|
What are acceptable temperatures for a GPU, specifically a 6950? Mine is unlocked, and HWiNFO shows 4 different GPU temps. After a bit of Hawken they are "GPU Thermal Diode (71C)", "GPU TS0 DispIO (72C)", "GPU TS1 MemIO (74.5C)", and "GPU TS2 Shader (77.5)". Which one should I be paying attention to?
|
#
?
Oct 1, 2012 21:13
|
|
|
|
| # ? May 30, 2024 00:10 |
|
|
Mid 70s are perfectly fine for GPU temperatures. The hottest recent GPU (GTX480) regularly ran up to 95 degrees C. Most cards have a hard shutoff around 105-120 though.
|
|
#
?
Oct 1, 2012 21:58
|
|
|
Nvidia is going to be PISSED. Kepler's manufacturing standards require that either you leave the Boost clock voltage control entirely stock and reference, or that you circumvent the system entirely with a fixed-voltage, hardware-based solution. This is neither; it games the hardware in order to influence the Boost logic, which is intended to be verboten. MSI is probably lucky that all Nvidia seems to be doing is enforcing the standard and making them undo the gimmicking.
|
|
#
?
Oct 1, 2012 22:09
|
|
|
Agreed posted:Thing is, I'd have said that about MSI until today. Unless of course we're talking about OCZ. That's so many failed SSDs. 
|
|
#
?
Oct 1, 2012 22:31
|
|
|
Oh god, I just got that MSI 660 Ti card the other day. Should I RMA it or what should I do? 
|
|
#
?
Oct 2, 2012 00:27
|
|
|
Zorato posted:Oh god, I just got that MSI 660 Ti card the other day. Should I RMA it or what should I do? I bought that 660ti card yesterday, I dunno what I'm doing either.
|
|
#
?
Oct 2, 2012 02:27
|
|
|
If it's the specific model, I'd return it to the retailer for an exchange, if possible. If it's not the specific model, maybe you'll be okay?
|
|
#
?
Oct 2, 2012 03:47
|
|
|
Kramjacks posted:So apparently MSI was overvolting their GTX 660 Ti and 670 Power Edition cards, which gave them performance gains but also caused some systems to fail to post or get black screens after a change in load. Bugger me, I just ordered one too. Guess I'll be swapping it for something else when it arrives.
|
|
#
?
Oct 2, 2012 04:18
|
|
|
I honestly have to wonder why MSI did that. From a warranty perspective they had to have realized that they'd be burning out cards left and right. The end cost to them is going to be far higher replacing all the burnt out cards than they'll make in profit. It's just utterly stupid. Goon Matchmaker fucked around with this message at 17:29 on Oct 2, 2012 |
|
#
?
Oct 2, 2012 17:21
|
|
|
I wonder what is being powered by the 9V, no way is that powering the GPU? Hope they don't find similar in AMD MSI cards or I'm SOL. e- I've been trying to get a response out of MSI re: my 7850 fan spinning up and they are just ignoring the messages. Compared to my experience with Corsair and Lian Li support (both awesome) I have to say MSI are poo poo at responding and it will be on my mind when I buy another piece of hardware. GRINDCORE MEGGIDO fucked around with this message at 17:40 on Oct 2, 2012 |
|
#
?
Oct 2, 2012 17:30
|
|
|
Goon Matchmaker posted:I honestly have to wonder why MSI did that. From a warranty perspective they had to have realized that they'd be burning out cards left and right. The end cost to them is going to be far higher replacing all the burnt out cards than they'll make in profit. These kinds of things are usually done by a single person or small group with something to gain in the short term, drat the long term consequences.
|
|
#
?
Oct 2, 2012 17:32
|
|
|
Goon Matchmaker posted:I honestly have to wonder why MSI did that. From a warranty perspective they had to have realized that they'd be burning out cards left and right. The end cost to them is going to be far higher replacing all the burnt out cards than they'll make in profit. See also: AMD Trinity APUs, 32nm process with a 1.45V Vcore... stock, plus Turbo. You don't make any money if you don't sell any in the first place. And many turbonerds who'll buy the juiced up stuff are also likely to upgrade sooner, so they'd never notice.
|
|
#
?
Oct 2, 2012 17:41
|
|
|
A note for anyone using Firefox with AMD videocards: the Catalyst 12.8 drivers have been blocklisted due to stability issues. Use the Catalyst 12.9 beta drivers instead.
|
|
#
?
Oct 4, 2012 19:25
|
|
|
AMD/AMD partners reduce HD7850 price again, just as Nvidia is preparing to release GTX 650 Ti, which is half of the GTX680(though on GK106 GPU): 768 shader cores, 128-bit memory bus and 1 gb GDDR5. Here are leaked performance results from Korea, which show that 650 Ti is clearly slower than HD7850. Not sure how much the price difference will be (and will Nvidia lower GTX 660 price), but definitely AMD wants the customer to fork up a little more cash than 650 Ti costs and gain a lot of performance. The memory bus might be the biggest problem as it has 2x the shaders of GTX 650 but no significant memory boost. Rosoboronexport fucked around with this message at 09:09 on Oct 8, 2012 |
|
#
?
Oct 8, 2012 09:07
|
|
|
Rosoboronexport posted:AMD/AMD partners reduce HD7850 price again, just as Nvidia is preparing to release GTX 650 Ti, which is half of the GTX680(though on GK106 GPU): 768 shader cores, 128-bit memory bus and 1 gb GDDR5. Here are leaked performance results from Korea, which show that 650 Ti is clearly slower than HD7850. Not sure how much the price difference will be (and will Nvidia lower GTX 660 price), but definitely AMD wants the customer to fork up a little more cash than 650 Ti costs and gain a lot of performance. That performance is aggressively middle of the ground. I hope it comes out at 150 or 140. Still, that means a better card is within reach. quote:The memory bus might be the biggest problem as it has 2x the shaders of GTX 650 but no significant memory boost. This seems to be a running theme now, does anyone know why?
|
|
#
?
Oct 8, 2012 09:20
|
|
|
In practice, memory bandwidth primarily effects resolutions and AA levels one can run at without bottlenecking. Perhaps Nvidia is subtly trying to position cards so that a pair of cheap cards in SLI is no longer as competitive with a single beefed up card at higher resolutions. For example, a pair of GeForce 460s was competitive with a 570/580 at the same resolutions while costing significantly less, so you could use them in a 1080p or 1440p setup without shelling out the extra $100-$200. By having cards with greater shader power but a gimpy ability to raise resolution and AA, it lets you use SLI as an upgrade to eye candy (e.g. Ultra detail instead of Medium) and framerate at a fixed resolution while maintaining an incentive to buy a higher-margin product.
|
|
#
?
Oct 8, 2012 10:11
|
|
|
Rosoboronexport posted:AMD/AMD partners reduce HD7850 price again, just as Nvidia is preparing to release GTX 650 Ti, which is half of the GTX680(though on GK106 GPU): 768 shader cores, 128-bit memory bus and 1 gb GDDR5. Here are leaked performance results from Korea, which show that 650 Ti is clearly slower than HD7850. Not sure how much the price difference will be (and will Nvidia lower GTX 660 price), but definitely AMD wants the customer to fork up a little more cash than 650 Ti costs and gain a lot of performance. http://wccftech.com/nvidia-geforce-gtx-650-ti-3d-ma/ So for $150 I get the 5850 class performance that I got for $150 1.5 years ago? That's progress from Nvidia! 
Palladium fucked around with this message at 11:37 on Oct 8, 2012 |
|
#
?
Oct 8, 2012 11:33
|
|
|
The GTX 650 Ti really should be positioned against the HD 7770, if nVidia isn't pricing the card appropriately that is a bit ridiculous.
|
|
#
?
Oct 8, 2012 13:50
|
|
|
Alereon posted:The GTX 650 Ti really should be positioned against the HD 7770, if nVidia isn't pricing the card appropriately that is a bit ridiculous. Let's not even consider a 7850 1GB can be had for $160 that wipes the floor with the 650 Ti, its just hilarious to see it barely edging out the GTX 460 that has been out for like what...2 years? The slightly faster 5850 has already celebrated its third birthday. Both of which can be bought for less than $150 a long time ago while they were still available. I can only conclude this card is Nvidia's idea of a sick joke. Palladium fucked around with this message at 15:42 on Oct 8, 2012 |
|
#
?
Oct 8, 2012 15:40
|
|
|
Eh, cut them some slack. They're running a very lean generation as far as fabbing goes, and making a lot of money in the discrete mobile GPU market while also offering a hell of a card selection for performance-oriented customers who prefer team green. Everything in the middle kinda sucks, but I'd say on the whole they're doing fine. They have every incentive with all the OEM ties they've got to make cards like the 650, 650Ti, etc., and no disincentives, so of course they'll do it even though it's silly from our highly informed, do-it-yourself mindset. Think of where these are going to end up and it makes more sense why they exist as products.
|
|
#
?
Oct 8, 2012 19:22
|
|
|
Kramjacks posted:So apparently MSI was overvolting their GTX 660 Ti and 670 Power Edition cards, which gave them performance gains but also caused some systems to fail to post or get black screens after a change in load. That circuit looks really weird, any other EEs floating around? I don't know where they sourced the diagram from, but it looks more like a design error than anything else IMHO. AC coupling the anode of a Zener regulator?  If anything, that cap would be on the output of the regulator (between cathode and anode). Maybe that sheet in Allegro/whatever was really messy and they missed it.e: looking at the datasheet of the RT8802A, I don't see what pin would benefit purposefully from getting >5V. I could see trying to game SS or a compensation network to get a more aggressive response (Type 2?) but that wouldn't make sense here. I'd hold off on flaying MSI for now, though it still is a bit odd that something this simple (and from the reference design apparently) made it through. movax fucked around with this message at 21:05 on Oct 8, 2012 |
|
#
?
Oct 8, 2012 20:51
|
|
|
Wozbo posted:http://www.xbitlabs.com/news/cpu/display/20110119204601_Nvidia_Maxwell_Graphics_Processors_to_Have_Integrated_ARM_General_Purpose_Cores.html That's pretty slick, not to mention clever and relatively cheap. You get the netlist for the ARM core from ARM, free to implement it on your chosen process, and make whatever changes you deem fit. Since it's ARM, I could see them leveraging AXI to create an interconnect between the ARM cores and their logic, or their own high-performance bus. It would have been pretty  if they just threw ARM cores on there and left the usage up to software developers, but having them autonomously (presumably) control GPU functionality is pretty neat. Maybe instead of drivers executing a fuckton of MMIO and hitting registers, it'll be some kind of "dispatch workqueue" thing. (Just guessing, I've only ever worked with Intel drivers).Maybe prettier textures will happen in-line with Maxwell release also, since the next-gen consoles should have more RAM & VRAM. Have to stress PCIe 3.0 bandwidth somehow!
|
|
#
?
Oct 8, 2012 21:23
|
|
|
movax posted:That's pretty slick, not to mention clever and relatively cheap. You get the netlist for the ARM core from ARM, free to implement it on your chosen process, and make whatever changes you deem fit. Since it's ARM, I could see them leveraging AXI to create an interconnect between the ARM cores and their logic, or their own high-performance bus.
|
|
#
?
Oct 9, 2012 01:02
|
|
|
Anandtech's Geforce GTX 650 Ti review is out. Leaks were correct: If you don't buy the Radeon HD 7850 (preferably 2GB) instead you are an idiot. If nVidia cuts $20 though it becomes a reasonable step up from the Radeon HD 7770, especially when overclocked. Interestingly enough, all tested cards overclocked to precisely identical settings.
|
|
#
?
Oct 10, 2012 04:05
|
|
|
Alereon posted:Anandtech's Geforce GTX 650 Ti review is out. Leaks were correct: If you don't buy the Radeon HD 7850 (preferably 2GB) instead you are an idiot. If nVidia cuts $20 though it becomes a reasonable step up from the Radeon HD 7770, especially when overclocked. Interestingly enough, all tested cards overclocked to precisely identical settings. http://www.pcper.com/reviews/Graphics-Cards/NVIDIA-GeForce-GTX-650-Ti-1GB-Review-GK106-Kepler-150 Well, the 650 Ti is a good 7770 competitor provided you want AND can get a free bundled copy of AC3 along with the card AND live with the so-so performance (a lot of ANDs there). although the clause of "check with AICs on participating bundles" strongly suggests the game won't be included in the baseline models, only the premium OC ones.
|
|
#
?
Oct 10, 2012 10:42
|
|
|
Alereon posted:The really interesting thing to me is that there actually are no ARM cores here. Project Denver is an implementation of Transmeta's Code Morphing technology to execute ARM code on a custom-designed nVidia core. The original plan was to execute both x86 and ARM on the same cores, but Intel successfully sued to block this x86 compatibility, arguing that the x86 license didn't transfer to nVidia when they acquired the corpse of Transmeta. Oops, I thought they were just licensing it, not rolling their own ARMvX-compatible core. Should be interesting times next year.
|
|
#
?
Oct 10, 2012 19:27
|
|
|
How does RMA-ing a card work if they no longer make the card? I returned a Sapphire 6950 I had bought last year for a fan problem, and doing a quick look on Newegg shows the card being deactivated. In the meantime, however:  
|
|
#
?
Oct 11, 2012 20:41
|
|
|
Endymion FRS MK1 posted:How does RMA-ing a card work if they no longer make the card? I returned a Sapphire 6950 I had bought last year for a fan problem, and doing a quick look on Newegg shows the card being deactivated. They'll probably send you a "better" one. Usually, this means the current-generation equivalent of where the 6950 used to live in the line-up. This can lead to hilarity in generation gaps where the newer card actually performs worse in certain games and conditions. Then again, I bought a single 6800GT from eVGA and my murderous PSU caused me to RMA for a 7900GT, followed by a 8800GTS 640, so that wasn't too bad. 
|
|
#
?
Oct 11, 2012 20:43
|
|
|
So it's possible I could end up with a 7850? I could definitely live with that.
|
|
#
?
Oct 11, 2012 21:01
|
|
|
Endymion FRS MK1 posted:So it's possible I could end up with a 7850? I could definitely live with that. Provided they actually give you a 7850. Anyway whatever happens do yourself and all of us a favor and don't settle for anything less than a 6950 or 7850 as a informed consumer.
|
|
#
?
Oct 12, 2012 13:58
|
|
|
Palladium posted:Provided they actually give you a 7850. Anyway whatever happens do yourself and all of us a favor and don't settle for anything less than a 6950 or 7850 as a informed consumer. I know, I didn't spend $250 last year for nothing. It's Sapphire, so I have faith that they will send me the correct equivalent.
|
|
#
?
Oct 12, 2012 16:08
|
|
|
movax posted:Then again, I bought a single 6800GT from eVGA and my murderous PSU caused me to RMA for a 7900GT, followed by a 8800GTS 640, so that wasn't too bad. Coincidentally, a similar replacement happened to me with XFX, and it highlights why I personally dislike XFX. I had a 7950 GT. Its VRAM conked, so I RMA'd it with XFX. They sent me an 8600 GTS. I complained to them about sending me a card with a 20% drop in performance and no video-in/video-out feature. They cherry-picked a review in which the Counter-Strike 1.6 performance was 30% greater. Then I pointed out the other ten benchmarks in the same review. So they swapped it for a 7900 GT, still with no video-in/video-out feature. But I figured it was the closest I would get from them. Then a driver update revealed a flaw in the card's BIOS, and any 3D acceleration (including Aero) would cause the monitor to look like this, tinted by background colors, within between 30 seconds and 30 minutes:  Occasionally, if you waited long enough, the screen would flip back to normal, but the 30s-30m timer would start again. XFX's response: "We're not RMA'ing that. Download Riva Tuner and overclock the VRAM by 1 MHz to make it stop." God help you if you updated your drivers months later and broke Riva Tuner or its autostart, because the patterning would start again well after you'd forgotten why it was there or how to solve it. Factory Factory fucked around with this message at 17:32 on Oct 12, 2012 |
|
#
?
Oct 12, 2012 17:30
|
|
|
Yeah, I've always been a little disappointed in XFX. I had a 4890 and noticed it was an extremely cut down version of the reference card (no volterra digital VRMs, uncooled) which did not overclock one bit. That said, your anecdote is a clusterfuck beyond belief. Why they couldn't send out a BIOS update is anyone's guess. Incompetence, most likely.
|
|
#
?
Oct 12, 2012 17:36
|
|
|
Factory Factory posted:XFX's response: "We're not RMA'ing that. Download Riva Tuner and overclock the VRAM by 1 MHz to make it stop." God help you if you updated your drivers months later and broke Riva Tuner or its autostart, because the patterning would start again well after you'd forgotten why it was there or how to solve it. That's terrible customer service. My experience with MSI was not in the same league of bad (although they did say to run Afterburner, which while it was running stopped the fans on my 7850 firing up every 20 seconds while it long idled - simply by stopping it going into long idle  ).e- that makes me wonder, has anyone ever had a good experience like that with a supplier? (reported fault, supplier fixed bios and sent it to the customer) GRINDCORE MEGGIDO fucked around with this message at 18:43 on Oct 12, 2012 |
|
#
?
Oct 12, 2012 18:36
|
|
|
Do SSD firmwares count? 'cause Intel patched up the 320 8MB bug pretty quick, and Crucial, well, sorta quickly got a fix out for some hangs on my C300 after a thread on their forums got to 100+ pages.
|
|
#
?
Oct 12, 2012 22:51
|
|
|
wipeout posted:e- that makes me wonder, has anyone ever had a good experience like that with a supplier? (reported fault, supplier fixed bios and sent it to the customer) 
future ghost fucked around with this message at 21:30 on Oct 13, 2012 |
|
#
?
Oct 13, 2012 21:23
|
|
|
Nothing to see here.
evilalien fucked around with this message at 21:52 on Oct 13, 2012 |
|
#
?
Oct 13, 2012 21:41
|
|
|
wipeout posted:That's terrible customer service. My experience with MSI was not in the same league of bad (although they did say to run Afterburner, which while it was running stopped the fans on my 7850 firing up every 20 seconds while it long idled - simply by stopping it going into long idle Powercolor was pretty good to me with my Radeon 9700pro way back in the day. The thing died, and after a ten minute conversation where I assured the Taiwanese phone rep that the "golden finger" had not broken off of the card, they overnighted me a new card with materials to ship it back in.
|
|
#
?
Oct 14, 2012 00:23
|
|
|
I am looking at Geforce 670's on Amazon. There is a $30 difference between the regular EVGA 670 and the FTW model, and a bigger jump to Superclocked. Does the factory overclock make a big difference on these cards? 1440p, Skyrim, BF3 etc
|
|
#
?
Oct 14, 2012 02:55
|
|
|
|
| # ? May 30, 2024 00:10 |
|
|
Animal posted:I am looking at Geforce 670's on Amazon. There is a $30 difference between the regular EVGA 670 and the FTW model, and a bigger jump to Superclocked. Does the factory overclock make a big difference on these cards? To make sure you're looking at the best prices, Newegg has the EVGA Geforce GTX 670 for $379.99-$20 MIR=$359.99 with free shipping.
|
|
#
?
Oct 14, 2012 03:56
|
|