Nvidia Pascal Reviews [1080XP, 1080ti, 1080, 1070ti, 1070, 1060, 1050, and 1030]

[Deleted member 2197]

Time Spy review ...
http://www.hardware.fr/articles/952-9/benchmark-3dmark-fire-strike-time-spy.html

 

[silent_guy]

Pure baseless and useless speculation until you are able to show any kind of solid proof of that.

- How much is the price-per-mm^2 on GF's 14FF and TSMC's 16FF?
- What exactly are the yields for P10 and GP104?
- How much is the cooling solution for either card? How much is the PCB for either card?
- How much is each 8Gbps DRAM module being sold for each OEM?
- Where's the official BoM + assembly cost for each card? Reference or OEM?

You've read my earlier posts.

I left the contracts between Nvidia/TSMC and AMD/GF at home this morning, but one has to make a lot of mental gymnastics to make AMD come ahead on this one.

I mean, are you seriously questioning that low volume player AMD would get a better deal on 8 GDDR5 chips than Nvidia would get on 6 of the same? That's one for the books...

This is that "HBM costs more than an entire building" argument again. Please, spare us of that speculative crap that tries to pass as fact.

That fact that it doesn't fit your AMD colored worldview doesn't make it crap. It's once again very logical when you consider volume, competition, and technical complexity. You're one of the last hold-overs, to be honest. With the introduction of the 480, most AMD fans have come to the conclusion as well.

 

[xEx]

what really surprise me is the difference in units, 1280 vs 2300; 80 vs 144 and still manage to be faster in many games.

 

[DavidGraham]

That is what I am saying in general, but some tests I have seen with the latest DX12/async patch of RoTR actually has the Fury X now about 3-5% faster than the 980 ti where before it trailed ever so slightly.
However this is not reflected with the 1060 vs 480 where it seems the 480 is more constrained.

It was one test, (it was doubtful because 980Ti lead was small in it to begin with). Now with all these updated Tomb Raider DX12 scores with the 1060 review, the situation is the same as before, even Maxwell cards still have a large lead over Fiji ones.

 

[Deleted member 13524]

The only GTX 1060 review I saw with the async compute patch mentioned for RoTR is the one from TechSpot:

However, the developer just recently released a DX12 update that adds async compute support which is said to boost performance on AMD hardware.

With the current version of the game, the RX 480 has 16% slower average, 10% faster minimum framerates.
Hardly what I would call "a large lead". Especially for a Gameworks title.

what really surprise me is the difference in units, 1280 vs 2300; 80 vs 144 and still manage to be faster in many games.

It's 1280 * 1700MHz on the GTX 1060 and 2304 * 1266MHz on the RX 480. In practice, the difference in theoretical compute performance is around 30%.
On the other hand, the GTX 1060 has 48 ROPs at 1700MHz vs. 32 ROPs at 1266MHz on the RX 480, meaning the GTX 1060 has about 100% more fillrate. It also has 10 geometry units (one per SM) whereas RX 480 has the 4 units general limit of GCN.

Depending on the game, both GPUs are bound to find advantages between one another.

 

[DavidGraham]

You've got DigitalFoundry as well, hell even a GTX 970 is faster than RX 480:

We've also re-benched Rise of the Tomb Raider on DirectX 12, which was recently kitted out with a new upgrade that supports asynchronous compute. This adds a small boost to overall AMD performance, but doesn't amount to any game-changing difference - Nvidia still has a strong lead here.

http://www.eurogamer.net/articles/digitalfoundry-2016-nvidia-geforce-gtx-1060-review

With the current version of the game, the RX 480 has 16% slower average, 10% faster minimum framerates.
Hardly what I would call "a large lead". Especially for a Gameworks title.

Same for Hitman and Ashes, AMD evolved titles, so should we count them out as well?

 

[CSI PC]

I think Nvidia will want Ashes to be included considering how close the 1060 is in nearly every review done so far at 1080p
Even 1440p it is within 3% depending upon review.
I would say Ashes performance is a bit disappointing from the 480.
Cheers

 

[Deleted member 13524]

On thing that might be important to mention is the GTX 1060's score in Steam's VR Test, which is really high at 7.8.
Either this is a direct result of SMP or just driver tweaks, it's a very good score. I don't know how Steam's VR test translates into real-world VR scenarios, though.

Same for Hitman and Ashes, AMD evolved titles, so should we count them out as well?

No.
Especially because if you count out every Gaming Evolved and Gameworks/TWIMTBP titles then for new APIs then all you'd have for comparison is Doom with the RX480 wiping the floor with the GTX 1060 even to the point of showing better performance-per-watt.
And you wouldn't want that, would you?

 

[DavidGraham]

No.
Especially because if you count out every Gaming Evolved and Gameworks/TWIMTBP titles then for new APIs then all you'd have for comparison is Doom with the RX480 wiping the floor with the GTX 1060 even to the point of showing better performance-per-watt.
And you wouldn't want that, would you?

Even with that overblown behavior, you would still have the 1060 wiping the floor of EVERY AMD card (FuryX included) @1080p in Talos Principle Vulcan mode:
http://www.gamersnexus.net/hwreviews/2518-nvidia-gtx-1060-review-and-benchmark-vs-rx-480/page-4

 

[Deleted member 13524]

the 1060 wiping the floor of EVERY AMD card (FuryX included) @1080p in Talos Principle Vulcan mode

You don't really know what "wiping the floor" means, do you?

It certainly doesn't mean a 6% difference.


A benchmark where the new API breaks performance on every single Vulkan-supporting card on the planet is so irrelevant it's no wonder I had never heard of it.
Kudos to Croteam for the effort in building the proof-of-concept though. I love their Serious Sam titles.

 

[homerdog]

1060 is a damn fine card, just as I expected.

As for the BOM for the 480 vs 1060, the 1060 is way cheaper for NV to make, there's no need for any mental gymnastics to figure that out.

I think the 4GB 480 at $200 is still a winner though so long as it runs 8Gbps memory (I think they generally do right?). Especially if it can be "unlocked" to 8GB. A lot of gamers can be convinced to stretch their budgets to $200, $250 maybe not so much. But I'm afraid AMD makes very little money on that card so it may not stick around for long

The moral of the story is those of us with GTX970 or R9 290/390 should be fine until the next round of cards

 

[Razor1]

Looks like many AIB 1060 stock cards are selling for $250 unlike the 1070 and 1080 launches MSRP looks to be being followed. As I was thinking before, if AIB partners or retailers put this card for any more, it just wouldn't be a good buy. (even the Mini cards are going for 250 too)

 

[Razor1]

http://videocardz.com/62364/nvidia-geforce-gtx-1060-reviews

list of reviews for the 1060

 

[DavidGraham]

You don't really know what "wiping the floor" means, do you?
It certainly doesn't mean a 6% difference.

It does when it's coming from a mid range card.

A benchmark where the new API breaks performance on every single Vulkan-supporting card on the planet is so irrelevant it's no wonder I had never heard of it.

Oh come on, be a sport and accept this facade for what it is.

The fact of the matter is: Right now, it is all crap, this whole new low level APIs are all BS, whether in doom or Talos, you just like to pick the one that suits your tainted glasses. Low level APIs shouldn't result in degradation without IQ increase, they should increase fps through close to the metal access on all hardware, not just GCN, because guess what? Every hardware can increase it's throughput if you code to it's metal.

it's no wonder I had never heard of it.

This game was tested by every major media outlet on the Internet as the first Vulkan game ever, and it did exactly the same. Break fps on all hardware. So much for gaining performance through low level code.

 

[Ext3h]

Looks like many AIB 1060 stock cards are selling for $250 unlike the 1070 and 1080 launches MSRP looks to be being followed. As I was thinking before, if AIB partners or retailers put this card for any more, it just wouldn't be a good buy. (even the Mini cards are going for 250 too)

So far, yes.

But then again, e.g. here in Germany, apart from the "Palit GeForce GTX 1060 Dual", essentially all custom models priced at MSRP are already sold out with no known resupply dates. For retailers (like mindfactory.de) which are publishing a rough estimate of sales, combined stocks of less than 200 cards across all models. For comparison, even the initial stock on the GTX 1080 was 5 times higher, at least for that specific retailer.

Guess we will have another look at the "MSRP" again in one or two weeks, see if it still holds.

 

[homerdog]

You don't really know what "wiping the floor" means, do you?

It certainly doesn't mean a 6% difference.

A benchmark where the new API breaks performance on every single Vulkan-supporting card on the planet is so irrelevant it's no wonder I had never heard of it.
Kudos to Croteam for the effort in building the proof-of-concept though. I love their Serious Sam titles.

I agree it's bollocks that Vulkan is slower than DX11 in that game (this should never happen for any card). Disregarding that, "wiping the floor" in this context is appropriate since even the midrange 1060 beats the FuryX.

I wouldn't put too much stock in any of the current DX12/Vulkan benchmarks. It's like this with every new API. Need to wait about a year to let everyone get their shit together.

 

[Deleted member 2197]

Palit GeForce GTX 1060 Super Jetstream Review
http://www.guru3d.com/articles-pages/palit-geforce-gtx-1060-super-jetstream-review,1.html

GTX 1060 FCAT Analysis
http://www.guru3d.com/articles-pages/geforce-gtx-1060-fcat-frametime-analysis-review,1.html

 

[ninelven]

Hmm... looks like it will be a tough choice. But apparently I will have plenty of time to mull it over...

 

[CSI PC]

So far, yes.

But then again, e.g. here in Germany, apart from the "Palit GeForce GTX 1060 Dual", essentially all custom models priced at MSRP are already sold out with no known resupply dates. For retailers (like mindfactory.de) which are publishing a rough estimate of sales, combined stocks of less than 200 cards across all models. For comparison, even the initial stock on the GTX 1080 was 5 times higher, at least for that specific retailer.

Guess we will have another look at the "MSRP" again in one or two weeks, see if it still holds.

We are more fortunate here in the UK, still quite a few main online retailers with custom 1060.
The cheapest custom models not out just yet such as the Asus Turbo/EVGA and MSI basic models, but they are only £12 cheaper than the current cheapest custom still in stock.
This puts the current custom 1080 models pretty competitive (does cover a broad range) to the 480 custom AIB when they finally release here in the UK.
Cheers

 

[gamervivek]

I agree it's bollocks that Vulkan is slower than DX11 in that game (this should never happen for any card). Disregarding that, "wiping the floor" in this context is appropriate since even the midrange 1060 beats the FuryX.

I wouldn't put too much stock in any of the current DX12/Vulkan benchmarks. It's like this with every new API. Need to wait about a year to let everyone get their shit together.

Actually it's inappropriate since his comment made it look like that poor gtx1060 was being made to clean floors of not only one or two but every AMD card.

As for Talos Principle Vulkan, the benchmarks were released back in February and AMD still saw pretty good increases compared to OpenGL though regression compared to dx11.