Keeping an eagle eye out for preorders from the usual suspects of online e-tailers. As well as Geforce.com website. Preorder starts May 20th, which is today in the States.

Let the scalping begin?

Is there a text version of that video review? I'm too "distracted" by the dialect to take him seriously. Plus, he's collating from reviews and making his own review out of those reviews? Reviewtiful!

This post has been edited by stringfellow: May 21 2016, 06:36 AM

I see his gripe as being more on the fact that the 1080 isn't THAT much more powerful than a heavily overclocked 980Ti and when overclocked, the 1080 throttles down heavily to the point that those who have heavily overclocked 980Ti should keep it because the difference is minimal.

I fall into that category, and yet I'm still excited about the 1080 because of what it can do at stock versus my mildly overclocked 980Ti. But my case is niche, a lot of the excitement from those who have held back from upgrading 2-3 generations back and are now prime for a replacement, and for someone who has been stuck with a 680 and 780/Ti, the 1080 is a valid enough purchase to get DESPITE this AdoredTV's claim that its overclocking capability is borked, according to him. The price/performance for this compared to Maxwell high end options cannot be overlooked.

At the risk of making me sound like an Nvidia shill, the 1080 is targeted for those who had held back from upgrading. That's why the performance delta is compared against a 980, NOT a 980Ti. AdoredTV's case is argued using the point of comparing the 1080 against the 980Ti, and a heavily overclocked one at that, which at that point isn't a favorable or sensible upgrade decision after all.

There is an even more niche group of people getting this for a combo of reasons: VR at a lower power consumption. Pascal VR-specific features may trickle down to previous Maxwell tier cards, but power efficiency isnt. And huge, big honking VR rig uglying up a living room versus a lower consumption 1080 card placed in a slim HTPC enclosure, or a Raven RVZ01/02 hiding innocuously as an HT deck, is an interesting enough proposition to upgrade, without having to deal with 980Ti overclock and fan-whining itself to the annoyance of those in the living room while doing VR.

In short, AdoredTv is disappointed because the performance increase isn't substantial enough (when compared with 980Ti) to his liking. The performance delta IS substantial enough if you compare to those who have held back from upgrading and is waiting for a price point good enough for them to jump in to 1440p or 4K. That was previously the province of 980Ti and Titan X which are not accessible to these potential buyer of 1080 because of the exorbitant (to them) pricetag. The 1080 is targeted at the 980 and below users, not the 980Ti. AdoredTV should've waited for the 1080Ti or the Titan Pascal if he wants that "significant enough" performance delta he craves, but.......from the way he worded himself (as much as I am amused by his dialect), he's also looking for affordability. He did say there are no free lunch in this world when he was talking about power consumption during the 1080 overclocking in his review, I now tell that same thing to him as well. You gotta pay to get that performance margin. There are no free lunches.

The argument that "you will get better options and cheaper price with better performance if you wait" is ETERNAL, why wait a few more weeks for the AIB non-reference/non-FE version when you can wait a few months more for the 1080Ti version? Why wait for that when you can wait even longer for Vega? What wait for Vega when you can wait a little longer for Volta and HBM2? Some lines were drawn where this is where most of those who waited with their 980 and below cards are ready to jump in. To them, getting a Titan X like performance for a non-Titan-X price is where they draw the line for them to jump in. Sure, they can always wait for a cheaper and better performance card on the horizon, but many have fallen into that trap and continued waiting, to the point where they never upgraded and play Dota and Counterstrike all their waiting life and not bothering to upgrade after that.

This post has been edited by stringfellow: May 21 2016, 07:20 AM

How much anyone wanna bet that there will never be an AIB card that is priced at USD599, as the suggested MSRP by Nvidia? I mean, if a card by partners are made with better overclocking ability, higher stock frequency and better cooling solution, why should they sell it at USD599? I'd even venture that it'll be even higher than USD699 that Nvidia is selling the FE for.

Not everyone's situation is the same. Some have been surviving on even worse situations than a poor 970 at 1.7K. Try a power hungry 480 that is close to burning down a house.

As for VR, it is a valid point. TO have all the benefit of overclocking, you need either a larger sized tower or liquid cooling (encased in large towers). Rooms big enough for VR are usually the living room, and unless you dont give a shit about aesthetics, a big honking tower isn't it. And putting a card into the smallest, aesthetically pleasing enclosure means rigs built for VR in this situation are made with little or no overclocking whatsoever. And these group of people are coming from the demographics that aren't geeky or nerdy enough to care about overclocking, they just want a rig that fits their need and fits their aesthetics of their living room. We're the big honking nerds who dont care about the size of our towers, we just want to eke out the last bit of performance out of our cards by overclocking.

Our interactions here evolves around folks who had been dabbling in the PC business for ages that it's insulting to even thinking about "little to no overclocking" our cards. Guess what? We're not the only ones who wants a PC. There are burgeoning interest in PC outside of these forums and hardcore PC audience but they dont wanna deal with the intricacies(to us)/hassle(to them) of overclocking. They just wanna run games. That's why apps like Geforce Experience with auto-optimization is made for. These newcomers, your commonly named derogatory "noob". You can mock them if you like, but they're there. They're a part of the demographics who wants to jump into gaming on PC right now, and 1080 looks good enough for them to game on their living room 4K TV.

We are the nerds. We are the more informed, and most importantly,the MORE INCLINED to dabble in tinkering with our purchases and pushing it to its limits. They (the noobs) just want to play their PC and VR games in their living room on their 4KTV using Xbox controllers (blasphemy!). This forum may consist of purely hardcore audience that look at this demographics with derision, but hey, like you said, they have the money and want to jump in into VR and PC gaming on the couch. Their disinterest in overclocking or waiting for better AIB solutions and/or/if price drops, does not make it any less valid on getting the 1080 now. The rest of us nerds are happy gaming in our stretched out 3 monitors surround (until Nvidia fix that with SMP) and our big honking towers in our rooms.

This post has been edited by stringfellow: May 21 2016, 08:37 PM

Landed on number #48 on the Firestrike Ultra Hall of Fame benchmark leaderboard for single card. Not gonna last long, once the horde of folks getting their cards starts benching, I'll be buried.



On a different note, Doom 2016, all setting max in 4K, 77fps on OCed GTX 1080 running at 2.062Ghz core clock. Not bad. Drops to lower 60s in demon-infested areas.





This post has been edited by stringfellow: May 28 2016, 07:38 AM

No access to wattmeter here. I'm using an SFF Silverstone 600W PSU here that fits in the Raven RVZ01.

Those who had me on Steam, would've saw me going in and out of Project Cars like a madman. I was troubleshooting how demanding Project Cars is in VR and these are the settings that I've finally settled on after tons of tweaking. I could probably fine-tune it further but, in VR, a game's telltale sign of dropping below the VR comfortable threshold of 90fps is when judder is evident while panning around with your VR headset.

I could remove AA altogether and flick other setting notches up, but the sheer jaggies in 1080p without AA makes my eyes bleed. Can't imagine gaming at this resolution, but I'm forced to accept the reality of it all playing Project Cars in VR on my HTC Vive.

These are my settings, on GTX 1080 .I was hoping for at least 1440p, but it chokes even on 1080p set on Medium in VR. Probably need to wait for the developer to implement Pascal-specific VR APIs to alleviate the workload stress on the card. We have a long way to go in VR, folks.



I chose rain weather in thunderstorm in every track I ran the solo run multiple times. Figured that those rain effects would be the most punishing effect of them all, and wanna see how far the card can go. The policy is here is use the worse case scenario (racing in thunderstorm), anything less than that would've been easy-peasy for the card to handle.

This post has been edited by stringfellow: May 29 2016, 12:26 AM

Did another benchmark run.

Currently the highest Core i7-4790K Devil's Canyon CPU based GTX 1080 benchmark score in the Hall of Fame Firestrike Ultra single card run. The rest have their scores either inflated by their 5930K or 6700K CPU scores.

Core clock ran at 2.126Ghz at the highest, lowest at 1.947Ghz. Memory clock rock solid stable at 5.500Ghz, 11Ghz effective(!), with voltage at maximum, power target at maximum 120%, temperature target at maximum 92C, which saw the temperature rising up to maximum 78C with fan blowing at maximum 100%. Yes, it sounded like a vacuum cleaner, but this is just for benching purposes.

Detailed link to the benchmark run: http://www.3dmark.com/fs/8608543

In short:
1. The highest 4790K + GTX1080 in the Hall of Fame Firestrike Ultra Single Card run.
2. 4th highest GTX1080 score in that same leaderboard.

World's Number #37 in ranking. And this score running from components residing in a PC footprint slightly bigger than an Xbox One. Who says PCs have to be big and ugly?



I have however settled with a mild overclock of 1.987Ghz core clock and 5.4Ghz/10.8ghz effective memory clock, voltage percentage at 75% that nabbed a 5452 score. http://www.3dmark.com/fs/8608333.

Checked with Doom, still running that laughable 77fps at 3840x2160 everything MAXed out.

Set a proper fan curve or set a fixed fan RPM on Afterburner/Firestorm/Precision. Done.

I had the fan "hunting" around when temperature spikes and drops before. After a set fan curve (previously) and a set fan RPM (now set at 60%, barely audible), no more "hunting".

Probably because Pascal has a smaller die size with even more transistors packed into an even smaller architechture, hence temperature spikes are concentrated around a smaller die size happens more often, fan have to ramp up quickly to mitigate temperature spikes to cool down, turns off when temperature spikes are gone, and the game you play demands more processing power, core frequency needs to increase again, temperature spikes again, fan ramps up again, rinse and repeat.

Going into Afterburner and set a fan curve or set fan RPM settles this. Also helps if you "Control-F" (only works on beta Afterburner 4.3.0 with detection of a Pascal card) to set GPU Boost 3.0 specific Voltage/Frequency curve as well, so that you have tighter control over temperature rise/fall.

http://www.guru3d.com/articles-pages/gefor...rner-4-3,1.html



I know the title of that article says "overclocking GTX1080", but setting that specific V/F curve definitely settles the temperature rise and fall to a steadier rate, hence stopping fan ramping up and down.

This post has been edited by stringfellow: May 29 2016, 01:00 PM

Works for me. With the Voltage/Frequency curve as well. I've been benching this morning for 6 hours straight and stress testing in and out to get my benching overclock and stable overclock, in and our of GTAV at 4K and PCars in VR. Using the beta EVGA Precision doesn't allow the Voltage/Precision curve to be set, but the latest Afterburner does allow that. My fan does ramp up and down in Precision, even when set in K mode, but stop doing that in Afterburner after setting the V/F curve.

This post has been edited by stringfellow: May 29 2016, 03:57 PM

Fan noise at 60% is barely audible. Demonstrated by jayTwoCents on his YouTube channel. Same case in mine in a smaller Raven RVZ01 case.

This post has been edited by stringfellow: May 29 2016, 04:05 PM

It's especially important to folks who have their rigs in the living room for living room couch gaming or VR. Fan noise sounding like vacuum cleaner will drive your guests crazy when you're trying to demonstrate your VR rig to them.



My Raven RVZ01 is behind that 4K TV on the right side. Any audible noise will be heard directly. I had to tune the overclocking so that it doesn't detract from the VR experience.

That pic is outdated though. The Rift has been replaced with the HTC Vive and that shitty Microsoft All In One Media Keyboard has been replaced with the Razer Turret.

This post has been edited by stringfellow: May 29 2016, 04:22 PM

The more fair comparison would be comparing the P scores. Unless that too is influenced by the type of CPU.

I'm in it just for the hell of it. Once everyone and their grandmas get their GTX1080 it wouldn't matter to me anymore. ...... until I get the Titan Pascal or 1080Ti. Then the vicious cycle begins again.

Only when it hits 82C. Set the V/F curve in tandem with temperature target below 82C and the fan curve you can tolerate. My stable overclock stayed at 78C with 60% fan RPM. Otherwise avoid the Founders Edition and get custom AIB cards.

Or get the Founders Edition and put it underwater. Waterblocks are more common for reference designed cards like the FE than AiB cards. That's what I'm planning to do if I ever need more power.

This post has been edited by stringfellow: May 29 2016, 07:03 PM

Not so much of an overclocker, even with all the extra power connector and power phases, eh? What a shame. and I thought the allure of AiB cards was its overclocking capabilities over the FE, because of those extra power connector and VRM power phases. The only trump card AiB cards have now is the price.

AiB cards are useless to me since the card is required to fit into a slim profile SFF case, and all those fat heat pipes, bling-ed out lighting, elongated PCB board to house the extra VRM power phases, would not fit into my case (Raven RVZ01). Plus, it's nice to be able to go liquid cooling later if I decide to ditch the case, with EKWB making waterblocks for FE cards more readily available, although with that voltage lockdown, the card may just overclock only as good on air as it is on water.

http://videocardz.com/60631/asus-rog-strix...or-overclocking



Teeheehee! Even my supposedly "poorly overclocking" FE card can hit that 2050Mhz clock and higher (my Firestrike Ultra bench clocks at 2088Mhz).

This post has been edited by stringfellow: May 31 2016, 07:48 PM

Looks like this generation, Nvidia is putting a stop on overlapping performance between its two cards (970/980 and now 1070/1080). Last gen we see 970 practically making itself a 980 with its overclocking capability (although with that 3.5GB debacle), but at least for now, the 1070, even overclocked, cannot even beat 1080 at stock. Nvidia is clearly segregating its market segmentation aggresively this time.

Tracing back that article picked up by these publications, the source(ChipHell) mention GTX1060 as GP104, when it should be GP106. I smell a fake.

Plus, Nvidia has no real motivation to release a 1080Ti, unless AMD responds to the GTX1080 onslaught with their own. Polaris is midrange at best, and Vega is 2017 (March to June, at the earliest).

This post has been edited by stringfellow: May 31 2016, 07:16 PM

People tend to forgive a lot of flaws and imperfections when it's priced that cheap. On the opposite end of the spectrum however, people get extremely picky and bitchy when they paid what they consider as "expensive" and expect the card to do miracles and impossible performance expectations.

Selection and confirmation bias.

http://videocardz.com/60631/asus-rog-strix...or-overclocking

I get the same overclocking headroom as with my FE. Without that extra 6 pin.

So much for better OC. Gamers Nexus posted that the Pascal architecture has voltage lockdown/hardware limited at 1.25V, if that is the case, you can plug in the whole PSU into the GPU also no point. The extra pins are for cleaner power delivery. And the only way to by pass this lockdown is custom vbios. Which will void warranties if done so, unless you can reflash it back after you've blown your card.