|
Hidden Information Found In HTML Source Code of NVIDIA’s GTX 1080 Page lol
|
# ? May 7, 2016 18:05 |
|
|
# ? May 30, 2024 11:06 |
|
Hey sorry if this is a stupid question but I wasn't able to watch the press event and just getting caught up. Did they actually describe the metrics of what "relative performance" is? Or give any real FPS (even benchmark) numbers? "Relative" sounds really nebulous.. like it could be using $$ as a factor or something which could be reasonable to use, but if for example performance is really not all that much higher but costs "relatively" less, could really skew a comparison graph.
|
# ? May 7, 2016 18:10 |
|
slidebite posted:Hey sorry if this is a stupid question but I wasn't able to watch the press event and just getting caught up. The "relative performance" numbers seem to be based on (synthetic) VR benchmarking. The game performance graph on the product page (which someone grabbed some raw numbers from and the game settings in that link I just posted) seems slightly more concrete. quote:The “Up to 3x performance” originally had a disclaimer that it was “based on test results in graphics intensive VR gaming applications”. This disclaimer was commented out quote:The graph that compares the GTX 980 to the 1080 in the ‘performance’ tab doesn’t display exact numbers, meaning most readers are forced to estimate the values. I opened up the inspect element to view the width in pixels of the chart bars, and used simple math to determine the percentage difference. The GTX 1080 performs 70% better in Witcher 3 and 80% better at Rise of the Tomb Raider than the GTX 980, according to the benchmark. These benchmarks were conducted at max settings @ 2560×1600, according to a comment in the source. Until we see real benchmarks conducted by another party, I'm going to assume the above is the closest thing to reality right now
|
# ? May 7, 2016 18:15 |
|
slidebite posted:Hey sorry if this is a stupid question but I wasn't able to watch the press event and just getting caught up. e: VVV Same, but I'm most anxious to see how it does in the mixed GPU vendor tests like Ashes. There's a perf hit in DX12 multiGPU (as well as just SLI alone) using nV cards vs. Fiji alone or CrossFire'd Fiji - if they resolved that with improved async I finally get to fulfill my dream of having both red and green GPUs at once. ZobarStyl fucked around with this message at 18:23 on May 7, 2016 |
# ? May 7, 2016 18:19 |
|
Thanks for the replies. I'll, I'm curious to see independent testing
|
# ? May 7, 2016 18:20 |
|
ElTipejoLoco posted:If current GPU I have is a GeForce 9500 GT that currently has no cooling fan in its heat-sink because it broke, should I wait for the fabled 1070/1080 to come out, or should I stop living way back in the past yet? If you have a 9500GT how old is the rest of your system? Is an old core 2 duo, or an earlier i5/i7? If you're doing a full PC upgrade (I am) then either card is going to be an unbelievable jump for you. Barry posted:I had a bolt of lightning hit like a block away from my house last week that fried my cable box and the HDMI inputs on my receiver. Sucks, but it's about 6 years old, so I'm not all that miffed about upgrading to the latest and greatest. Get one (or more) of these: http://www.amazon.com/gp/product/B000HPV3RW/ You can run your cable through it as well which ensures more than just regular electronics are covered. I use one behind my TV and a PSU with my computer since I had a similar thing happen about a decade ago, only it was my PC that got fried because it was on a cheap and lovely surge protector.
|
# ? May 7, 2016 18:25 |
|
repiv posted:Depends if the monitor supports HDMI 2.0. Older versions maxed out at 1920x1080, unless you dropped to 24/30hz. It's HDMI 1.4
|
# ? May 7, 2016 18:25 |
|
A guy just broke nda from the deep dive presentation thing going on right now over at beyond3d :quote:If I'm getting this right, the current presentation just said that Pascal can finally change the allocation of SM-blocks to either compute or graphic tasks at runtime, with the same granularity as preemption. Controlled by the driver.
|
# ? May 7, 2016 18:26 |
|
Jesus well now I am torn because while a 1080 is awesome it looks, you know the 1080Ti is going to be stupid with HBM2 if the performance increase is real. And I have to decide if I want to stick it out with my 980Ti Strix and loose a tone of value, or see if I can sell it now and recoup the cost for a 1080(ti) in a few months. Grr That would mean no VR gaming between now and then though so... Crap. Not like I have had time for it anyway, but man being down a system for the summer doesn't sound fun......
|
# ? May 7, 2016 18:29 |
|
EdEddnEddy posted:Jesus well now I am torn because while a 1080 is awesome it looks, you know the 1080Ti is going to be stupid with HBM2 if the performance increase is real. And I have to decide if I want to stick it out with my 980Ti Strix and loose a tone of value, or see if I can sell it now and recoup the cost for a 1080(ti) in a few months. Grr See if you can still get a good price for the 980 Ti to trade it in for a 1080 cheap?
|
# ? May 7, 2016 18:32 |
|
Regardless on whether real world shows as much headroom over a 980ti under normal applications, the VR enhancements make it totally worth it to me. Also, the power consumption means that it should consume maybe 20w more than my GTX460 which means I can get away with not buying a new power supply. Hopefully supply won't be constrained too much. My rift ships between June 6th and 16th, so the timing couldn't be more perfect as long as they don't get backordered for months.
|
# ? May 7, 2016 18:33 |
|
fozzy fosbourne posted:The "relative performance" numbers seem to be based on (synthetic) VR benchmarking. The game performance graph on the product page (which someone grabbed some raw numbers from and the game settings in that link I just posted) seems slightly more concrete. The chart you're referring to showed a TitanX at 3.6 and a 1080 at 4.4 That's about a 22% improvement (my math could be a little fuzzy on that) So a 22% gaming performance improvement and it only draws 150ish watts versus the Titan X drawing 250 or so about 66% more efficient while providing 22% more performance. That's not terrible. I see a lot of people crying about the 599 price tag (on other forums,) I think they've failed to realize that the cost of 16nm finfet is much greater than 28nm. Nvidia's margins probably aren't going to be as good for 1 or 2 generations as they were at the end of 28nm. I'm running a 980 and plan on using the 1080 in my broadwell - E build (wish Intel would get off their asses and release BW-E already I'm only missing the CPU to complete the build.) We can probably expect the Titan Whatever it's called this gen in 8 to 12 months and a pricetag of 1199. The full size die won't be cheap.
|
# ? May 7, 2016 18:33 |
|
xthetenth posted:See if you can still get a good price for the 980 Ti to trade it in for a 1080 cheap? Darkpriest667 posted:I see a lot of people crying about the 599 price tag (on other forums,) DrDork fucked around with this message at 18:37 on May 7, 2016 |
# ? May 7, 2016 18:34 |
|
DrDork posted:Sell the 980Ti on Amazon (where they're still going for ~$500). Buy a EVGA 980Ti. Use step-up to 1070/1080 once real benchmarks hit. Don't worry about the 1080Ti because we won't see that guy for quite awhile. You're right but I do wonder about this founder's edition thing. I like my reference coolers and I don't think I should have to pay 100 bucks for a reference cooler. Someone said it's possible the 100 is for a vapor chamber only to be on the founder's edition. I'm really curious what the hell the differences are. Probably binning but who knows. I really would have liked some more information on that. EDIT --- according to hardware canucks the founders card has the same reference blower but machined aluminium instead of plastic. Darkpriest667 fucked around with this message at 18:42 on May 7, 2016 |
# ? May 7, 2016 18:39 |
|
I seem to remember the presentation mentioning the Founder's Edition would be for "sick overclocks" or whatever banal phrase he used. My assumption is just better binned chips. Either way I'd expect the same reference cooler.
|
# ? May 7, 2016 18:42 |
|
Ak Gara posted:I run my monitor at 3264x1836 @ 60 fps via dvi. If I swapped to a GTX 1080, would the HDMI out from that, plugged into the HDMI in on the monitor work? First, WTF is up with that resolution? What monitor are you using? The only mentions of that resolution I can find are smartphone camera sensors. Anyways, 3264x1836 is just short of 18 MB per frame, assuming we're talking about standard 8 bit per channel/24bpp RGB. At 60 Hz that's 8.6 gigabits per second. Technically you're exceeding the official DVI spec already. DVI-DL doesn't actually have a maximum clock, but 165MHz is the highest official speed just like single link mode. resulting in a spec maximum of 7.92 gigabits per second. HDMI 2.0 is the only HDMI variant with enough bandwidth to handle that.
|
# ? May 7, 2016 18:42 |
|
980 came out at $549 if what I'm reading is correct. How big was the performance margin for the 980 over the 780 in benchmarks comparable for the time to the 2560x1600 max Witcher and Tomb Raider benchmarks?
|
# ? May 7, 2016 18:43 |
|
I thought that Founder's Edition is primarily a way of nVidia to sell some "reference" cards for a mark-up and to help determine what the underlying demand is based upon the order volume for those cards. That and the mark-up is a way of paying a little more for early adopter access more explicitly.
|
# ? May 7, 2016 18:43 |
|
wolrah posted:HDMI 2.0 is the only HDMI variant with enough bandwidth to handle that.
|
# ? May 7, 2016 18:44 |
|
DrDork posted:I seem to remember the presentation mentioning the Founder's Edition would be for "sick overclocks" or whatever banal phrase he used. My assumption is just better binned chips. Either way I'd expect the same reference cooler. Well you'd think that but it's powered by a single 8 pin. My 980 is powered by 2 6 pins. How much voltage are you going to be able to push through it for better overclocks. These monsters already boost clock to 1750. Seems to me the 8 pin would be a limiting factor for heavy overclocking. Perhaps someone with more GPU overclocking experience can comment on that.
|
# ? May 7, 2016 18:45 |
|
Darkpriest667 posted:I see a lot of people crying about the 599 price tag (on other forums,) I think they've failed to realize that the cost of 16nm finfet is much greater than 28nm. Nvidia's margins probably aren't going to be as good for 1 or 2 generations as they were at the end of 28nm. I'm running a 980 and plan on using the 1080 in my broadwell - E build (wish Intel would get off their asses and release BW-E already I'm only missing the CPU to complete the build.) We can probably expect the Titan Whatever it's called this gen in 8 to 12 months and a pricetag of 1199. The full size die won't be cheap. They report their margins. They've been steadily climbing.
|
# ? May 7, 2016 18:46 |
|
xthetenth posted:They report their margins. They've been steadily climbing. Right because the process had matured on 28nm. The 16nm process was much more expensive and TMSC charged more for it. So their margins will either stay the same or go down. Also the number of good wafers was a lot less on 16nm. Nvidia has never sucked at making money so I wouldn't expect them to lose money on this I just think that their margins aren't going to be as good as they were at the end of 28nm. This is why the price had to go up.
|
# ? May 7, 2016 18:49 |
|
Darkpriest667 posted:I think they've failed to realize that the cost of 16nm finfet is much greater than 28nm.
|
# ? May 7, 2016 18:49 |
|
xthetenth posted:They report their margins. They've been steadily climbing. Really? I thought that was seriously hidden info. 4 billion is a pretty giant investment for an architecture if that figure was correct, was pretty surprised at that.
|
# ? May 7, 2016 18:50 |
|
Darkpriest667 posted:Well you'd think that but it's powered by a single 8 pin. My 980 is powered by 2 6 pins. How much voltage are you going to be able to push through it for better overclocks. These monsters already boost clock to 1750. Seems to me the 8 pin would be a limiting factor for heavy overclocking. Perhaps someone with more GPU overclocking experience can comment on that. The specs list the 1080 as running at 180W. A 8-pin + board power gets you 225W before you even break the PCIe specs, so that's 25% more wattage overhead available. And it's not like there's anything stopping you from breaking said PCIe specs, either *ahem 295x2 ahem* But yes, I'd expect that all the usual partners will be more than happy to slap another 6 or 8 pin connector on there if wattage looks like it's a limiting factor.
|
# ? May 7, 2016 18:51 |
|
Also, seriously, if you 980Ti holders are lookin' to sell, Amazon is the place to be. Just managed to sell mine for $480, having listed it last night.
|
# ? May 7, 2016 18:54 |
|
DrDork posted:Sell the 980Ti on Amazon (where they're still going for ~$500). Buy a EVGA 980Ti. Use step-up to 1070/1080 once real benchmarks hit. Don't worry about the 1080Ti because we won't see that guy for quite awhile. Where on Amazon do you see a 980Ti for $500? Oh nm, used.
|
# ? May 7, 2016 19:01 |
|
fozzy fosbourne posted:How big was the performance margin for the 980 over the 780 in benchmarks The 1080 and 1070 look to be a good step up from previous gen cards both in terms of performance per watt and total peak performance. I'd expect AMD to get similar gains in at least performance per watt just due to the new process. A 480X still probably won't beat a 1070, much less the 1080, but it should get close to the 1070 and if they can keep the price difference around $50-100 that seems like a solid value to me that would sell well. Lets just hope AMD doesn't make any more mistakes.
|
# ? May 7, 2016 19:08 |
|
GTX 670 to GTX 1080 sounds like a perfectly reasonable thing for me to drop cash on and start playing PC games more often. Can't wait for this poo poo. Now just have to find time (or one of you goons) to perform a similar quantum leap in the OP of this thread. Look at that eye diagram movax fucked around with this message at 19:13 on May 7, 2016 |
# ? May 7, 2016 19:10 |
|
DrDork posted:Also, seriously, if you 980Ti holders are lookin' to sell, Amazon is the place to be. Just managed to sell mine for $480, having listed it last night. About how long do you expect that gravy train to last? I previously said I should keep my 980Ti, but new toys are always nice! If we truly are looking for 12 months for the 1080Ti, and I want a stop gap 1080, the sooner I pull that trigger the better.
|
# ? May 7, 2016 19:20 |
|
Anti-Hero posted:About how long do you expect that gravy train to last? I previously said I should keep my 980Ti, but new toys are always nice! If we truly are looking for 12 months for the 1080Ti, and I want a stop gap 1080, the sooner I pull that trigger the better.
|
# ? May 7, 2016 19:23 |
|
Nvidia reference blowers are amazing for itx builds
|
# ? May 7, 2016 19:24 |
|
fozzy fosbourne posted:980 came out at $549 if what I'm reading is correct. How big was the performance margin for the 980 over the 780 in benchmarks comparable for the time to the 2560x1600 max Witcher and Tomb Raider benchmarks? It looks like the 980 is about 35% faster than the 780. https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_980_Ti/31.html http://www.anandtech.com/bench/product/1036?vs=1351
|
# ? May 7, 2016 19:26 |
|
Evil Fluffy posted:If you have a 9500GT how old is the rest of your system? Is an old core 2 duo, or an earlier i5/i7? If you're doing a full PC upgrade (I am) then either card is going to be an unbelievable jump for you. Yeah, I just learned my lesson that any old "surge protector" isn't really a good idea to put a few thousand dollars of electronics behind. Definitely getting something legit.
|
# ? May 7, 2016 19:30 |
|
Malcolm XML posted:Nvidia reference blowers are amazing for itx builds How loud do they tend to be? I like my itx pc to remain pretty quite, even while gaming.
|
# ? May 7, 2016 19:32 |
|
PC LOAD LETTER posted:980gtx was around 30% faster than the 780gtx I believe. The 780Ti actually came pretty close to the 980's performance though, around 10% difference on avg. MaxxBot posted:It looks like the 980 is about 35% faster than the 780. Nice! Now I'm curious about how the 980ti does on 2560x1600 "max" benchmarks for the Witcher 3 and Rise of the Tomb Raider, alongside 980 gtx benchmarks. We could use the percent different from that graph to model where the 1080 gtx might land in terms of fps as well as a percent difference between that and the 980ti
|
# ? May 7, 2016 19:55 |
|
Trent Razor1911 posted:How loud do they tend to be? I like my itx pc to remain pretty quite, even while gaming. The blowers in NVidia reference are actually higher quality than third party blowers (Asus, PNY) or AMD reference. They have a sound profile that is low pitched and not super annoying. As always though, I recommend the Thermaltake Core V1 for new ITX builds, if you're OK with a cubic-foot shaped case. They are sub $50, fit all but the hugest cards (i.e. Any card but triple-fan Gigabyte size), and have side vents so fan-style cards work great too. They also fit a full-size ATX power supply and 5 of the 6 sides of the case pop off so it's super easy to build in. http://pcpartpicker.com/part/thermaltake-case-ca1b800s1wn00
|
# ? May 7, 2016 20:01 |
|
Evil Fluffy posted:If you have a 9500GT how old is the rest of your system? Is an old core 2 duo, or an earlier i5/i7? If you're doing a full PC upgrade (I am) then either card is going to be an unbelievable jump for you.
|
# ? May 7, 2016 20:03 |
|
Man are people really trying to justify prices increasing substantially year after year, just seems crazy to me. I guess good for nvidia if they can keep doing it and selling cards? Shareholders must be loving it.
|
# ? May 7, 2016 20:09 |
|
|
# ? May 30, 2024 11:06 |
|
fozzy fosbourne posted:Now I'm curious about how the 980ti does on 2560x1600 "max" benchmarks for the Witcher 3 and Rise of the Tomb Raider http://www.techspot.com/review/1011-nvidia-geforce-gtx-980-ti/ Too lazy to do the number crunching though sorry
|
# ? May 7, 2016 20:11 |