|
Anime Schoolgirl posted:If Pascal is what it's cranked up to be (which is still fairly likely given the node shrink despite apparent yield issues) you're probably going to see enthusiasts jettisoning 980tis going for around half of what they usually do now. Yeah I'm not making that mistake again. I'll be selling off mine weeks before the first 1070 equivalent release and picking up one of those and hope its good enough to hold me over to the 1080ti release. Unless AMD pulls through. Although I probably won't preemptively sell for that until I see some numbers.
|
# ? Jan 29, 2016 21:28 |
|
|
# ? Apr 29, 2024 15:52 |
|
THE DOG HOUSE posted:Yeah I'm not making that mistake again. I'll be selling off mine weeks before the first 1070 equivalent release and picking up one of those and hope its good enough to hold me over to the 1080ti release. Unless AMD pulls through. Although I probably won't preemptively sell for that until I see some numbers. IMHO, AMDs 470X needs to meet a 290X in capability for like ~150-160W and roughly similar TDP, the 480X needs to match a Nano and the 490X (possibly new Fury) Need to match the 295X2 for similar power and thermal envelope as a 290X. I kind of have my hopes up because what appears to be the R7 450X is performing like a 7870/ R9 270X while drawing power like a GTX 750. Will definitely hand me down a 290X if a 470X is as good or better for the overall better thermals and new uarch.
|
# ? Jan 29, 2016 21:55 |
|
Interesting take on ROTR CPU scaling at Guru3D. AMD GPU systems perform better with 1 core than 8 cores
|
# ? Jan 30, 2016 01:18 |
|
That would possibly explain why for Furies and Furies alone computerbase.de saw light stuttering.
|
# ? Jan 30, 2016 01:29 |
|
I wonder if similar results occur if you restrict the processor affinity of the game in Task Manager to one core/two cores/four cores.
|
# ? Jan 30, 2016 02:58 |
|
What's the best way to test if a GPU overclock is stable? My 980 Ti is showing some visual artifacting in Black Ops 3 but I can't really find any good guides on proper overclocking of a GPU.
|
# ? Jan 30, 2016 11:45 |
|
Unigine Heaven is a benchmark that can be run for hours and is moderately taxing.
|
# ? Jan 30, 2016 12:01 |
|
Rumours, rumours, get your rumours here ... (nvidia to
|
# ? Jan 30, 2016 15:58 |
|
If AMD doesn't get the fury x2 out soon its gonna be dead in the water.
|
# ? Jan 30, 2016 16:11 |
|
Don Lapre posted:If AMD doesn't get the fury x2 out soon its gonna be dead in the water. Pretty much consider it dead in the water as of now. Thing needed to be out months ago.
|
# ? Jan 30, 2016 16:20 |
|
mcbexx posted:Rumours, rumours, get your rumours here ... Honestly, the only nugget in there that isn't aiming for the moon is a GTX Titan X2, but Maxwell, specifically to counter Fury X2. SLI Titans should be able to run VR pretty well IMHO, at least first gen. Don Lapre posted:If AMD doesn't get the fury x2 out soon its gonna be dead in the water. I honestly see it maybe being dead already, a kind of 6990. Polaris is coming really quick, so even with a late 2016 release for the new Fury/490, no one is really going to want to touch it except VR nerds, and as I posited above, Nvidia is likely to use a dual GPU Maxwell solution, maybe like 980ti VRX2.
|
# ? Jan 30, 2016 16:42 |
|
repiv posted:Interesting take on ROTR CPU scaling at Guru3D. AMD GPU systems perform better with 1 core than 8 cores
|
# ? Jan 30, 2016 17:31 |
|
Fiji
|
# ? Jan 30, 2016 17:38 |
|
Malloc Voidstar posted:How in the world did AMD manage to do this Isn't it more likely to be bad coding on the part of the game designers, or is that ?
|
# ? Jan 30, 2016 17:38 |
|
Eletriarnation posted:Isn't it more likely to be bad coding on the part of the game designers, or is that ? I mean I guess the devs could have done it somehow, but I don't know how
|
# ? Jan 30, 2016 17:40 |
|
Based on other benchmarks, it's a Fiji only issue. I kind of want a Fury X as a collector's item, and that's looking increasingly viable.
|
# ? Jan 30, 2016 19:02 |
|
So they think Titan X2 will be Pascal? With a name like that it could easily be a dual-GPU card too. But I don't see a Maxwell HBM card launching, it seems like that would probably take a pretty substantial redesign to make it work. I'm also pretty skeptical that NVIDIA will be ready to launch Pascal that soon, but I guess we'll see. If they do I would expect it to be (a) the paperiest launch that you've ever seen, and (b) to cost a billionty dollars. Seriously I can't see that card launching at less than $1500 or $2000. FaustianQ posted:I honestly see it maybe being dead already, a kind of 6990. Polaris is coming really quick, so even with a late 2016 release for the new Fury/490, no one is really going to want to touch it except VR nerds, and as I posited above, Nvidia is likely to use a dual GPU Maxwell solution, maybe like 980ti VRX2. Yeah Fiji VR is way late at this point, and a Crossfire solution with 4 GB of VRAM is dumb. Like, it's the same amount they put out with the 295x2 a full 2 years ago. Launching a dual-GPU card like a year after the single cards always seemed stupid to me. At that point in time, everyone's hyped up for a new arch and will wait for the new single-GPU card that will be 90% as good with none of the wacky problems of CrossFire/SLI. That poo poo needs to be out like 1-3 months after the single cards.
|
# ? Jan 30, 2016 21:04 |
|
Actually speaking of the Fury X2 The way this is presented makes me think the Fury X2 isn't going to be marketed directly at consumers, hmmmm. AMD is claiming their new midrange (380 equivalent) card will be able to meet/beat minimum VR spec while beating a 970 for power and temp EmpyreanFlux fucked around with this message at 23:34 on Jan 30, 2016 |
# ? Jan 30, 2016 22:57 |
|
My 980Ti with a hybrid cooler and the fans running at VTOL volume is...90C. So something is not right there. I guess I'll try reseating and repasting. The gurgling probably means it's not circulating optimally either.
|
# ? Jan 30, 2016 23:38 |
|
My air cooled 980 Ti HOF does 60c max at 70% fan. You may be better if contacting them and see what they, the manufacturer, say on the matter.
|
# ? Jan 31, 2016 00:00 |
|
Subjunctive posted:My 980Ti with a hybrid cooler and the fans running at VTOL volume is...90C. So something is not right there. I guess I'll try reseating and repasting. The gurgling probably means it's not circulating optimally either. Does tell you to make sure the fan/rad is above the card?
|
# ? Jan 31, 2016 00:10 |
|
FaustianQ posted:Actually speaking of the Fury X2 I haven't listened to the presentation but the specs basically suggest what amounts to an air-cooled Fury Nano X2. Nano is 175W TDP, which is just about half of 375W. Throw in a bit extra for board electronics or whatever and there you go. Like the single-card Furies, it's possible that they are launching multiple versions aimed at various power levels/cooling solutions. Or it could just be a one-off. Or AMD could just have hosed up again and put out yet another product that's no faster than their last generation. quote:AMD is claiming their new midrange (380 equivalent) card will be able to meet/beat minimum VR spec while beating a 970 for power and temp Assuming that means that it beats a 970 (which is actually not supported if you are finely parsing the statement you gave) then that works out to a 36% improvement over their current lineup. Let's also remember the bullshit early Fury X benchmarks vs 980 Ti and the "overclocker's dream" thing too. I mean don't get me wrong, that's good, but you don't get a handjob for slightly edging out your competitor's last-generation card in power/temp/performance after there's been a node shrink. If NVIDIA just does a die shrink of Maxwell they'll likely achieve the same thing at even less power. I'm willing to give AMD a little bit of a pass because they're devoting resources to async shading/compute stuff that NVIDIA currently isn't. But they gotta get their lineup in order, their GPUs are starting to look about like their CPU lineup in terms of competitiveness. Paul MaudDib fucked around with this message at 00:23 on Jan 31, 2016 |
# ? Jan 31, 2016 00:20 |
|
Subjunctive posted:My 980Ti with a hybrid cooler and the fans running at VTOL volume is...90C. So something is not right there. I guess I'll try reseating and repasting. The gurgling probably means it's not circulating optimally either. There's absolutely no way that's right. Try mounting the rad above the card. If that doesn't help RMA.
|
# ? Jan 31, 2016 00:22 |
|
is it beating the fictional 970 that runs at 150w or is it beating the actual 970 that runs at 190w is the question because the latter just isn't that impressive
|
# ? Jan 31, 2016 00:29 |
|
Subjunctive posted:My 980Ti with a hybrid cooler and the fans running at VTOL volume is...90C. So something is not right there. I guess I'll try reseating and repasting. The gurgling probably means it's not circulating optimally either. Yeah ... That's just plain not working. I would be paranoid here and not even run it at all..
|
# ? Jan 31, 2016 00:55 |
|
xthetenth posted:Does tell you to make sure the fan/rad is above the card? The fans and rad are on the other side of the case blowing out, so I don't think it's recirculating hot air into the card.
|
# ? Jan 31, 2016 01:10 |
|
Subjunctive posted:The fans and rad are on the other side of the case blowing out, so I don't think it's recirculating hot air into the card. I think he means the rad in a position above location of the gpu. A lot of them have a particular orientation that they work best in.
|
# ? Jan 31, 2016 01:14 |
|
I wasn't getting excited guys, I was kind of being satirical because the only hard numbers they gave is a midrange price target (below what the 970 costs, so a price bracket of 200-260$), and the minimum spec for VR, which Roy says (implies) it'll be better than the 970 in every way. We also know that what appears to be their low end, a 450 was demoed for 1080p60 for Star Wars battlefront and consumed apparently 86W at the wall. The requirements for VR means that this hypothetical midrange card has 3 times the rendering power of the demoed card. Power scaling isn't perfect, so I want to say the next gen 470X pulls ~140-150W from the wall and beats the 290X, possibly not due to raw horsepower but having a better scheduler (less driver overhead) and better tessellation capability - zero driver over head is one of Rajas big goals and super necessary for VR. Either way, Raja Koduri is in charge now, especially regards to press releases, I'm expecting they're less likely to bullshit and only release on what internally they can guarantee. I want to say that released cards for June/July will be the Ellesmere LE, Ellesmere Pro, Ellesmere XT R7 440: 80$, 30W, R7 260 equivalent R7 450: 110$, 50W, R9 270 equivalent R7 450X: 130$, 60W, R9 270X equivalent Baffin LE, Baffin Pro, Baffin XT R7 460: 160$ 90W, R9 280 equivalent R9 470: 200$ 110W, R9 290 equivalent R9 470X: 230$ 140W, R9 290X equivalent for desktop. Well binned Baffin and Ellesmere will likely be M490X, M480, M470 and M450. Greenland/Polaris10/Vega will be released alongside or just after Zen desktop, Raja likely doesn't want another Fury debacle. I'm going to say that part of the hold up on Zen APUs is likely so they can meet market demand for Polaris GPUs first, not that the silicon can't be ready, and that they'll be using cut down Vega and Baffin parts. I'll put my crystal ball away.
|
# ? Jan 31, 2016 01:24 |
|
Fauxtool posted:I think he means the rad in a position above location of the gpu. A lot of them have a particular orientation that they work best in. Oh, I'll check. Probably not, I didn't think about that. (And I don't know how I would manage that, tbh.)
|
# ? Jan 31, 2016 01:28 |
|
It'd be hard for to fathom any sort of placement issue that would cause you to run at 90 degrees water cooled, presumably throttling, at 100% fan speed. There are more ideal and less ideal orientations but we're talking like 10 degrees or less. My guess is that water is not circulating correctly for whatever variety of reason - assuming you reseat it already. Leave the side of the case off and if its still doing the same thing something is just plain wrong. If orientation is an issue, say if there is excessive air in there and in the position its in now it its causing blocking flow, you can find that out by taking the radiator out and hand turning it over while doing its stress test (with gloves on..). But even if that "solved" the issue its still an unacceptable condition and I would consider that faulty.
|
# ? Jan 31, 2016 01:41 |
|
Fauxtool posted:I think he means the rad in a position above location of the gpu. A lot of them have a particular orientation that they work best in. Yep, some of them have a very definite preference about what they want up at the highest point of the overall loop, because otherwise for example, if the card with the pump in it is up top, that means any air in the loop collects in the pump, and running the pump dry is a definite not-goodness.
|
# ? Jan 31, 2016 04:20 |
|
FaustianQ posted:I wasn't getting excited guys, I was kind of being satirical because the only hard numbers they gave is a midrange price target (below what the 970 costs, so a price bracket of 200-260$), and the minimum spec for VR, which Roy says (implies) it'll be better than the 970 in every way. [citation needed]
|
# ? Jan 31, 2016 12:07 |
|
THE DOG HOUSE posted:Yeah I'm not making that mistake again. I'll be selling off mine weeks before the first 1070 equivalent release and picking up one of those and hope its good enough to hold me over to the 1080ti release. Unless AMD pulls through. Although I probably won't preemptively sell for that until I see some numbers. I was just going to get the *80 to replace my 780 but getting a *70 then *80 GTi might not be a bad idea.
|
# ? Jan 31, 2016 12:19 |
|
SwissArmyDruid posted:[citation needed] ??? Seems he clearly stated he was doing some tea leaf reading to me. What is wrong with that? Based on the information we have right now (which admittedly isn't much) his post isn't all that bad to me. If AMD pulls off those levels of performance with those prices and power usage they should do alright. Unless nV messes up somehow, either on pricing or performance, I don't see AMD clawing back huge amounts of dGPU marketshare though.
|
# ? Jan 31, 2016 14:08 |
|
xthetenth posted:Yep, some of them have a very definite preference about what they want up at the highest point of the overall loop, because otherwise for example, if the card with the pump in it is up top, that means any air in the loop collects in the pump, and running the pump dry is a definite not-goodness. I thought the whole point of AIO was not having to worry about this poo poo.
|
# ? Jan 31, 2016 16:54 |
Subjunctive posted:I thought the whole point of AIO was not having to worry about this poo poo. No? The point of AIOs is that they are much cheaper and easier than building a custom loop and they don't need to be drained and cleaned regularly. Pump & reservoir placement will always be a thing that you need to worry about because that's how physics works and no loop is 100% free of air bubbles.
|
|
# ? Jan 31, 2016 17:19 |
|
ijyt posted:I was just going to get the *80 to replace my 780 but getting a *70 then *80 GTi might not be a bad idea. It's what I did this time around. It was a pleasant progression.
|
# ? Jan 31, 2016 20:49 |
|
Subjunctive posted:I thought the whole point of AIO was not having to worry about this poo poo. AIO is so all you have to do is put the card in and screw the rad in rather than picking out the block, the rad, the fans, the barbs, the liquid, the reservoir, buying the tubing, cutting it, checking it for leaks and so on. THE DOG HOUSE posted:It's what I did this time around. It was a pleasant progression. Yeah, it's nickel and diming you by getting you to buy two cards, but the high midrange holds value well so you can flip it for not much price.
|
# ? Jan 31, 2016 21:27 |
|
But if it's AIO, how are you supposed to bleed bubbles out? This can't be the desired state of operation.
|
# ? Jan 31, 2016 21:38 |
|
|
# ? Apr 29, 2024 15:52 |
|
xthetenth posted:
With the free game codes, you actually get paid a little to do it :x Subjunctive posted:But if it's AIO, how are you supposed to bleed bubbles out? This can't be the desired state of operation. Its broken
|
# ? Jan 31, 2016 21:53 |