Problems? See this thread at archive.org.
Overheat badly. Possibly catch fire (HP).
They won't drop by more than about $100 from what we currently have, and even that is going to be a rather surprising occurrence.
Doesn't matter, really.
In the respect you're talking about, it is overblown, and nVidia has mucked up MORE than AMD in the last two years, especially with almost every driver between 350.12 (GTA V driver in february 2015) and 365.19 (driver of a few months ago). However... the "driver thing" isn't overblown at all for this other reason:
AMD's directX 11 performance is atrocious. Every single GCN card in DX11 is crippled right now.
A couple years ago, nVidia basically got fed up with Mantle and removed as much CPU overhead from their DX11 drivers as possible. A lot of games got huge boosts in performance, most notably at the time Thief, which was a mantle game where AMD was beating nVidia with it. Suddenly, DX11 nVidia was rivaling and beating Mantle AMD, and DX11 AMD could not keep up.
AMD has yet to do this, or anything close to this. You've noticed how AMD gets ~30% performance boosts in DX12 over DX11, and FAR more in Vulkan over OpenGL, correct? And how nVidia has generally performed worse in DX12/Vulkan over DX11/OGL? Well, that's because of a few things.
- DirectX 12 and Vulkan "allow" developers the "ability" to bake in optimizations into video games directly, like they do with a console. They are not "required" to do it, and they do not, because time/money & no profit benefits from doing so.
- DX12 and Vulkan can use driver-side optimizations for performance, just like DX11/DX9/OGL/etc. This is what developers are doing. This means that the ONLY benefit from DX12/Vulkan they receive is lower CPU load, and, you guessed it, no CPU overhead in the driver.
- Since driver-side optimizations are needed, immature DX12/Vulkan drivers from both parties result in lowered performance than DirectX 11/OGL... or so you'd think. For nVidia, this was the case. DX11 was more optimized as it was much more mature. AMD on the other hand... even with immature optimization in the driver, the simple act of killing the CPU overhead instantly boosts the performance of the cards that much.
This means that AMD's drivers are so bad that their cards could easily be performing 30 to 40% better RIGHT NOW. Every. Single. GCN. Card. From the 7970M to the RX 480. All of them. Could be 30-40% faster in the absolute majority of titles most people play right now... if AMD's drivers didn't suck that badly.
So yes, I do agree that AMD's drivers are awful. It's just people don't generally talk about them, and they're happy to take at face value that "DX12 performance is better with AMD". But if everybody understood that what they're seeing is the performance they should have now, they wouldn't really be celebrating at all.
-
-
Thanks a lot. I may buy a cheap BGA Ultrabook for work, and a 17.3" monster for play. I'm leaning GT73VR or P775DM3 as luggable DTR. I still have to figure out the money though
Sent from my SM-G935F using Tapatalk -
-
-
-
-
TWENTY TIMES. I HAD TO REFRESH THE MAIN PAGE 20 TIMES BEFORE IT WOULD LOAD PROPERLY AND TELL ME I HAD ALERTS.
Where do I find tapatalk I give up with this flipping forum
Vsync on? No. Temperatures? Only if sitting at main menu with unlocked FPS, and only on GPU.
CPU... eh. Maybe.
Evolve Stage 2 now... that would eviscerate most of these laptops. It's by far the hottest game I've played in terms of pushing my whole system.
Ironically, 7 days to die unlocked FPS is even worse on my single GPU than anything else too.
Yeah, those machines you can get a bit far depending on what you're doing, but it's still only for light loads. Some machines are just that awfully designed. Would you like to see my sister's netbook? I could use my mouth and blow more air on the freaking heatsink than that pathetic fan does.
![[IMG]](images/storyImages/CoowRjTUsAAKWqz.jpg)
IN FACT. SHOVING A FAN NEXT TO OR BLOWING A CANNED AIR INTO THE EXHAUST VENT PUSHES IN SO MUCH AIR THAT IT REDUCES THE TEMPERATURES. THE. EXHAUST. VENT. NOT THE INTAKE VENT.
THE.
*EXHAUST*
VENT.Ashtrix, ThePerfectStorm, TomJGX and 3 others like this. -
-
-
-
-
I have actual TDP data as well, and while I agree wccftech gets a lot of stuff wrong, I need to anchor my discussion to something other than NDA-protected data
Point is, everything I have seen still contradicts you, in that the 1060 is not hotter than a 980M. If there was some concrete data to refute that, I would be interested in seeing it!
I think most everyone is aware of Gigabyte's shortcomings in cooling, they are the worst offender, and I think picking them out as representative of all ODMs is also a little misleading. That video was pretty interesting though: one fan, just 3 heatpipes, for the CPU and GPU, and the guy says there is no throttling. The CPU maxed in the low 80s, the GPU was in the early 90s but did not throttle (presumably didn't do much boosting either, though). Given that one fan and 3 heatpipes can pull that off and most other designs are better, I feel like you are painting the OEM/ODM situation to be worse than it is in general.
I spent a little time on the Aorus forums recently because I'm scoping out their X3/X5 and some people did report that they actually fixed throttling with a re-paste, because their machines went from 90s to 80s. It will definitely be case-by-case, and this was without undervolting. If the machine is just on the cusp of throttling, a re-paste can take you over to the other side.hmscott likes this. -
TDP. Does. Not. Matter. The problem with Pascal is thermal density. Huge number of cores and high clockspeed on tiny die sizes. It doesn't matter how much power the cards draw. The desktop GTX 1080 is a 180W card and uses the 250W TDP-rated vapor-chamber cooling system that was on the Maxwell GTX Titan X, and the cooler performs worse on the 1080 than it does on the Titan X. That alone should explain that TDP means nothing. As for "proof" that the 1060 is hotter than a 980M, everybody here but you seems to realize this. And I gave it to you already. The Clevo P650RG houses a 980M:
The P650RP6 houses a 1060:
Their cooling is the same design, correct? The 980M in the RG runs cooler when overclocked than the 1060 does at stock in the RP6. The 1060 is a hotter card.![[IMG]](images/storyImages/xmg_p507_1.jpg)
Well firstly, GPU-Z was showing all sorts of throttling under the perfCap reasons. Green and a magenta one that I don't even understand where it came from. Also, GPU load under the "high clockspeed" portions were often well under 100%, and fluctuating, indicating some other limitation somewhere else. And the ENTIRE stretch of GPU-Z they showed in general had limit reasons shown. If it wasn't vREL it was "util" which means the game didn't require it clocking up any more due to lack of load etc, and then the purple/magenta-ish ones I don't even know what they are... because my cards never actually throttle. Years ago I saw green for the first time when maxwell came out and was amazed, because it doesn't happen on my system. Now there's magenta and I don't know what THAT is. But anything in "perfcap" when you want high performance is not good.
Next, that thing sat well above 90c the entire time. Its max was 95c. They "claimed" no throttling either, but the performance was lackluster. Octiceps' 980M in his P650SG gets better performance in the Deus Ex benchmark and the portion of DOOM they were in (he claims they were hitting 50fps often when he was at 65 minimum in the same spot). Note again that this card is supposed to be 30% faster than his 980M is. So whatever they're telling us, there was some pretty heavy throttle going on there.
At the end of it all though, the point I made was to show that these things are considered "acceptable" for OEMs to sell. You said how gigabyte is known for bad cooling. I agree. I also know repastes can help, but oftentimes not fix, the bad throttle issues. HOWEVER. Gigabyte is a huge company. They're charging $1800 for that machine in that video which for all intents and purposes is trash for a gamer, and it's "fine" for them to push that out. And they're a huge company. They make their own motherboards and are an AIB partner for both AMD and nVidia. The G1 gaming cards were considered cream of the crop for Maxwell on launch. They are no stranger to cooling, or making good systems. But this is what we get. Razer too, just tossed the 1060 into the same chassis that barely handled a 970M and called it a day. MSI added quite a bit of heatpipes to their GS series, but the 14" is a hazard to the other components, and the rest of them are still in the wilderness as far as I can find them now. Alienware is the only company I've seen appear to make an actual effort to get pascal in. Clevo has not had to change much on their models because... well... their maxwell models were already well-designed. In fact, I always felt that swapping the CPU and GPU cooling systems on their P6xx models was a great idea, because the one fan could cool maxwell better and the dual fan could handle haswell better, or an overclocked skylake better. But now... it's like they were waiting for Pascal to be a bloody sun in video card form. Suddenly the overcompensated GPU cooling is "sufficient" and even for the 1070 they felt a need to offload some cooling onto the CPU.
What you need to do man, is go hunting about Pascal temperatures. Also, feel free to ask @SRSR333 who felt the same way that Pascal "can't be that hot" and brought up the TDP discussion.
Yeah. If that's 1070 only, 60Hz only, and they expect people to play with vSync, then ok. It's not handling an unlocked 6820HK and 1070 very well is my judgement. There's not enough heatpipes and I'm sure it'll need a vapor chamber contact point to truly be effective in that scenario. High refresh rate gaming is *EXTREMELY* hot. Most of these things are 60Hz. There's a user on LTT with a GT72VR using a 1070. He/She was playing WoW gsynced/vsynced to 60Hz and ended up at 82c. Same for Overwatch. They weren't using max fans or throttling really, but that's still... atrocious. I asked them to use max fans via MSI's dragon center and test also with unlocked framerates and to see why their screen wasn't at 75Hz as the 1080p Gsync panels usually end up being, and I'll have some updates later. Unlike NBR where most people who comment here are in wait-and-see states, other forums and such usually have people who don't know, buy, then come asking information/advice. Then I get to treat them like guinea pigs and source much moar information =D.
likes guinea pigs and informationSirSaltsAlot, ajc9988, Ashtrix and 1 other person like this. -
Ditto. I'd be happy to have a 1070 inside a 14" and not really care if it throttled asking as it gave decent fps and more than a 1060 and 980. And besides I wouldn't expect a 1070 to max every game even without throttling so I'd be ok to tinker settings which we all do. Non of us settle for presets if we can't use a preset on max settings.
The 1060 is a lil below what I would like (says the guy using a 960m) whilst a 1070 far exceeds what I could of imagined I'd need.
hmmm. Actually a 1060 won't be TOO bad. That way I can't be spoilt downgrading laptops with weaker gpus like I did with the 970m vs 960m.
I can wait abit longer. Wanna see aw15 1070 prices. (Yes that contradicts what I wrote above) or clevos 14" with a 1060. I'd think it's better money spent spending cheaper than say msi as good as it is. I was expetting to pay those prices with a 1070...at least in the uk anyway.
Sent from my iPhone using TapatalkPMF likes this. -
*Official* nVidia GTX 10xx Series notebook discussion thread
Discussion in 'Gaming (Software and Graphics Cards)' started by Orgrimm, Aug 15, 2016.
