play.

https://hardforum.com/threads/gtx-1080-overclocking-results-thread-post-yours-here.1901047/#post-1042332423

The good news for FE is that from the user reports I've read, unlike the 290x reference, the sound profile of the card is not bad even when pushing the fan to 70+%.

 

[jabroni619]

So because in your situation you don't see throttling, no one does? Hard to believe one user experience with no actual evidence or context against a large group of professional reviewers all experiencing throttling and now users also reporting the exact same behavior even if they don't include the actual frequencies. Just for fun though, here's another:



https://hardforum.com/threads/gtx-1...read-post-yours-here.1901047/#post-1042332423

The good news for FE is that from the user reports I've read, unlike the 290x reference, the sound profile of the card is not bad even when pushing the fan to 70+%.

If you also read the HardOCP review, they say the same thing he did. It throttles but even after the throttling it maintains higher then advertised boost clocks. I don't see the fuss, and this non issue will be even less of an issue with AIB cards.

 

[Face2Face]

From what I gather, the GTX 1080 FE's cooler is about as loud as the cooler that was on my reference GTX 780. While 70% fan speeds sounds like a lot, the cooler was extremely quiet at 70%. I compared it to the Accelero cooler that was on my 7950 @ 100% fan speed, which is very quiet.

I guess the real issue is why didn't NVIDIA set A more aggressive fan curve on the FE cards? Especially since the blower is pretty quiet at 70%.

Joker kept his from throttling using 74% fan speed, and keep in mind this was overclocked 2GHz + in a Define R5 case.

https://www.youtube.com/watch?v=mrRhV6qfrr0 Around the 4min mark.

 

[Keysplayr]

Retail cards dont seem to throttle. Sorry to bust that one

All cards throttle.