[VC] NVIDIA GeForce GTX 1060 Specifications Leaked, Faster than RX 480

[Bacon1]

Well, I think I've only seen these pics:

http://videocardz.com/61868/nvidia-geforce-gtx-1060-rumors-part-3-more-pictures

http://videocardz.com/61826/nvidia-geforce-gtx-1060-rumors-part-2

The 6 pin power connector placement, or lack thereof, is odd.

Where is the 6 pin lacking in those pictures?

 

[William Gaatjes]

Where is the 6 pin lacking in those pictures?

It is not lacking. But look at where the pcb seems to end and where the 6 pin connector is positioned.

 

[LTC8K6]

Where is the 6 pin lacking in those pictures?

I don't know.

There is the one pic of the back of the board which does not show where the 6 pin socket would/should be connected.

It seems unlikely that the connecting wires that go over to the 6 pin socket would be surface mounted. They would normally be "through hole" and largish for the high power. We should see the solder points for the wires, but we don't?

I guess one could interpret that a few ways.

Mostly, all we can say is that we aren't sure what's going on with the power connector.

 

[William Gaatjes]

Here is a GTX1080 :

and a GTX 1070, both for comparison.

and a GTX 1060

 

[LTC8K6]

There are 6 big solder spots near the middle of the back of the board. Slightly towards the front.

http://cdn.videocardz.com/1/2016/07/nvidia_geforce_gtx_1060_purepc_2-900x437.jpg

Perhaps the power wires go that far?

 

[LTC8K6]

Here is a GTX1080 :

and a GTX 1070, both for comparison.

and a GTX 1060

Yes, the 1080 is a full length board with the power connector mounted at the back of the board.

The 1060 is a short board with the power connector apparently mounted off of the board, behind the blower.

 

[William Gaatjes]

It would be odd if the GTX1060 has such a long pcb. And the pcb does seem shorter. So , the question is how is the 6 pin pcei pcb connector connected ?
There is molex minifit jr wire to wire plug, but that would make the gtx1060 needlessly expensive and labour intensive to make.

Or pictures are fake.

 

[MrTeal]

If I had to hazard a guess, I'd say it's either the two big holes at the end with the wires joined together earlier, but it could be surface mount or the 6 holes above the PCIe notch. The whole thing is going to be a PITA for water cooling though.

 

[IEC]

It would be odd if the GTX1060 has such a long pcb. And the pcb does seem shorter. So , the question is how is the 6 pin pcei pcb connector connected ?
There is molex minifit jr wire to wire plug, but that would make the gtx1060 needlessly expensive and labour intensive to make.

Or pictures are fake.

That certainly does seem odd. You'd also be running a not insignificant amount of current over a longer distance + adapter for no reason as well. What engineer would do that?

 

[William Gaatjes]

There are 6 big solder spots near the middle of the back of the board. Slightly towards the front.

http://cdn.videocardz.com/1/2016/07/nvidia_geforce_gtx_1060_purepc_2-900x437.jpg

Perhaps the power wires go that far?

Nah, there is also a heatsink present, the wires would be intruding the same space. Seems to me that is.

It could also be that they have a thin riser pcb that connects to the top left of the main pcb and has on one end a 6 pin pcie-power connector and on the other end a pcb to pcb connector that can handle the current.

Great catch with the pictures : LTC8K6

 

[sm625]

There are 6 big solder spots near the middle of the back of the board. Slightly towards the front.

http://cdn.videocardz.com/1/2016/07/nvidia_geforce_gtx_1060_purepc_2-900x437.jpg

Perhaps the power wires go that far?

I think those are thru-hole inductors. Here is a picture of a card that shows what I mean:

Note: This is NOT a 1060!

I have circled in red what appears to be a large single conductor or wire or post or something. It could be that this big one and the other big one are grounds, and the 4 small holes are for +12V that has been fed through some sort of adapter as William Gaatjes said. It does kind of make sense since soldering wires is not manufacturing-process friendly. You want to be able to just plug the connector in during assembly. And you also want it to line up really easy despite the shroud blocking the vision. But it is pretty dirty to bring on a totally new type of header for this.

 

[.vodka]

There was nothing wrong with leaving the 6 pin plug at the edge of the PCB as usual, apart from it looking ugly. A 1060 isn't something that you're gonna be showing off in a build, so it baffles me even more seeing what they've done with the power connector.

That, and the lack of a SLI connector... with DX12 bringing benefits to multiGPU setups and the news of MS' doing some groundwork for the feature to be more easily implemented... it's another weird move.

Unless nV has developed something like AMD's XDMA crossfire that uses the PCIe bus instead of a separate connector and are rolling the feature out in the 1060. Haven't heard any rumors of the sort so I'm probably mistaken and just trying to rationalize skipping the SLI connector.

 

[Silverforce11]

@.vodka

NV's probably done their homework and find that the % of gamers who SLI mainstream GPUs is next to zero. So it's a pointless and costly move to support it via drivers for the next few years.

They did the maths for 3 and 4 way SLI and found the same conclusion.

It's actually the correct move. They need to stop wasting time on supporting such minority setups and focus more on single cards, extending it to previous generation architectures too. ie. Keep the focus on Pascal but also Maxwell for the next few years to be seen like they are NOT gimping older flagships.

 

[guskline]

If I had to hazard a guess, I'd say it's either the two big holes at the end with the wires joined together earlier, but it could be surface mount or the 6 holes above the PCIe notch. The whole thing is going to be a PITA for water cooling though.

I really doubt many will water cool the GTX 1060.

 

[Insomniator]

If its really $250 I may just sell my R9 390 even if its mostly a side grade. 390 is a furnace in my mATX case, can feel the heat coming out of the system.

 

[guskline]

@.vodka

NV's probably done their homework and find that the % of gamers who SLI mainstream GPUs is next to zero. So it's a pointless and costly move to support it via drivers for the next few years.

They did the maths for 3 and 4 way SLI and found the same conclusion.

It's actually the correct move. They need to stop wasting time on supporting such minority setups and focus more on single cards, extending it to previous generation architectures too. ie. Keep the focus on Pascal but also Maxwell for the next few years to be seen like they are NOT gimping older flagships.

SilverForce, I made the decision to go high end single cards and drop CF/SLI.
I applaud AMD for improving some of the profiles for CF support and likewise Nvidia. However, things just seem so much smoother running the gpus I have below.

 

[sze5003]

If its really $250 I may just sell my R9 390 even if its mostly a side grade. 390 is a furnace in my mATX case, can feel the heat coming out of the system.

We will see prices tomorrow. I think the 3gb would be that much at least.

 

[MrTeal]

I really doubt many will water cool the GTX 1060.

That's because people are lazy.

 

[LTC8K6]

I have circled in red what appears to be a large single conductor or wire or post or something. It could be that this big one and the other big one are grounds, and the 4 small holes are for +12V that has been fed through some sort of adapter as William Gaatjes said. It does kind of make sense since soldering wires is not manufacturing-process friendly. You want to be able to just plug the connector in during assembly. And you also want it to line up really easy despite the shroud blocking the vision. But it is pretty dirty to bring on a totally new type of header for this.

The 4 little solder points make me think of a fan connector.

 

[Bacon1]

Also remember that DX12 MGPU doesn't require bridges, hence using Fury X + 980 TI. So "SLI"/"CFX" will be obsolete with DX12.

 

[BSim500]

By the time the 960 came out there were very quiet 290s, namely the XFX DD and the Sapphire Tri X, so at the time that argument was just another smokescreen...

Have you used the Tri-X before? The only way to describe it is quiet.

"Quiet", "Very quiet", "Silent", etc, are entirely subjective. TPU reviews show the tri-fan Sapphire 290 still 37-41db (vs 49db reference). The MSI 960 however was way down at 27db (and 13c cooler) at full load, and 0db for light games / idle / HTPC playback. Same test equipment, methodology and noise floor. Not to mention the obvious secondary effects +170w of extra heat in low airflow cases (800-1500rpm vs 600-750rpm case fans, or a semi-passive PSU being pushed over a 200w trigger point) which can easily add another 5-10db of noise, yet is never measured as a whole system when benchmarking individual components (and thus never shows up on such charts, especially open air test beds).

The "R9 290 at fire sale prices" was certainly a better buy vs the GTX 960 if all you cared about was raw performance & perf-per-$, but to be honest, I think some people here with an increasingly unhealthy anti-960 personal daily obsession are "over-talking" their point when they start advocating 275w 40db cards for sub 20db light gaming HTPC's on the back of "because the rest of my rig has a 43db noise floor, I declare my 40db 290 is as completely silent as 0-27db cards, and anyone who says otherwise is erecting a smokescreen". It really doesn't work like that. TPU's measurements agree, not to mention the laws of physics don't care about brand warfare either, hence the higher objectively measurable temps and noise on the higher wattage card.

I hope we don't see another situation where the inferior price/perf card is recommended on incredibly flimsy grounds to smoke screen for brand loyalty with the 1060 like we did with the 960/290.

Personally I'm looking forward to the RX 460/470's. Why? Because if they're any good it will both provide healthy competition for nVidia in the low/lower-mid ranges (750/950/960), but most importantly it will finally force +250w space heater advocates to start offering something that actually competes "like for like" vs 90-120w nVidia cards for lower end gaming beyond single metric price dumping and "brand fanboy" accusations vs the same non poverty stricken consumers who are also quite happy to buy a Platinum PSU over a Bronze, a 65w i5/i7 over a 220w FX-9590, etc, and really don't care what labels a few "Forum Warriors" stick on them...

Here's a totally outrageous suggestion to those filling up 135 page threads day after day with this "why can't everyone be exactly like me" community 'analysis' stuff:-

- Those who want a GTX 1060 and are happy to pay whatever premium should just buy one, without needing to talk down any decent AMD offering (RX 480 really isn't a bad card) nor feel any "need" to justify their purchase decision to any Red Team Forum Warrior.

- Those who want an RX 480 should also just buy one without constantly self-inflicting daily panic attacks over watching someone else "daring" to spend even $1 more on another dGPU than their own personal budget permits, or feel any "need" to justify their purchase decision to any Green Team Forum Warrior.

These boards would be a damn site healthier if people cut back on the "whine" and stopped acting like that last $10, $20, $30, etc, difference is like curing cancer vs making 1,000 disabled orphans starve (an attitude wildly out of proportion to any other PC component, eg, no-one bats an eyelid over $60 vs $25 gaming mice, $60-$120 water vs $30 air cooling, +$70 on color co-ordinated braided PSU cables & lighting, $320 950 PRO vs $160 850 EVO, 32 vs 16GB RAM, etc). Not just spend page after page raging against hardware they don't own (and had zero intention of buying) whilst moaning that they never have enough time to clear that Steam backlog because they're too busy arguing to ever get around to actually using what they're arguing over to play any games...

^ Comment isn't aimed at anyone specifically, but it will certainly "resonate" with some more than others recently who expend an extraordinary amount of energy 'not buying something'...

 

[sze5003]

So what time are they having this release event?

 

[Headfoot]

"Quiet", "Very quiet", "Silent", etc, are entirely subjective. TPU reviews show the tri-fan Sapphire 290 still 37-41db (vs 49db reference). The MSI 960 however was way down at 27db (and 13c cooler) at full load, and 0db for light games / idle / HTPC playback. Same test equipment, methodology and noise floor. Not to mention the obvious secondary effects +170w of extra heat in low airflow cases (800-1500rpm vs 600-750rpm case fans, or a semi-passive PSU being pushed over a 200w trigger point) which can easily add another 5-10db of noise, yet is never measured as a whole system when benchmarking individual components (and thus never shows up on such charts, especially open air test beds).

The "R9 290 at fire sale prices" was certainly a better buy vs the GTX 960 if all you cared about was raw performance & perf-per-$, but to be honest, I think some people here with an increasingly unhealthy anti-960 personal daily obsession are "over-talking" their point when they start advocating 275w 40db cards for sub 20db light gaming HTPC's on the back of "because the rest of my rig has a 43db noise floor, I declare my 40db 290 is as completely silent as 0-27db cards, and anyone who says otherwise is erecting a smokescreen". It really doesn't work like that. TPU's measurements agree, not to mention the laws of physics don't care about brand warfare either, hence the higher objectively measurable temps and noise on the higher wattage card.

That test measured 37 db.

I agree you must re-measure your noise floor after adding a new component, if your case air flow isnt enough to handle it you might increase your noise floor from other components. It also might not. So your point is you have to measure the noise floor under proper conditions, which I agree, but if you're under the noise floor you're under the noise floor. That's the definition of a noise floor.

But the real fact of the matter is that wattage never mattered until nVidia was better at it. This is not coincidental. For every 1 genuine commenter that cared about power usage forever, there are 20 fanboys that use it as a smoke screen talking point. Its intentional deception and anyone who values the truth and honesty should come out against the blatant dishonesty being perpetuated against new posters who come to this forum for unbiased advice.

This is an enthusiast forum where performance matters. 50% more performance is a massive difference. It's the difference between having to upgrade in a year and being fine for multiple years. More performance uses more power, this is basic and well known.

The 960 was only ever a good card for HTPC when the 950 didnt exist. The 950 was the good HTPC card. It had all the benefits of the 960 but it was cheaper and used even less power. The only downside was less gaming performance which doesnt matter if its just an HTPC card.

The "space heater" garbage is exactly that. Emotional appeals that are totally meaningless. The extra electricity costs me a few bucks a year. People have to resort to weasel words like that when they have no real, factual case to stand on. The cost of electricity on video cards is absolutely negligible. I agree with you that we shouldn't be making massive stinks about $10. That little money just doesn't matter. If you want to reduce your electricity usage there are MANY better and more effective ways to do it. If you sit down and calculate how much energy you would save just by air drying your clothes on a clothesline in the summer, it will dwarf the difference in videocard power in the same performance class. It's a mountain out of a molehill that's been done intentionally as pure marketing and brand loyalty. I'm not accusing you personally of any of this to be clear -- but the people who do this certainly exist. Power use didnt matter when the 9700 came out. It didn't matter when Fermi came out, even though fanboys turned that molehill into a mountain. And then the team of fanboys switched on 28nm with the other side making a mountain out of a molehill. AIB versions of all of those cards solved the noise problem.

NOTE: None of this applies to the people who have always built silent PCs (SILENT, not just low noise). Obviously a totally-silent passive PC requires different concerns where you do actually need to minimize heat use and you have to make significant tradeoffs to get there. This is an infinitesimally small portion of the market.

This time around both vendor's have very solid HTPC features so I agree that the 460/470 will be very interesting for that use case. Likely the 1060 will be too expensive as a pure HTPC card like the 960 was, but the 1050 might get there like the 950 did.

 

[Headfoot]

Just to be clear - the issue (in my mind, and I assume in most other similar minded folks minds) is not people choosing one way or the other. If someone accidentally bought a $200 960 4gb without understanding it was a bad buy in light of a $250 290, I don't really care at all.

The problem was that certain people would lie, tell half truths, and generally be dishonest in their presentation of the facts in order to dissuade less knowledgeable people seeking advice from making an objectively better choice on the most important metrics to most buyers: price/perf, absolute perf, absolute price. The dishonesty bothers me. Not the products themselves. And I predict the same thing will happen with the 1060 and surrounding cards (though obviously the facts are different this time around, I doubt the dishonesty will be)

 

[96Firebird]

Do you have an example of this dishonesty and lies?