Geforce GTX 1060 Thread: faster than RX 480, 120W, $249

[Maverick177]

I don't know it's true or not.

http://www.3dmark.com/compare/fs/9202637/fs/9188962#

+200Mhz OC for the 1060 vs +84Mhz OC for the 480 and yet 1060 still loses 8.2%. Thought GP106 has no trouble winning against Polaris?

 

[AtenRa]

Would probably be fine at 1080p and also depends on how good Nvidias compression algorithms are. They've been pretty darn good.

2GB was fine for the GTX 770 two/three years ago but look how it affects its performance today. And this includes 11 GameWorks titles.

GTX 770 2GB review , GTX 770 4% faster than HD7970 GHz (R9 280X)

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_770/27.html

RX 480 review, R9 280X ~19% faster than GTX 770

https://www.techpowerup.com/reviews/AMD/RX_480/24.html

 

[Sweepr]

+200Mhz OC for the 1060 vs +84Mhz OC for the 480 and yet 1060 still loses 6.6%. Thought GP106 has no trouble winning against Polaris?

As already explained, 3DMark is listing maximum boost clock here. Same happens with Geforce GTX 1080, this is Founder's Edition at stock:

All cards in this comparison are using stock settings. The variance for GTX 1080 clocks is between 1860 to 1886 MHz, probably caused by new GPU Boost 3.0 technology.

The maximum achievable 1,886 MHz, the GeForce GTX 1080 Founders Edition are more of a theoretical value: In games you never see this clock. There, the highest ever measured frequency was during the tests 1,835 MHz, but was also the clock after a few seconds history. In demanding titles 1.785 MHz has finally placed as the highest, realistic clock out.

www.computerbase.de/2016-05/geforce-gtx-1080-test/6/#abschnitt_bis_zu_1785_mhz_takt_unter_dauerlast

 

[USER8000]

+200Mhz OC for the 1060 vs +84Mhz OC for the 480 and yet 1060 still loses 8.2%. Thought GP106 has no trouble winning against Polaris?

Yep,the GTX1060 boost clocks are meant to be 1.709GHZ and AIB AMD RX480 cards are meant to run around 1.325GHZ/1.35GHZ - that means an AIB overclocked GTX1060 is slower than an overclocked RX480.

 

[antihelten]

Probably the exact opposite reason you are not.

Faster is faster to us. (whoever you think "you guys" are)

And faster to you has to be a certain percentage faster, I.E. >5%
to be considered faster.

Who is right here, technically? The people saying it's faster even if it's a smidge? Or the people saying it has to be >5% faster to really be considered faster?

Who's argument would hold up in court if it had to?

I'm curious, why is it you think I'm not willing to accept the bare minimum from Nvidia (my reason apparently being the opposite of yours)?

You guys are you and Sweepr, seeing as you are the only ones I've been arguing with that seem perfectly happy with the 1060 only matching or being just 0.0001% faster than the 980 when Nvidia said the 1060 would be faster than GTX 980.

And of course I expect at least 5% faster, otherwise it's pointless. There isn't a person alive who would be able to tell the difference between a card running at X fps and another card running at X + 0-4.9% faster. So if the difference is less than 5% then in practise it's a draw.

And again I don't understand why you are happy with Nvidia being "technically correct". I mean sure if you work for Nvidia's PR department then you would feel fine about it, but as a consumer you can't possibly be satisfied with that.

And as far as who's argument would hold up in court, again what does it matter, are you a lawyer or a consumer?

As I said, for some reason you seem happy to accept the bare minimum from Nvidia, which as I said to Sweepr, seems like a really defeatist attitude to have for a consumer.

Irrelevant. Doesn't mean all products they claim to be faster need to be >5-10% faster.

Of course it doesn't need to be faster, But as I said above, do you work for Nvidia's PR department or are you a consumer? As the later I simply can't understand how you could be happy with anything less than a 5% improvement when Nvidia says the 1060 will be faster than the 980.

Because it's a 4.4 TFLOPs VGA based on a 1280 SPs die half a size of GM204 targeting $249+. Only matching GTX 980 or a tiny bit faster would already be a technical accomplishment.

So Nvidia lying about the performance of the 1060 would be okay to you because the specs are poor relatively speaking and it's cheap?

Yeah I'm sorry but I really don't agree with that logic.

It would also be a technical accomplishment if it actually lived up to Nvidia's claims.

 

[Sweepr]

Yep,the GTX1060 boost clocks are meant to be 1.709GHZ and AIB AMD RX480 cards are meant to run around 1.325GHZ/1.35GHZ - that means an AIB overclocked GTX1060 is slower than an overclocked RX480.

Right. If you think that's an OCed GTX 1060 and not simply 3DMark listing maximum GPU Boost 3.0 like it does with other Pascal cards, please explain to us how this 'stock' GTX 1060 scores above 13K.

www.techpowerup.com/223920/nvidia-geforce-gtx-1060-3dmark-firestrike-performance-revealed

Stock RX 480 8GB is slower in all three Fire Strike benchmarks:

www.3dmark.com/compare/fs/9202637/fs/9202684/fs/9202725/fs/9180075/fs/9181008#

Fire Strike Graphics Score: 12.951
Fire Strike Extreme Graphics Score: 6.063
Fire Strike Ultra Graphics Score: 2.902-2.911

 

[dark zero]

More GTX 1060 leaks.

http://videocardz.com/62086/nvidia-geforce-gtx-1060-rumors-part-5-full-specs-2-0-ghz-overclocking


Best Compubench Results

GTX 1060
https://compubench.com/device.jsp?benchmark=compu15d&os=Windows&api=cl&cpu-arch=x86&hwtype=dGPU&hwname=NVIDIA%20GeForce%20GTX%201060%206GB&did=36085769&D=NVIDIA%20GeForce%20GTX%201060%206GB

RX 480
https://compubench.com/device.jsp?b...Radeon+(TM)+RX+480+Graphics&testgroup=overall

Trading blows it seems. RX 480 wining in crypto and compute and GTX 1060 wining on gaming. Seems that both cards are great, but are purpoused in totally different areas.

 

[USER8000]

Right. If you think that's an OCed GTX 1060 and not simply 3DMark listing maximum GPU Boost 3.0 like it does with other Pascal cards, please explain to us how this 'stock' GTX 1060 scores above 13K.

www.techpowerup.com/223920/nvidia-geforce-gtx-1060-3dmark-firestrike-performance-revealed

Stock RX 480 8GB is slower in all three Fire Strike benchmarks:

www.3dmark.com/compare/fs/9202637/fs/9202684/fs/9202725/fs/9180075/fs/9181008#

Fire Strike Graphics Score: 12.951
Fire Strike Extreme Graphics Score: 6.063
Fire Strike Ultra Graphics Score: 2.902-2.911

Apart from the fact videocardz listed the maximum boost clock as 1.709GHZ?? That link shows the card running at 1.911GHZ and the clockspeed of the RX480 seems to be similar to what has been hinted for the AMD AIB RX480 cards. You don't seem to realise if it is an AIB RX480 it has a better cooler and a 8 pin power connector,meaning it will actually boost to the rated clockspeed and probably won't power throttle like the reference cards too. Plus look at the name of the submitter,Parm - that is one of the first names of the reviewers on Hexus(Parm Mann).

Edit to post.

Just because you have gotten overexcited and hyped the GTX1060 massively,you only have to blame yourself if the reality is not as clear cut as it seems.

You are suffering from the same hype train issue with all the people overhyping the RX480. Heck,people overhyped the GTX960 and it did the same(although I have one myself,and it does the job).

 

[Sweepr]

As the later I simply can't understand how you could be happy with anything less than a 5% improvement when Nvidia says the 1060 will be faster than the 980.

I'm just pointing that their definition of faster might not be the same as yours. 5-10% would be great but I don't see it happening in a broad spectrum of games. It might fare better against GM204 in newer titles though, perhaps that what they're quoting when they say it's faster.

 

[Sweepr]

Apart from the fact videocardz listed the maximum boost clock as 1.709GHZ??

Geforce GTX 1080's boost clock is 1733 MHz and 3DMark (sometimes) lists it as 1886 MHz due to GPU Boost 3.0, probably the same thing here. Note that only 1 GTX 1060 score from Parm showed 1911 MHz, all the others list the same 1886 MHz as its bigger brother.

http://cdn.videocardz.com/1/2016/05/NVIDIA-GeForce-GTX-1080-FireStrike-Ultra.jpg

All cards in this comparison are using stock settings. The variance for GTX 1080 clocks is between 1860 to 1886 MHz, probably caused by new GPU Boost 3.0 technology.

The maximum achievable 1,886 MHz, the GeForce GTX 1080 Founders Edition are more of a theoretical value: In games you never see this clock. There, the highest ever measured frequency was during the tests 1,835 MHz, but was also the clock after a few seconds history. In demanding titles 1.785 MHz has finally placed as the highest, realistic clock out.

And if the other cards are any indication, it's likely running below these speeds in the actual benchmark.

www.computerbase.de/2016-05/geforce-gtx-1080-test/6/#abschnitt_bis_zu_1785_mhz_takt_unter_dauerlast

Edit:

Just because you have gotten overexcited and hyped the GTX1060 massively,you only have to blame yourself if the reality is not as clear cut as it seems.

Learn how to differentiate stock vs OC for NVIDIA cards and then we'll talk.

 

[Keysplayr]

Just look at this

270X vs GTX770, thats what happen when you run out of VRAM, even at 1080P.

770 was 2GB card, no?

Same graph and 3GB 780 does fine.

 

[Shivansps]

770 was 2GB card, no?

Same graph and 3GB 780 does fine.

until 3GB runs intro the same problem as 2GB today.

 

[USER8000]

Learn how to differentiate stock vs OC for NVIDIA cards and then we'll talk.

Learn to not get so overexcited about the GTX1060 like the same people who got overexcited with the RX480 - hypetrain incoming. Also,I love who you are trying to teach somebody who actually has a GTX960 4GB about a 60 series part. The clocks I see in 3DMark are actually around the same I see in normal games,and they will tail off after a longer session. I had a GTX660 and a GTX660TI too. Early boost 1 use to overdo itself and throttle much easier than later iterations. However,the 60 series parts tend to do it far less. The GTX660 didn't suffer from the issue for example AFAIK. Also if it is Parm Mann from Hexus,they said they pre-heat the cards before benchmarks anyway,so it is unlikely to be a run from cold.

 

[Maverick177]

@Sweepr: Wait what? 1060 boost clock is 1k7 and you are telling me in Firestrike it boosts wayy over to 1911Mhz? What ?

 

[Sweepr]

Also if it is Parm Mann from Hexus,they said they pre-heat the cards before benchmarks anyway,so it is unlikely to a run from cold.

Pre-launch Geforce GTX 1080 FE's:

1860 MHz
1886 MHz

All cards in this comparison are using stock settings. The variance for GTX 1080 clocks is between 1860 to 1886 MHz, probably caused by new GPU Boost 3.0 technology.

Launch reviews confirm max theoretical clocks:

The maximum achievable 1,886 MHz, the GeForce GTX 1080 Founders Edition are more of a theoretical value: In games you never see this clock. There, the highest ever measured frequency was during the tests 1,835 MHz, but was also the clock after a few seconds history. In demanding titles 1.785 MHz has finally placed as the highest, realistic clock out.

Today GTX 1060 gets listed at 1886 MHz and 1911 MHz, curiously the exact same 25/26 MHz gap. This is GPU Boost 3.0 for you.

@Sweepr: Wait what? 1060 boost clock is 1k7 and you are telling me in Firestrike it boosts wayy over to 1911Mhz? What ?

Yes, I'm telling you max theoretical clocks listed by 3DMark go above NVIDIA specs, just like the Geforce GTX 1080.

Here's a leak from another GTX 1060 posted a few days ago, nearly same score:

 

[USER8000]

@Sweepr: Wait what? 1060 boost clock is 1k7 and you are telling me in Firestrike it boosts wayy over to 1911Mhz? What ?

No,he is saying it is only running at 1.7GHZ in that benchmark and not boosting past the reference max boost clocks since apparently Nvidia cards never boost consistently past the reference boost clocks. Funny how all my Kepler and Maxwell cards seem to have run consistently at much higher than the boost clocks when gaming or benching. But apparently he is the expert at all things Nvidia,so we will have to just sort of yield to him.

 

[Maverick177]

Pre-launch Geforce GTX 1080 FE's:

1860 MHz
1886 MHz



Launch reviews confirm max theoretical clocks:



Today GTX 1060 gets listed at 1886 MHz and 1911 MHz, curiously the exact same 26 MHz gap. This is GPU Boost 3.0 for you.

Yes, I'm telling you max theoretical clocks listed by 3DMark go above NVIDIA specs, just like the Geforce GTX 1080.

So what's the actual clock in Firestrike? Anyone has stock 1080 Firestrike restuls? Also, do you have a link to that Firestrike score you posted? I can do 17K in graphics score with GPUz showing my Fury@1Ghz. I don't trust screenshot.

 

[Sweepr]

No,he is saying it is only running at 1.7GHZ in that benchmark and not boosting past the reference max boost clocks

No, you just made this up.

since apparently Nvidia cards never boost consistently past the reference boost clocks.

They can, but still below max theoretical clocks.

The maximum achievable 1,886 MHz, the GeForce GTX 1080 Founders Edition are more of a theoretical value: In games you never see this clock. There, the highest ever measured frequency was during the tests 1,835 MHz, but was also the clock after a few seconds history. In demanding titles 1.785 MHz has finally placed as the highest, realistic clock out.

 

[Sweepr]

So what's the actual clock in Firestrike? Anyone has stock 1080 Firestrike restuls? Also, do you have a link to that Firestrike score you posted? I can do 17K in graphics score with GPUz showing my Fury@1Ghz. I don't trust screenshot.

We'll have to wait for launch reviews for that answer. Might as well be above 1.7 GHz but most likely below the listed clocks from 3DMark. My point is, the card is at stock here.

 

[USER8000]

snip

No,you are the one who is trying to say the cards never boost much past the 1.709GHZ clockspeed and the 1.911GHZ is not reached in real life. Yet,very single one of my Kepler and Maxwell cards have boosted much more than the max stated boost frequency - so I maybe I have special edition cards apparently and everybody knows Nvidia boost does start to go down over longer periods -so what??

This is the other issue,say it is an AIB RX480(as a worse case scenario for the RX480),and using your view it only boosts to 1.785GHZ,or 76MHZ above the base boost clock,so even expecting a linear clock to performance increase,the card would have to running at close to 1.95GHZ to match that RX480 at 1.35GHZ since the latter apparently has a 8.2% higher graphics score anyway.

But from my experience of having a AIB OC GTX960 myself,it consistently boosted much higher out of the box,ie, 140MHZ to 200MHZ past the reference boost clock. The GTX1060 listed at boosts to 1.9GHZ,which again seems more like an OC model since it is 200MHZ more than the official 1.7GHZ boost clock - also BTW,my card is only listed as having a 40MHZ to 50MHZ higher base clockspeed too.

But,you apparently think this is not an OC model,so we will need to agree to disagree and leave it at that then. I just hope your expectations are satisfied!!

 

[Sweepr]

No,you are the one who is trying to say the cards never boost much past the 1.709GHZ clockspeed and the 1.911GHZ is not reached in real life.

No. You're basically changing the goalposts again. Let's go back to the begining:

Yep,the GTX1060 boost clocks are meant to be 1.709GHZ and AIB AMD RX480 cards are meant to run around 1.325GHZ/1.35GHZ - that means an AIB overclocked GTX1060 is slower than an overclocked RX480.

Your overclock argument got proven wrong with examples of stock Geforce GTX 1080 FE being listed as a 1860-1886 MHz card due to GPU Boost 3.0, way above NVIDIA's boost clocks of 1733 MHz (without overclock). This doesn't mean they hit the mythical 1886 MHz in actual games (still above 1733 MHz, never said the opposite):

The maximum achievable 1,886 MHz, the GeForce GTX 1080 Founders Edition are more of a theoretical value: In games you never see this clock. There, the highest ever measured frequency was during the tests 1,835 MHz, but was also the clock after a few seconds history. In demanding titles 1.785 MHz has finally placed as the highest, realistic clock out.

This is the other issue,say it is an AIB RX480(as a worse case scenario for the RX480),and using your view it only boosts to 1.785GHZ,or 76MHZ above the base boost clock,so even expecting a linear clock to performance increase,the card would have to running at close to 1.95GHZ to match that RX480 at 1.35GHZ since the latter apparently has a 8.2% higher graphics score anyway.

We have a leak >2.0 GHz from today.

which again seems more like an OC model since it is 200MHZ more than the 1.7GHZ boost clock

No it doesn't. Only 26 MHz above what Geforce GTX 1080 was listed at, and that's only 1 result out of 5 (all the others show the same 1886 MHz).

I just hope your expectations are satisfied!!

I don't overhype so I'm sure they will be. Geforce GTX 980 performance at $249+ would be more than enough to give the closest competitor (RX 480 8GB) a run for its money given NVIDIA's strong brand. I'd probably still recommend RX 480 4GB over a GTX 1060 3GB though.

 

[Maverick177]

But from my experience of having a AIB OC GTX960 myself,it consistently boosted much higher out of the box,ie, 140MHZ to 200MHZ past the reference boost clock. The GTX1060 listed at boosts to 1.9GHZ,which again seems more like an OC model since it is 200MHZ more than the official 1.7GHZ boost clock - also BTW,my card is only listed as having a 40MHZ to 50MHZ higher base clockspeed too.

But,you apparently think this is not an OC model,so we will need to agree to disagree and leave it at that then. I just hope your expectations are satisfied!!

Yeah, I have an eVGA 960 SC, boost clock is around 1.2Ghz but the actual in game frequency is around high 1k3 almost to 1k4.

 

[boozzer]

until 3GB runs intro the same problem as 2GB today.

right? that is the entire point. with the precedent of 2gb cards, it is a 1000% no brainer the right choice is a 4gb or higher card, even for 1080p. if you want to keep your card for more than 1 year.

for anyone asking what to buy, and your advice is a 3gb card. that is sabotaging. plain and simple. D:

unless your advice comes with an advisory that the person asking help might need to buy a new one in 1 year or less.

 

[Kenmitch]

We'll have to wait for launch reviews for that answer. Might as well be above 1.7 GHz but most likely below the listed clocks from 3DMark. My point is, the card is at stock here.

You know how nvidia boost works. Using the stated boost clock for comparisons like fps/watt or any others is dishonest at minimum. One must monitor them if they really want to know what it really is.

Using my Zotac 1070 Amp as an example. Stated boost clock is 1797 I believe. Stock settings it will boost to 1999MHz at times. I've never seen it revert to 1797 yet. Only had it a couple days so fare and haven't really tested it thoroughly yet.

 

[JustMe21]

The Legit Review #s are from an i7-4960X @ 3.6 GHz. If you want to do a comparison with the RX 480, go with the same i7-6700K like in the leaked benchmark results. The leak's results could still be off because it doesn't show what the clock speeds are.

This RX 480 result has a score of 10,911 with the RX 480 is running at stock speed and the processor running at stock speeds as well.
http://www.3dmark.com/fs/9106797


The link to the comparison that Sweepr posted shows the max clock at 4400 Mhz.

The 2 comparable speed RX 480 stock scores are one with a max clock speed of 4237 MHz, with a score of 11,005
http://www.3dmark.com/fs/9187493

and one with a max clock speed of 4497 MHz, with a score of 11,066.
http://www.3dmark.com/fs/9187493

Interesting how this GTX 980 gets lower scores than the 1060, but still has the same overall score. You gotta love the varied results that you get from benchmarks.
http://www.3dmark.com/fs/6776500

I am interested to see what the new 3DMark Time Spy DX 12 benchmark results will look like. Does anyone know when that will be out?