|
Truga posted:Don't worry, SLI for VR is really easy you just render each eye on one gpu and will drop any moment now! IT worked great when we had the Extended mode, but now with the actual VR direct to HMD mode, they have yet to do anything with it, and I feel it's them waiting until the DX12 VR SLI support to roll out. Really though, the other issue is the latency between the 2 cards. While normal SLI doesn't have a problem with it, with VR, the latency between the cards can increase a ton based on the PCI-E Lanes and Spec you are running. Technically you need PCI-E 3.0 X16 to have near .00X latency where 2.0 X16 or 3.0 X8 = 1-4~.XXX latency numbers which is bad. That limits your market not only to the few with SLI, but the few with X79/X99 systems with 40 3.0 lanes...
|
#
¿
Apr 21, 2016 17:45
|
|
|
|
| # ¿ May 12, 2024 05:02 |
|
|
Pollyzoid posted:Has anyone had Afterburner somehow set absurdly high core and memory clocks by itself? I just had to figure out why I suddenly started getting artifacts under a minute after booting. Turns out Afterburner had set both clocks to +1000 MHz. I haven't even been running overclocks for months. Why does Anyone (Everyone) seem to still be using MSI Afterburner. Afterburner hasn't been good since the 500 series era or earlier. Switch to EVGA Precision and watch nearly all the Clock/Fan problems disappear since it actually works.
|
|
#
¿
Apr 21, 2016 23:18
|
|
|
Paul MaudDib posted:All of the initial GP104 releases are going to be die harvests, reportedly. The big question is whether GP104 has any DP capability, because that could increase the SP-GFLOPS/mm^2 rating over GP100. Considering the Move to GDDR5X, I would imagine the 1070 would be closer to 980Ti speeds than the 1080. However until we actually see some #'s, we are all pulling hope and dreams out of the Unicorn's rear end here. If after 2 years of Maxwell Nvidia can't make a new generation of GPU's faster than the previous then something would be seriously wrong.
|
|
#
¿
Apr 22, 2016 21:55
|
|
|
Jeez, thank God I have a desk with a Tower drawer that hold my PC safely inside away from anything that can get to it. If I am not actually putting a load on it I can even shut the front door and be a lot quieter. However if there is a Load on it, the heat inside will make that thing throttle up to F-14 loud if the front is shut. Recirculating heat from the rear is bad.
|
|
#
¿
Apr 25, 2016 20:23
|
|
|
Edmond Dantes posted:No, it wasn't that, Skyrim was just CTD as soon as I loaded a game, it wasn't my OC settings; I googled a bit and it looks like it doesn't play well with the OSD server Precision X has going. Yea the Onscreen crap Precision (or any of those GPU apps for that matter) seems to be a crapshoot on causing issues for games or not. However they work perfectly for everything sending the OSD stuff only to my G19's LCD and just using it for Fan Curves Otherwise. Also I got a Corsair 500R at the Nvidia Lan in 2011 and swapped from a P180 to it. The P180 at the time was that high quality, quiet case, however man it had terrible ventilation when you started getting crazy with SLI or bigger hardware (such as a 6 Core 3930K) and wasn't any good for a water setup unless you wanted to go all custom and loose out on the quiet that the case offered. Switching to the 500R though ended up being great as it had more room, much better airflow, and overall seems to be a much better case overall except for the front USB 3.0 port dying (1 out of 2,) I got the replacement front panel for free under warranty, but been to lazy to rip the entire thing apart to replace it yet. It isn't as quiet due to it being more open and having 4 fans for the H100 I threw on there, but overall when everything is idling it can quiet down rather well, and keeps everything rather cool with how much you can shove into it. Not many other cases have caught my eye for a future build but I guess partially the reason is I haven't really been looking. I can't do one of the cubes because my old style desk (hey I got it in 2000 after building my first PC OK. Its big and wood and has lots of space for junk) can't really fit a cube over a mid tower with the 500R being about as tall as It can take before airflow out the top becomes a issue unless I scrap the top vent entirely. Not looking to rebuild anytime soon really as outside of whatever the next Ti/Fury level cards offer, the rest of my rig really isn't being held back by any major point requiring a X99 build or anything....yet.
|
|
#
¿
Apr 28, 2016 01:43
|
|
|
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
|
|
|
If I sold the 980Ti now, then what is the wait time till a 1080 arrives for purchase? I heard end of May to September which is pretty broad. I am not going without VR for the Summer, no sir. I may just hold onto the Ti until the 1080Ti arrives and if I flip it for $300 id be happy. Also do we know how cut down the 1070/1080 are from a full GP100 chip? (and not the newest Tesla which is still cut down from a full chip).
|
|
#
¿
May 7, 2016 21:47
|
|
|
Why is the talk about vapor chambers for the reference blower cooler a new thing? The Vapor Chamber Blower Reference Design has been the same since the 690. It is a really good design and the newest one really only looks like they redesigned the outer shroud to make it more, angular. Whatever the "Founders Edition" is it is probably some special binned, Referenced OC'ed model, or something like a MARS card of past generations. The fact that it might be available in May tempts me to no end. The 980Ti runs VR quite well currently, but man, if the performance numbers are to be trusted, being able to truly play Elite Dangerous at 90FPS solid even on planets with a 1080 (even before the new TI arrives) is tempting as all hell. 
|
|
#
¿
May 8, 2016 03:56
|
|
|
Ok currently I just seem to be a slight bit confused. First Gsync/Freesync, from what I believe I understand, both sort of work together to either enable V-Sync above the Monitors refresh rate and disables it when framerate goes below, and/or adjust the refresh rate to sync the frames when it is below as well? When you have >60hz screen (120/144Hz range) unless you are going above 120FPS then what benefit does Freesync/G-Sync provide again?
|
|
#
¿
May 11, 2016 17:42
|
|
|
Last I checked though, tearing only happens when the FPS > Hz of the screen so if you have a 120/144hz screen and are rendering at 90fps, there is no reason to have anySYNC turned on because your screen doesn't have any problem keeping up. Now for the COD/CS players that game at 200FPS+ then I can understand that sync/input lag/studder benefit as it is back above the Hz range and tearing would most definitely happen with that high of a FPS/Hz spread.
|
|
#
¿
May 11, 2016 18:20
|
|
|
Gwaihir posted:The NV compute servers using NVlink are effectively doing 8x SLI so that seems like a pretty safe bet! I would think NVLink isn't quite the same as SLI as they are working on Workloads, but not needing to pass the information from all the cards back to a single card to output anywhere like SLI has to do. Another reason TriSLI may have been dropped is unless you get a X99 system (or AMD Equivalent), you lack the PCI-E Lanes to even run 2 cards at X16 let alone 3+, and with the memory speed increasing like they are and the much more important transfer latency for SLI in VR, they need to get either PCI-E X16 or come out with 4.0 before more than 2 cards are actually useful to anything but the super niche again. The only time I did Tri-SLI was with 3 8800GTX's from BFG (RIP). The performance gain from 2 to 3 was certainly a lot less then 1 to 2, but it still brought it to swinging distance of a 4870X2 at the time lol. Those cards only got retired last year with SLI 560Ti's that are purring away nicely to this day. It will be interesting with the way DX12/Vulkan is supposed to handle SLI/Crossfire in later cards and being able to toss work around especially with VR, but I feel VR may be the kicker that limited SLI back to 2 cards. 1 Card for each eye that is easier to test/validate/control. The performance gain with the new bridge I look forward to checking out too. And thanks for the GSync/Freesync cleanup, I completely forgot how it controls the redraw below the Hz level and syncs the frames accordingly. That does sound good/useful for sure. However since most my gaming is going to increasingly be on VR headsets, I feel it has less of a need on my end as async timewarp and such is sort of the VR solution to that. EdEddnEddy fucked around with this message at 22:01 on May 11, 2016 |
|
#
¿
May 11, 2016 21:58
|
|
|
Jesus if those numbers are real that thing has some legs on it. I was within (distant) striking distance stock vs my 980Ti at 1514Mhz, but once it OC'ed, it literally is running away from the pack. I can see why they say it is faster than 2 980's, they weren't kidding. This make holding onto my 980ti a little harder, however considering the numbers here, the Ti version (Hoping to god it is close to a full GP100) may be another 25% faster than the 1080 and then OC headroom on top of that. Literally over Titan X SLI performance in a single GPU? Yes Please. Elite in VR would finally be GPU bottleneck free landing on planets, and even have some headroom maybe for running it at 1.5X Res.
|
|
#
¿
May 13, 2016 16:57
|
|
|
SlayVus posted:NDA ends on the 17th I think. Then you still have to wait another 10 days to even pay $100 over msrp. Anyone who owns a 980 Ti should just wait for big Pascal, IMHO. We may not even see big Pascal until the end of the year though. Either way it goes, I am not really tempted too much to get a 1080. Look at how bad the 980 faired vs the 970/980Ti's. It was sort of a price point chip, but really didn't provide the worthy bang that the 970 or eventually the 980Ti's did. If that has any chance of happening again, I am ok with waiting till the end of the year for a Ti that literally curbstomps the 1080 with sheer performance for $200 more as I really don't see VR SLI picking up steam for another year or two. So having the best single GPU I can get will still be my priority for the near future. Currently the only VR titles that kill the 980Ti is Project Cars and Elite, so a 1080 might fix that, but a 1080Ti might allow me to have everything maxed and more room to spare. Guarantee Project Cars in VR can still tax that 1080 stock though. My 980Ti only gets like 45-55FPS with a full grid and some settings turned down (shadows/reflections/etc) and I have a target of 90FPS solid to reach.
|
|
#
¿
May 13, 2016 18:32
|
|
|
That 1080 does look good, but I think I can hold off with my OC'ed 980Ti for a while still. The number seem to point to like said earlier, 980 level performance over say a 780(Ti) and with the Founders Card power limits, really limits its OC potential it appears. That card probably has another 10-20% OC headroom in it with more power as the cooler can easily keep that thing a lot cooler than what they seem to be testing for. I have no idea why they just don't set that fan to 100% to see what you CAN get out of it, rather then just setting a heat target and letting the heat be a factor in the Turbo OC. Power is a wall you might not be able to go past yet, but heat shouldn't be a factor currently with the hardware thats available to them.  I do look forward to seeing what a 1070 can do and what AMD has up their sleeve because drat do they need something good. I also really want to see what a fully power fed unnerfed chip can do. All the Cores at 2.2Ghz~ and HBM2.... I could see it being 50%+ faster than the 1080 and a godsend for VR (Project CARS @ 90FPS on the dot) Yes! 
|
|
#
¿
May 17, 2016 18:52
|
|
|
THE DOG HOUSE posted:Yeah I've been griping about this for a looooooooong time now. Someone will do it but most wont and take the results at face value (a 1080 OC'd is a billionty times faster than a 980ti!! etc), but at least for the time being most reviews are apples to apples (no OC). Though once OC models start coming around with drivers its going to be more and more of this where they throw in 980ti benchmarks from literally release day reference card values that they've had stored for 20 months. Also it is not like the 980Ti is lacking in performance on its own with a nice OC. 17528 FireStrike @ 1518mhz 9391 FireStrike Extreme @ 1518mhz 4725 FireStrike Ultra @ 1418mhz Which appears to be within spitting distance of the 1080 per ARS lol. Not bad paired with my old 3930K @ 4.6Ghz (Though that is just a Benchmark, real game performance of course the 1080 just seems to walk away from the pack.  ) 
EdEddnEddy fucked around with this message at 19:17 on May 17, 2016 |
|
#
¿
May 17, 2016 19:12
|
|
|
The 700 series and 900 series did have a bit of a difference for the midrange and value though. In the 700 series, the 780 did arrive with great performance compared to the OG Titan (hell in most cases, it was actually faster but with 3G less ram). The 770 was a good card as it was literally a 680, but nothing like the performance/value of the 970 to come. The 780Ti arrived really late and was speced ABOVE the Titan of the time (it was a full Kepler which the OG Titan wasn't) which was an odd move IMO, however the Titan was the compute monster vs the 780's or so I believe so it kept some value there. Up until this point, since about the 500 series, it seemed that the performance of the cards moved down X10 each generation or so. So a 580 -> 670 / 680->770 give or take. The 970 sort of flipped that a bit hitting what could have been 2 X10's above. (If you count a Ti as a X90 or something like AMD would do). The 900 series arrived and Nvidia changed their game a bit. The 980 and 970 came out, but the 970 being as fast or faster at 1/2 the price of the 780Ti made it a steal where the 980 was faster still, sure, but it really wasn't $300 faster on top of that. Then The Titan X arrives with a true ungimped Maxwell chip and shows some great potential for a Ti which then arrives at a much more timely manner, and at a much more understandable price point for its performance, rendering the 980 practically moot until it dropped into the ~$400 price range where it makes a lot more sense. I don't really regret my purchase of 1 then a 2nd 780 back in the day, as they were badass cards for sure, however it was the lack of SLI support for VR that pushed me to upgrade earlier than I had planned, as the wait for the Pascale series was much longer than I had expected coming from the nearly yearly+-6 months of the previous 3 generations or so. The Maxwell series felt like it was going on for freaking ever. Now that the 1080 has arrived, and we all know that it is a higher clocked but 75% the cores of a P100 (which itself isn't a full 100% core card either) makes me take pause a lot on whats going to happen next... IF the P100 was clocked at the same as the 1080 (and OC similarly with enough power available) then you could have really good performance that would be worth the actual ~$650 a Ti should sit at, and $1K for the fully cored Titan as I doubt Nvidia will ever again made a Ti a full core part over a Titan again. Though the 1080 look drat impressive currently, the 1070 may or may not be another 970 due to the non X memory, Unless some sort of 3rd party makes a 1070 with GDDR5X memory too, which could make it a killer value and performance card probably, but we won't know until we get some benchmarks/stock/etc. If the 1070 ends out being sort of a flop, Nvidia would just have to get stock of the 5X memory and just make a 1070Ti to fix that really. Once the 1080Ti drops though, if it is as good as I can imagine it being (going off the Ti performance over the standard 780/980's) that should instantly shove the non Ti cards down ~$200 and if not, it will potentially just make it a poor purchase for the remainder of the generation. It does all weigh on how the 1070 does as if it stays stuck at just ~980Ti performance, then the 1080 might actually sit comfortably at the purchase point between it and the Titan then the Ti can compress the price spread a little but keep things comfortably $100 apart for ~20-30% performance each point..... This is all my own head gearworks though so until we see what we end up seeing, I am just basing it off of what I have seen and cannot be unseen because I spend way too much time looking at this stuff. I also mainly just want the best single card performance I can get this next generation for VR as VR SLI is still probably not going to be a reliable thing for another 2+ years.
|
|
#
¿
May 17, 2016 21:40
|
|
|
DrDork posted:Or NVidia believes that AMD won't have any compelling high-end parts to compete with anything above the 1070, that the 980Ti price was low simply to crush any possible life Fury might have had, and the price lineup ends up being: This is truly possible for sure... Hopefully AMD does have SOMETHING to throw against the Green Wall and make prices competitive again. (hell would be great if they surprised us even more with HBM2 and hit above the 1080 while pricing roughly the same or $50 more.) Guess we will see. I look forward to the 1070 numbers though. Its stock and OC numbers will be telling if GDDR5 is enough or not. If it ends up being another 770, I fully expect a Ti or Aftermarket model with 5X on it since the chips are supposed to be drop ins. I don't "Want" a 1080, but drat if I am not tempted to "rent" one from Frys and try one though when they appear.... just have to actually return it this time...
|
|
#
¿
May 17, 2016 22:06
|
|
|
japtor posted:Still waiting for 6880x2880 with high refresh and free sync, and whatever new version of DisplayPort to actually be able to run all that for a billion dollars. Considering I was able to game on 3440X1440 screen I rented for a lan with my HTPC comp that has a 20th Anniversary Pentium at 4.4Ghz with a GTX 760 in it. It played War Thunder, Rocket League, EVE, and a few other game perfectly fine. War Thunder was turned down a hair on AA, but besides that the little 760 really does impress me with what I can throw at it. Benchmarks don't always tell the hands on usage performance.
|
|
#
¿
May 19, 2016 01:29
|
|
|
Blackfyre posted:Urrrm I forgot to cancel my ebay auction for my 980Ti and it has sold for £390-ish. Just after I decided I would wait and save my money for the 1080Ti. Ouch, well it is up to you for sure, however when you can actually get your hands on a 1080 is anyones guess between now and 3 months from now with the way this launch is going. I am sorta sour from not waiting out with my SLI 780's before getting the 980Ti, but the Ti really has helped me not miss the SLI glitches and inability to do SLI in VR so I plan to hold onto it for a good while until the next Ti drops. In your case, you might have to play the waiting game on the IGP or something unless you have a Frys nearby and can "rent" a GPU for a month. (grab an open box one) While I loved the blowers on the 780s (They were real vapor chambers then, no VC in the 900 series apparently, but it's back for the "FE" cards... Yay?) The nerfed power systems of the FE cards seem to really make it a bad deal unless you absolutely have to have the card as soon as possible. What bums me out too, is the Blowers are actually better sometimes than the aftermarket coolers and yet, you will probably never see a reference blower on a fully powered card from any of the 3rd parties. That thing runs quiet and keeps cards really cool. My 780's were OC'ed to 1202mhz and never broke past 70C with their reference blowers.
|
|
#
¿
May 19, 2016 16:01
|
|
|
Blackfyre posted:Excuse my ignorance but IGP? Integrated Graphics Processor. The GPU on the CPU if you have an i5/7 non -E chip and you don't have a backup card to use while the 980Ti is gone. :/ Yea it is lovely but maybe the guy will understand that you kinda screwed yourself on timing and don't care to blow $$$ on the latest and greatest card that will be out, whenever. :/
|
|
#
¿
May 19, 2016 16:26
|
|
|
SlayVus posted:Petition: Henceforth the GTX 1080 shall be known as the GTX18 or GTX 10.8 It really has been way, way, waaaaay too long since either them or AMD(ATI) freaking come up with a new name for their cards. GeForce has been around since what, 1999? Before that we had TNT and TNT2, Vanta, etc. ATI had the Rage, Fury (hey! At least that's back), Maxx, Pro, etc. The Radeon has been around almost as long as GeForce has. Titan is like the only newish name to come out in forever.  I am just sour how stale the naming scheme has become and how the numbers/names just seem to flip around each generation sometimes making sense, sometimes just changing to screw with peoples heads once again. (Looking at you Intel... 3960X/4960X/5960X/6950X?)
|
|
#
¿
May 19, 2016 22:20
|
|
|
I agree that AMD really needs to get this right this time, and I believe they know that too, but the question is, will they? I love Intel, but dammit to hell they need a real kick in the nuts not only because of the current crapshow that is the fallout from selling ARM all those years ago and going all in with Atom, but also their current complacent CPU view with the endless drivel of "The PC is Dead" talk. The only reason the PC is Dead to them, is because they haven't really released anything outside of mobile, that has garnered any real drive to push people to that "I need to upgrade" level that they were able to do in the P4 -> Core2 -> Core i7/i5 days. Now they are going all in with IoT and 5G, and while their CPU stuff isn't going away, it just continues to be super unexciting where the GPU side of things for Nvidia/AMD has the potential to be not only exciting, but also needing more CPU power now than ever (If DX12/Vulkan is to be believed). Back on Topic, AMD has the ultimate chance to hit Intel right in the jewels with hopefully a good CPU architecture as well as both Nvidia/Intel with a good GPU architecture that not only takes Nvidia down a notch (if anything bringing some price wars back) as well as making their APU's competitive to Intel's offerings on a performance per watt solution. It currently continues to be a big wait and see, but man it would suck if we wait this long, and AMD completely drops the ball on both ends to continue this stale Monopoly style market that while still exciting, is a bit hard on the wallet with the lack of competition.
|
|
#
¿
May 20, 2016 16:08
|
|
|
Froist posted:Hang on, when did Intel ever own ARM? I may have missed something (or less publicised share ownership) but I'm 99% sure it's been a Cambridge-based/owned company since the get-go back in the 80s, and never linked with Intel ownership. Left out some details, but Intel had their own ARM tech that hey sold off (XScale) which at the time was pretty badass and them selling it off to go 100% x86 was literally the worst thing they could do at the best time to be in Mobile.  That and buying McAfee. Fvck that CEO screwed Intel hard and now BK is taking the blame for something he didn't really cause, but is doing a poo poo all to fix.
|
|
#
¿
May 20, 2016 16:32
|
|
|
HMS Boromir posted:Shouldn't DX12/Vulkan need less CPU power than ever? I thought one of the big improvements they bring is less single threaded overhead - that hardly seems like an impetus for Intel to squeeze more blood out of that particular stone. It might improve games' ability to use 4+ threads, but increasing core/thread count isn't really that exciting either, is it? I doubt people bored by Ivy Bridge / Haswell thought the 5820K was a revelation. Like "Edited" in in the post above yours, the Potential of being able to use More than 4 threads in the rendering pipelines has the potential to use more CPU power than current games seem to have been coded for (so why OC'ing/Clock Speed has been a bigger boost than just throwing cores at the problem) but at the same time, with GPU performance needing to scale up in the next year or two to handle not only 4K, but whatever is in the future of 4K and VR, they soon are going to start getting hit with the need to feed these things and other number crunching that just throwing more cores at the problem isn't going to fix. Latency between the CPU/GPU('s) and frame time is only going to become more and more of an important factor too with VR really going to be the driving force, not only for games too. Considering we are starting VR slowly with simple graphic games and such, it will only be a matter of time before they start wanting to push Photo Realism into VR similar to what they have been doing on flat screens in recent years, and the only way to do that, is massively improved hardware if Project CARS/Elite Dangerous is anything to go by. When it looks that good in VR, it is downright breathtaking and immersive as all hell. What has pissed me off is the efficiency has gone up sure, but even with OC'ing, the sheer performance of a single core hasn't really moved much more than 10% per generation which is pretty sad even coming from the jump we got from Core 2 to the Core i Series. I know new architectures take a lot of time and engineering to create (look at AMD I guess), but dammit if Intel is due for a new chip that isn't just a Atom. Nothing since SandyBridge - E was announced, has really gotten me excited for a CPU outside of some of the tech that the chipsets going with those CPU's has to offer. (Sandy Bridge - E brought back what Sandy Bride's chips seem to have lost from the X48/X58 platform) > 16/20 PCI-E lanes and Quad Channel RAM rather than Tripple/Dual channel. Now the new Z170 series has some neat bells and whistles that even X99 doesn't have natively, but nothing is quite pushing me to need to upgrade my aging X79 as outside of the more cores that even Haswell - E brought, my 4.6Ghz 6 core can still hit in the same ballpark as a mildly overclocked 8 core, which is great for the old tech, but sad considering how old it is in comparison now.
|
|
#
¿
May 20, 2016 17:05
|
|
|
Animal posted:Is PCIe 8x gonna bottleneck the Geforce 1080? I am building a new rig and thinking of putting an m.2 SSD drive which is gonna bring the video card from 16x to 8x. If it's gonna reduce performance, then I'll just get a regular SATA SSD. If you are going VR, it may as well as higher res stuff. Forget SLI as well unfortunately. Finally with the higher end 900 stuff and now with 1080, we may finally start to saturate the 8X lanes and we already have for VR (need X16 3.0 for VR SLI especially.) Also on a separate note, I wonder if it would be possible to Build your own Razer Core for a external GPU. Talking with some Intel guys at the lan and they laugh at how simple the Core really is, and how it is currently just 99% markup for what it provides. You would think if that is the case though, that they and everyone else would get in on the external GPU box bandwagon.. I am guessing there may be some form of Thunderbolt licensing BS to jump around before you can bring one to market.
|
|
#
¿
May 20, 2016 19:25
|
|
|
https://twitter.com/OC3D/status/733723998853324800 What is this? I like how the graph looks HUGE but is only in the 100's of points difference, however a "Mobile" chip that passes up a Titan X? I really do look forward to upgrading from my ASUS G73JH and it's 5870M and waiting till after the 900 series was the right choice it's looking.
|
|
#
¿
May 20, 2016 19:35
|
|
|
xthetenth posted:So the successor to the "gently caress it we're stuffing a desktop 980 in this heatsink barge of a laptop", then? Desktop 1080 being 30% faster because of a TDP limit sounds plenty reasonable. Yea that's what I was thinking, and if they used the desktop 980 designed chassis, they could have lots of room to OC even so all that R&D doesn't go to waste.
|
|
#
¿
May 20, 2016 19:49
|
|
|
Ok if that chart is accurate about the 1080Ti/Titan, I am sort of disappointed about the 5X still being used, however having another near 1000 cuda cores and a big chunk of ROP's will make that card downright scream either way. That stock clock will more than likely not hold it back from a good OC either, but to keep the cards "quiet and still cool enough" will required them to pull some clock like seen in these OC videos. What I don't get though, is the stock cooler should be able to keep a stock 1080 easily below 70C at max and still be quiet enough to not annoy the crap out of people compared to a AMD blower, why in the heck when they do these OC benchmarks they don't max it out or run a custom fan curve to really say if the stock cooler is inadequate or not? (I know the video rundown says they did it at 100% and it was fine then, but I believe Guru3D just used a certain set % (not 100%) and hit 83C in all their OC runs when talking about OC'ing. Very strange) Since forever has the fan curves been set to balance out a heat target vs noise level, and usually it always ends up sucking except for under ideal conditions. Zotac 560Ti SC was the absolute worst as the stock fan profile with the after market coolers, only went to 50% and if heat went above what that could handle, went into a sort of emergency 100% setting that was somewhere in the bios until the chip cooled back down. Didn't affect gameplay, but was loud as crap going from comfortably quiet to "WTF IS GOING ON IN MY PC!?!" Tweaked the BIOS and flashed that to control the fans (Breaks the auto fan profile completely, but allows the full % range from 0-100) and EVGA Precision fan curved it to perfection. EdEddnEddy fucked around with this message at 17:30 on May 23, 2016 |
|
#
¿
May 23, 2016 17:20
|
|
|
Animal posted:An open cooler plus the CPU cooler (Corsair H100i) is a lot of hot air being dumped into such a small case. I can experiment with a top exhaust but it just seems like a bad idea overall. Everywhere I looked it's not recommended. You have a H100i Venting into your case?
|
|
#
¿
May 23, 2016 18:46
|
|
|
https://twitter.com/OC3D/status/734778445339627524 Jesus, didn't know that the 6/7XX series was going to get Legacy'ed so drat fast considering they still support most of the major DX12 stuff that isn't VR related. Poor 780/Titan owners now.
|
|
#
¿
May 23, 2016 19:05
|
|
|
Ahhhh. Ok. I thought it would have been a really weird/stupid move considering that even though its older tech, the hardware still supports nearly everything currently and beyond still. Only the DX10/11 stuff from Nvidia/ATI got moved to Legacy which makes sense since the hardware cannot support DX12+ going forward no matter what the drivers do. Luckily both sides have given some of the legacy stuff a break with Windows 10 drivers and even AMD gave the old 5000 series a crimson driver which was more than I expected for 2010 tech. Speaking of Old Legacy tech. I was able to shoehorn the latest Legacy driver for the 8000/9000 Nvidia drivers onto an old 2006 Sony laptop with a 8400GS/Intel switchable setup and man what a performance/support difference it made in Windows 10. The GPU still sucks, but it fixed all the graphical issues/glitches/support stuff I was having with the only old Windows 7 driver I could find before.  fozzy fosbourne posted:Frys is apparently accepting preorders of FE cards right now: http://www.frys.com/product/8842862 I may have to go in Day1 and see if I can get my hands on a 1080 to "test"...
|
|
#
¿
May 23, 2016 19:56
|
|
|
AVeryLargeRadish posted:Actually according to the video poster earlier the default fan curve causes massive judder and other problems in actual games because once the GPU hits it's throttle point it rapidly switches between its base clock and much lower clocks. The only way they found to fix that was to peg the fan at 100% at which point it is extremely loud at 60dBs but the GPU can now sit at its boost clock the whole time. I thought it was only throttling when they were running Overclocked and it was hitting the Temp limit. I thought running stock it was chilling in the 60C range most of the time?
|
|
#
¿
May 23, 2016 20:14
|
|
|
AVeryLargeRadish posted:Nope, it gets to its temp limit of 83C just running at the default clocks/boost clocks and then starts throttling like mad, it's just that it takes well over ten minutes to reach full temp and start doing that so most reviewers never saw that behavior because they were just running benchmarks long enough to get a score, when tested in a case(Fractal Design Define R5) and playing games for extended periods it displays the judder and throttling behavior. You can fix it by ramping up the fan but that will make it much louder. BTW That is surprising considering how much power savings this car is supposed to have vs its performance. It's like they either broke the reference cooler that was soo good on the 690/780+ (The 900 series wasn't vapor chamber I believe), but at the same time, won't let it ramp up to keep it just below say 80C. WTF is the temp cap doing at 83C anyway? That feels surprisingly low for a GPU to start spike throttling so badly that it would cause judder so badly in games after only 10 minutes. Wasn't the 780's and other cards sort of 85C with a hard throttling cap at 90C or am i thinking back farther to my AMD days? (got 130C VRMs on a 4870X2 Refurb that was artifacting and getting replaced anyway lol  )If Nvidia doesn't have a fix in place before they hit stores in their drivers they might get hit like the 970 issue all over again, which wouldn't be bad for potential free stuff or discounts/partial refunds (-$100 maybe?) if they don't. If they try and say that under "normal" gameplay it doesn't reach that temp, then well... 
|
|
#
¿
May 23, 2016 21:52
|
|
forever if the FE 1080 ends up with the same problems that the stock cooler R9 290/X had.
|
Considering the Blower Fan looks like the same one they have been using since the 690 days, even at 100% it was nowhere near the loudness of any AMD one of note. Unless they somehow completely borked all the work they did on the fan shroud in the 690/780(Ti) day with this new angular design or something... Even then you shouldn't need 100% fans to keep the card at say 70C (when I max out my fan curve on all my cards these days) Usually on the 780 (even the one thats stuck close to the 1st in SLI so it is usually 10C hotter) didn't take more than 75-80% fan speed to keep it at or below 70C. The only time I had the fans ever sit around 100% for any period of time was Benchmarking while I had them OC'ed as far as I could and even then, it still stayed at or just above 70C. Now considering ASUS dropped this little hint yesterday, I'd say if you are worried at all about the stock fan not keeping up with a limited OCable card, this should do it if you want a single card you plan on using on air. https://twitter.com/ASUS_ROG/status/734368525133983744
|
|
#
¿
May 23, 2016 22:26
|
|
|
teh_Broseph posted:Some questionable numbers for benchmarks/temps/etc. on the 1080s, selling them before the 1070 info is released, and AMD stating they're targeting midrange for now...yee haw fresh die shrink, eh. The wait for Computex is painful. What adapter are you trying? A DVI to HDMI adapter should work perfectly for the Rift. DP on the other hand is more of a pain in the rear end to convert as unless you buy an Active Converter, you are stuck at around 1080P/60Hz with the passive ones (I can get 1080P/100Hz out of mine, but the Monitor hates it and tells me every time I power it on).
|
|
#
¿
May 24, 2016 16:08
|
|
|
teh_Broseph posted:Tried this one. Some more stuff from the other person with a 7970 in that Oculus thread, starts to get a little outside my experience level. That is, strange. You got a DVI to HDMI (Male) adapter? seems a bit odd. Usually you use a HDMI to DVI (Female) adapter and just plug the HDMI cable into that. ALSO, check that the cable is plugged in solid in the rift itself. The cable going to the HMD is removable and it may have gotten slightly loose somehow possibly. If that isn't the case, try another adapter if you can, or see if you can plug the Rift into any other system possibly. If you live near a Frys, pick up a GPU to "Try" as the no restocking fee makes it easy to test and return (Buy an open box/returned one to be less of a dick about it though). Though no promises if you get a good card, and end up wanting to keep it because VR. IF it happens with another adapter/card though, then your rift does sound boned and needs an RMA. Edit:READING that you both have 7970's with the same issue, if the adapters for DVI-HDMI don't work, then it does sound like the GPU's fault more than the Rift at this point... Weird for sure. EdEddnEddy fucked around with this message at 17:28 on May 24, 2016 |
|
#
¿
May 24, 2016 17:25
|
|
|
And SLI would be great, if VR SLI was a thing yet. Until then, I feel we are sort of stuck with waiting to see what the 1080Ti offers and how it performs in VR. Now if you aren't using VR at all. Then go hog wild. My 780 SLI setup could run anything I threw at them at 1440P ultra so SLI 980Ti's would be stupid good. Also if the 980Ti is roughly equal to a 1070, wouldn't that make them just a hair cheaper than a 1070 which would probably be around the $400+ range itself for a while?
|
|
#
¿
May 24, 2016 21:47
|
|
|
Josh Lyman posted:Doesn't SLI not work half the time anyway? SLI works fine in the past few years for most any games that aren't day 1 AAA launch titles. It isn't near as bad as it was back in the GTX500 series era at least. However, it doesn't work at all in VR currently where it would be massively beneficial currently.
|
|
#
¿
May 24, 2016 22:10
|
|
|
Subjunctive posted:I thought the card-per-eye stuff was released now. Is it out and ready from a driver standpoint, yes. Has anyone started using it yet? Not that I have seen unfortunately. I believe Valve just recently had a Release Note in one of the SteamVR updates that disabled the 2nd card if a user had a MultiGPU system. Games like Elite Dangerous benefited from it a ton with my 780's in Extended Mode, as did Asseto Corsa back then. Direct mode never worked in SLI however and I believe that remains true today still unfortunately. VR SLI was supposed to do the per eye instead of AFR I believe too which would have been neat, but also PCI-E 3.0 X16 is a must as well due to latency between the cards and all. Probably won't see it much at all until the 10XX series takes a good chunk of the market with its SLI Bridge of magic that might fix the SLI Latency that would depend on users having a X79/X99 system to do VR SLI. :/
|
|
#
¿
May 24, 2016 22:54
|
|
|
|
| # ¿ May 12, 2024 05:02 |
|
|
DrDork posted:Some of us have X99 systems and would really like to pick up 2x 980Ti's at firesale prices and get the best of all worlds, thanks. Hey my X79 loved it some SLI goodness with PCI-E 3.0 enabled. Though now I may have a problem fitting a 2nd ASUS Strix in there with the PCI-E USB 3.0 card I have in there now for the Oculus. drat ASMedia USB Host. If I could find another Strix for $400 or less and VR SLI starts to pick up steam, I might as well pick one of those up and wait out until Volta and some HBM2 goodness.
|
|
#
¿
May 24, 2016 23:43
|
|