Haswell vs Kaveri

I'm not saying it doesn't include an integrated southbridge, just that this doesn't bring much support to the idea. It's also not clear whether Toronto features an on-die southbridge or just an on-package one. And it's not certain that Toronto == Carrizo, though I admit it's very likely.

What's odd is that other leaks point to FM2+ compatibility for Carrizo, and I'm not sure how that could work if the southbridge is integrated.

 

http://www.phoronix.com/scan.php?page=news_item&px=MTYxMDg

 

Intel 10.18.10.3412 drivers for Haswell and Ivy Bridge released.

32bit
https://downloadcenter.intel.com/Detail_Desc.aspx?agr=Y&DwnldID=23638

64bit
https://downloadcenter.intel.com/Detail_Desc.aspx?agr=Y&DwnldID=23639

 

I realize this is going to sound strange come from me , but how do you notice when these new Intel drivers are released DSC? Is there a notification somewhere?

 

http://forums.laptopvideo2go.com/topic/24340-intel-gma-drivers/page-26

From this thread, Intel's Download Center search is useless as hell but this thread finds the latest Intel drivers when posted on Intel.com.

 

Thanks. These drivers work together with the amd catalyst 14.1 beta drivers. I.e. Both running together at the same time. My monitor has two hdmi inputs , I plug one into the mother board and the other into my hd 7770. I do this so I can use quick sync.
Every time amd releases a new driver it breaks the intel driver (code 43).

This is one advantage Kaveri has over Haswell. A single driver that works for both the igpu and the dgpu. For amd dgpu's at least.

 

That sounds suspicious... properly operating drivers should not be interfering with one another assuming Win8+.

 

I don't know why you would considering it's marketshare.

 

20% according to the Steam Hardware survey.
More than XP and Vista combined.

I run Windows 8.1 because of the upgrades that are under the hood, not so much for the "improved" user interface..

The problem started with Win8.1.
I'm guessing the Intel driver team know exactly what the problem is.
From the Intel forums
"Users with 3rd party cards installed alongside Intel graphics are experiencing various degrees of Code 31 and Code 43 issues. These are being actively investigated, and we are working with 3rd party vendors to resolve them. The latest we are aware of is that AMD* drivers are force-unloading our drivers, resulting in various issues such as loss of multiple-displays, brightness controls, etc."

I doubt AMD is doing anything deliberate here, however it highlights the advantage of an all AMD system over a NVIDIA/AMD gpu with an Intel "APU".
The question in my mind is , can AMD use this tight integration to gain further advantages. e.g. Put an r9 290x in an AMD system and get 20%+ better performance (due to HSA?)

It seems like a waste of silicon to disable the igpu in a Haswell.
There is no reason the Intel igpu couldn't be used as a physics accelerator for example while the Nvidia or amd gpu is used for the game graphics.
Or perhaps use Quick Sync to compress game video coming from an AMD graphics card like shadowplay?

 

Have you tried to re-install Intel driver any time you update your AMD driver?

 

Order of installation is important as well as which card is primary.
I've experienced similar issues on my system with Intel HD3000 and when it was set to be primary GPU both W7 and W8.1 installed all GPU drivers more or less correctly. Had one or two small issues with 2nd AMD R9 card but I blamed it on old and tired W7 I run.
Funnily enough once installed I can switch my primary adapter back to AMD cards and everything works fine

 

I remember when an IGP able to work at all next to a graphics card was a new feature, maybe at the same time non-Intel IGPs (in chipset) got able to do dual screen. But then Vista was incompatible with running two cards/GPU from different vendors (more accurately, all GPUs had to run the same driver).

It's nice that stuff like this works at all. (But hell, running different cards worked in Windows 9x). Regarding Intel, they need to work on laptops that may harbor a competitor's GPU, and preferably have some working kind of dynamic switching between them, that's a whole 'nother set of complications.

 

I find that kind of stuff pretty unlikely.
Though for game recording, e.g. FRAPS could specifically support using Quicksync.
I think you need some Lucidlogix software to easily use Quicksync, if the software is still available and updated. You wouldn't have to use a dummy output. (/edit : but maybe that lucidlogix stuff is not a very good idea)

 

Indeed it is usually that stuff that causes the issues. If games would simply allow users to select an adaptor instead of always taking the "default" it wouldn't really be necessary and all of this would be way cleaner.

 

A bit like trying to use Mantle which fails with any iGPU enabled.

 

http://www.phoronix.com/scan.php?page=news_item&px=MTYxNzE

 

https://communities.intel.com/thread/49322

https://downloadcenter.intel.com/Detail_Desc.aspx?agr=Y&DwnldID=23644

 

http://www.extremetech.com/gaming/1...bugged-introduces-game-breaking-frame-stutter

 

<sarcasm>I am so surprised that asymmetric multi-gpu has problems</sarcasm>

In all seriousness though, I don't get why anyone has any hope for this mode outside of "wanting to believe".

 

integrated GPGPGPGPU physics