Thursday, January 3rd 2013
AMD to Fix GCN Latency Issues with Driver Updates
Last month, an investigative report by The TechReport found out that despite being faster, AMD's Radeon HD 7950 graphics card isn't "smoother" than NVIDIA GeForce GTX 660 Ti, in that shows signs of higher frame-delivery latency, a theory proven by high-speed camera recordings. Over the holiday, AMD's David Baumann responded in discussions around the web talking about the issue, in which he put AMD's stand.
Apparently, AMD Catalyst drivers still have refinement left in working perfectly with GPUs based on the Graphics CoreNext (GCN) architecture. Baumann explained that GCN, and AMD's older Very Long Instruction Word (VLIW) architectures feature fundamentally different memory management, and drivers that make the most of it are still a work in progress. Baumann stated "Over the early part of the year you'll see a few driver updates help this across a variety of games."
He continued "Additionally, when we switched from the old VLIW architecture to the GCN core there was a significant updates to all parts of the driver was needed - although not really spoken about the entire memory management on GCN is different to prior GPU's and the initial software management for that was primarily driven by schedule and in the meantime we've been rewriting it again and we have discovered that the new version has also improved frame latency in a number of cases so we are accelerating the QA and implementation of that."
Source:
The TechReport
Apparently, AMD Catalyst drivers still have refinement left in working perfectly with GPUs based on the Graphics CoreNext (GCN) architecture. Baumann explained that GCN, and AMD's older Very Long Instruction Word (VLIW) architectures feature fundamentally different memory management, and drivers that make the most of it are still a work in progress. Baumann stated "Over the early part of the year you'll see a few driver updates help this across a variety of games."
He continued "Additionally, when we switched from the old VLIW architecture to the GCN core there was a significant updates to all parts of the driver was needed - although not really spoken about the entire memory management on GCN is different to prior GPU's and the initial software management for that was primarily driven by schedule and in the meantime we've been rewriting it again and we have discovered that the new version has also improved frame latency in a number of cases so we are accelerating the QA and implementation of that."
95 Comments on AMD to Fix GCN Latency Issues with Driver Updates
Nvidia 600 series
forums.geforce.com/default/board/34/geforce-600-series/
AMD 7 series
forums.amd.com/game/categorie...454&entercat=y
Granted I get higher framerates with a 7970 oc'd. You can use FRAFS bench viewer to make these graphs using frametimes from FRAPS.
Also, the graph doesn't lie however my experience with this card and any game have been a good one and I don't find "smoothness" to be any different from using a Nvidia card.
Get FRAFS bench viewer here: sourceforge.net/projects/frafsbenchview/
No problem with my 7970 IceQ His, it never heats and yet I also amd fx 8350 and at all as I overclock to 4.6 ghz without any problems and does not heat up that much even with the fan original, the only problem AMD chips currently have drivers ...
I looked at the youtube latency thing and saw nothing that I can discern as micro stutter.
If so both cards had it and none of them seemed to be worse than the other.
Plus Skyrim does have a micro stutter on Windows 7 even when I use an HD4850 if I don't use windowed mode with SWB mod.
Either way AMD see it to be a serious enough issue and enough to work on it.
Maybe this will resolve the more significant DX9c issue which some people are plagued with still.
Maybe it's certain monitors which bring out the problem more than others?
It's due to VGA memory management. It's even worse in crossfire, because the effect doubles, leaving what appears to be worse performance than a single card, even though the actual frame rate is higher.
And frankly, this problem is present in ANY high-end card from AMD, since 2900XT. I can't comment on anything other than high-end cards, because that's all that I have used.
What is even funnier, is that I can run 5760x1200(6912000 pixels) far easier than 2560x1600 (4096000 pixels). This was the key indicator that memory management, to me, was at fault, but since I've brought up this issue for years now, and none have seemed to care, I haven't reported this fact to anyone but a few users here.
How is it possible, that rendering 50% more pixels in Eyefinity performs better than 50% less pixels on a single monitor?
:wtf:
There's more to this than anyone wants to admit, unfortunately.
Truly, I dunno know WTF is the problem, really, but there HUGE issues here. I almost bought a GTX690, because I'm sick of AMD and their issues as well, but I feel I need I need GTX690 performance because of problems with performance on AMD cards.
What's more interesting is that system memory speed seems to smooth things out a wee bit for me. Like 2400 MHz on IVB and then some.
That's not so bad then and helps to explain why AMD didn't bother fixing it for a year. Still naughty of them though, tsk.
When running a single GPU it goes up to 80%-90% and it "seems" smooth.
Still can't figure out why quadCFX doesn't work on Skyrim but this doesn't concern this topic.