LinkBack URL
About LinkBacksYeah but the differences in this example are pretty massive - check the screenshots from the third link in my post:
http://1.1m.yt/6voZccp13.jpg
http://3.1m.yt/bvOSUNPr1.jpg
Yeah but the differences in this example are pretty massive - check the screenshots from the third link in my post:
http://1.1m.yt/6voZccp13.jpg
http://3.1m.yt/bvOSUNPr1.jpg
That clearly shows high AF vs no AF, but could be any number of reasons why that users setup is like that - another user(1) shows default for 980ti that does apply AF correctly. Either there's a bug in the titanX (only, not 980ti) driver that's missing default AF settings or more likely the original user has some stuff hanging around in registry from a previous install and/or something else is setting it to no AF for him.Originally Posted by watercooled
1: http://forums.overclockers.co.uk/sho...&postcount=348
That's the point - the in-game settings are both identical but something is causing the game to behave weirdly with the TX. But as people in those threads have said it seems like there are LoD differences too, not just AF.
OFC it could just be a bug but it would be interesting to see if it's a potentially common one, especially if it happens on systems with many driver swaps e.g. review test benches as the lowered IQ also increases the FPS so could skew results.
And it's not the first time weird stuff has happened with IQ settings in Nvidia drivers: http://www.bsimracing.com/nvidia-ima...n-newer-cards/
It is more that one person though? So far I've seen one person with a TX have the problem, and one person with a 980ti not having a problem. That's not enough for me to say something is causing the game to behave weirdly with the TX - sample size way too low. And certainly it means you can't say anything about the drivers in general since the problem wasn't seen on the 980ti.
I didn't say otherwise, just said something funky was going on and it wouldn't hurt to check. However as I also said, if it's a bug more likely to occur on test benches (e.g. if re-installing drivers has something to do with it) it could have an impact on reviews.
Easy enough to check. If the titanX is performing much better than the 980ti (more than expected) then it would suggest the bug is there. So far the 980ti reviews seem to place them exactly where you expect.
That wouldn't prove one way or the other though as it may not be exclusive to the TX (other people have claimed no issue with their TX). The guy with the 980Ti probably hasn't been reinstalling drivers so might not have triggered the bug if that's indeed the cause.
Given other people see no issue with the TX, and no-one sees an issue with any other cards, wouldn't the most rational explanation therefore be there's no problem, and this one guy happens to have a problem with his particular setup/the way he did things?
It's quite possible and to reiterate, I never said any different. But since it's a discussion forum, where's the harm in confirming that?
It's like Google's bug tracker - a large number of people can post fairly major bugs but they get ignored and/or flagged as 'cannot reproduce' because one other guy couldn't get it to happen on his system. Just because they don't affect every single user, or even a large percentage of them, doesn't mean the bug doesn't exist.
So far after reading those threads I saw a large amount of speculation, one proof of it happening reliably on a TX after several driver reinstalls, and one proof each of it not happening on a 980Ti and a Fury X. As I've said all along, it would be interesting to know what is causing it and if it's perhaps more common in certain configurations.
In the other unrelated link I posted, it shows different image quality with the same in-game settings on different generations of Nvidia cards, so it's not the first time something like this has happened in drivers.
Plus, I followed bugs like the 840/Evo first popping up on forums with some users claiming it was all some conspiracy theory until more people started noticing it. And then others started testing for it until it became obvious there was a major problem. The TX thing is new, and indeed it could be a one-off but bringing it to people's attention helps figure out what it is, one way or the other. If there is a problem, just ignoring it would be harmful.
Last edited by watercooled; 06-07-2015 at 06:11 PM.
Glad to see the 300 series driver features have been added to existing cards in the latest drivers. As I said the move didn't make sense IMO unless there was a technical reason; I guess it's just that they were added to the 300 release drivers before the next full release.
It would be interesting to see the framerate limiter compared between 200 and 300 series to see if there are differences in implementation.
This video shows the 280X getting substantially better performance in project cars in Win 10: https://www.youtube.com/watch?v=XzFe5OOHZko
It will be interesting to see how it affects performance of other games/cards too, so I hope we see sites like Hexus doing some comparisons!
A cheap & fast 22nm SOI process (the ST Microelectronics one) is available at GlobalFoundaries https://www.semiwiki.com/forum/conte...-its-true.html
Sounds like it would make nice AM2/FM2 style chips. Not what we want for the next FX chip though.
It's interesting that an FD-SOI process has finally been formally announced by a foundry. However it doesn't look like there's a 'high performance' version of it yet and risk production in H2 2016.
Last edited by watercooled; 14-07-2015 at 03:08 PM.
I believe STM have it in production, albeit the capacity isn't that big.
Interestingly, while the slide deck doesn't once mention the mainstream PC market or even tablets, in the market segment slide the arrow for 22FDX creeps just over the boundary between "Wired Networking/Consumer Applications/Mid-Range Smartphone" and "High End Mobile Application Processor", overlapping with the low end of the 28HPP node. Reading between the lines, it would look like there's scope for a low end x86 processor in the capabilities of the node...
EDIT: also, haven't GloFo already announced a 14(16?)nm FinFet node?
Yeah GloFo are licensing Samsung's 14nm FinFET process and if some rumours are to be believed may already be running some capacity for Samsung.
However the current shipping nodes are 14LPE whereas high performance stuff like Zen would most likely wait for 14LPP. As far as I know though, AMD still haven't explicitly said where they're producing their next processors, only that it's a FinFET node.
There are currently 2 users browsing this thread. (0 members and 2 guests)
Posting Permissions
Reply With Quote