GTX 1060

Orangey · 17 Jul 2016 at 15:43

The 480 needs those extra phases to deliver the extra power to its redlined failure of a chip lol. Perhaps you might have noticed it doesn't OC at all. The 1060 will OC quite nicely I'm sure, even with "only" half the phases. Because quality.

Griffildur · 17 Jul 2016 at 16:11

RanxZy said:
Vulkan has some NVIDIA cards that were once neck and neck with AMD equivalents (280X vs GTX 770, 290X vs 780Ti) being left in the dust. If the Fury X was even stevens in terms of perormance with the 980Ti, Vulkan would give it a clear lead.

once?
if ?
would?

Scougar · 17 Jul 2016 at 16:11

Orangey said:
The 480 needs those extra phases to deliver the extra power to its redlined failure of a chip lol. Perhaps you might have noticed it doesn't OC at all. The 1060 will OC quite nicely I'm sure, even with "only" half the phases. Because quality.

Please submit evidence of 'at all'.

(p.s. I am actually completely with you on the 1060 doesn't need the extra phases, otherwise they would have been on there).

Orangey · 17 Jul 2016 at 16:42

Dunno if you missed it but all the evidence points to the chip being underperformant on the new process and thus pushed way beyond where the clocks and volts should sit naturally (which is what Kyle from [H]ardOCP was talking about pre-launch) in order to meet some kind of semi-acceptable level but discarding the previous ppw claims. This is borne out by AIB models also having disappointing clocks and LN2 barely taking it to 1500.

RavenXXX2 · 17 Jul 2016 at 16:56

480 was designed for the PS neo, AMD just pushed the clocks to max and dumped it hastily on the PC market. It's running way out of spec for what it is designed for hence the high power draw, high temps and poor OC.

Meaker · 17 Jul 2016 at 16:57

Orangey said:
Dunno if you missed it but all the evidence points to the chip being underperformant on the new process and thus pushed way beyond where the clocks and volts should sit naturally (which is what Kyle from [H]ardOCP was talking about pre-launch) in order to meet some kind of semi-acceptable level but discarding the previous ppw claims. This is borne out by AIB models also having disappointing clocks and LN2 barely taking it to 1500.

Without knowing the target frequency for the process that's a hard conclusion to reach.

The 1080 is not reacting hugely well to LN2 either past the max air clocks.

In the end the actual frequency matters little. The performance per £ and what extra performance you can get from overclocking matter more.

Mr Latte · 17 Jul 2016 at 17:29

humbug said:
Look at the difference in power phase, the RX 480 is far better quality.

Its the 1060 that does look positively cheap.

It also has room for 8GB but only uses 6GB

Clearly Nvidia wanting to maximise profits but to be fair the outer casing and appearance of the 1060 to many folk does look more expensive/fancier than the 480.

Mauller · 17 Jul 2016 at 17:44

Orangey said:
Dunno if you missed it but all the evidence points to the chip being underperformant on the new process and thus pushed way beyond where the clocks and volts should sit naturally (which is what Kyle from [H]ardOCP was talking about pre-launch) in order to meet some kind of semi-acceptable level but discarding the previous ppw claims. This is borne out by AIB models also having disappointing clocks and LN2 barely taking it to 1500.

Hyperbole much?

The frequency a part can attain is a larger part to do with its design and partly to do with the node.

chaparral · 17 Jul 2016 at 18:23

Mr Latte said:
It also has room for 8GB but only uses 6GB

I think that because it only has a 192-bit Memory Interface and can't have 8GB ...

But am most likely wrong about this as I know crap about how a GPU is made..

Scougar · 17 Jul 2016 at 20:13

Orangey said:
Dunno if you missed it but all the evidence points to the chip being underperformant on the new process and thus pushed way beyond where the clocks and volts should sit naturally (which is what Kyle from [H]ardOCP was talking about pre-launch) in order to meet some kind of semi-acceptable level but discarding the previous ppw claims. This is borne out by AIB models also having disappointing clocks and LN2 barely taking it to 1500.

I wouldn't be surprised in the slightest if it were true. But saying it isn't able to overclock at all is a bit extreme. It seems to clock quite well from what I see, held back by bizarre design decisions from AMD. (Disclaimer: I am not an engineer, they are obviously wayyyy smarter than me!)

Orangey · 18 Jul 2016 at 00:29

Quite well? It's less than 10%. To be considered a good OCer a card would need something like 30% OC at least, like the 5850.

If they had a high-performance process to make the chips on the power would not swing so wildly with so little frequency adjustment.

humbug · 18 Jul 2016 at 00:39

Orangey said:
Quite well? It's less than 10%. To be considered a good OCer a card would need something like 30% OC at least, like the 5850.

If they had a high-performance process to make the chips on the power would not swing so wildly with so little frequency adjustment.

30% at least? are you trolling? the best Maxwell cards overclock by 30%.

mark-san · 18 Jul 2016 at 01:10

Orangey said:
Quite well? It's less than 10%. To be considered a good OCer a card would need something like 30% OC at least, like the 5850.

If they had a high-performance process to make the chips on the power would not swing so wildly with so little frequency adjustment.

I hope your former maths teacher isn't reading this, otherwise you might get to be the next one using the donkey hat...

1120 (base clock) -> 1350 (usually the maximum stable clock with reference model) is approximately a 20.5% increase.

Scougar · 18 Jul 2016 at 01:37

I think it is better to take the boost clock rather than the base clock given that boost will be being used nearly all the time.

revanchrist · 18 Jul 2016 at 03:17

Orangey said:
Dunno if you missed it but all the evidence points to the chip being underperformant on the new process and thus pushed way beyond where the clocks and volts should sit naturally (which is what Kyle from [H]ardOCP was talking about pre-launch) in order to meet some kind of semi-acceptable level but discarding the previous ppw claims. This is borne out by AIB models also having disappointing clocks and LN2 barely taking it to 1500.

LMAO. Der8auer overclocked the ref 480 to 1500 on water and 1700 on LN2. Enio overclocked a ref 480 to 1520 on water. It is rumored that Powercolor Devil is 1400 boost while Sapphire Nitro is 1390 boost.

RanxZy · 18 Jul 2016 at 06:21

Griffildur said:
look at it from another point of view, when you are on OpenGL NVidia cards do a lot a better. On Vulkan, which is supposed to still be heavily optimised for AMD cards, Nvidia ones still do better.

AlamoX said:

Griffildur said:
once?
if ?
would?

You could have just had the humility to admit your opinion was clearly wrong but "once, if, would" is a much more competent reply I guess.

mahius · 18 Jul 2016 at 07:45

humbug said:
Look at the difference in power phase, the RX 480 is far better quality.

Its the 1060 that does look positively cheap.

And the funny thing is... the 480 will be the actual cheaper card to buy, with more RAM to boot too. Nvidia's 'quality' conmponents in the founder's edition and all that. King of selling us less for more this gen.

Can't wait to see the inevitable moaning and groaning tomorrow come the actual release of the 1060. I've got the day off for it too. Place your bets folks: Asus Strix/MSI Gaming 1060 at 980 perf for over £300 and the same terrible overclocking as the bigger 1000 cards is my guess. Oh, and of course the reference version will have bad thermals like the rest.

Nvidia could stick a thorn in AMD's foot if they wanted to just by pricing the 1060 right... but we all know they won't. I miss the Maxwell days already... (though we have nice deals on Maxwell cards that make them better buys than current gen cards lol).

Kaapstad · 18 Jul 2016 at 09:31

Will there be non reference 1060s available on launch day ?

Gibbo · 18 Jul 2016 at 10:05

Kaapstad said:
Will there be non reference 1060s available on launch day ?

Yes, lots of them in stock right now, 14:00 tomorrow you can see and buy them.

madmaca · 18 Jul 2016 at 10:18

Gibbo said:
Yes, lots of them in stock right now, 14:00 tomorrow you can see and buy them.

Woohoo!! Hoping for some keen pricing