AMD: Sea Islands R1100 (8*** series) Speculation/ Rumour Thread
Silent_Buddha
Legend
If AMD are to be believed then a more general fix should be available in approximately two weeks that also applies to Dx10 and 11 titles. Although we won't know how effective or comprehensive it will be until it is out and can be tested.
Regards,
SB
Regards,
SB
There was no casual attitude. A problem was reported and is being fixed. It's simple. No one paid much attention to frame latency until The Tech Report showed how and why to do it.
The latency issue wasn't really there in many of these games back with the 12.3 and similar cat drivers. When I bought my 7950 I still had those older reviews in memory and remember that latency in theTechReport's reviews was very similar to Nvidia GK104 cards, only the 7900 series had more memory, better compute, and were priced better. I'm still very happy with my purchase as it was cheaper than anything I can get for that performance range and still is to this day. I do also believe that with future games the extra memory and compute performance will help. It's nice to see the issue is being fixed - it just seems odd that it didn't seem to exist before the 12.7+ performance drivers from the reviews I looked at...
No, the drivers Scott tested will be out in two weeks. The memory manager re-write won't be due for a while still. That's a Big Fucking Job(TM).
Basically, three specifically targeted games will improve, and perhaps a few others if the fixes aimed at the non BL2 games(BL2 was an app specific fix, sadly) are generalized enough.
SB is a little closer. Target for the driver Scott tested is still next week, we're just adding a different fix in that before release.
Many are asking (I've noticed in other forums) will there be any improvements (direct or residual) for 5000/6000 series cards. Or is this strictly 7000 series improvements.
Glad to be wrong!
Jawed
Legend
I have to say I'm exceptionally disappointed.
To be honest, my mind boggles at the lack of curiosity this demonstrates. NVidia got there quite a while a go and AMD had to be shoved in this direction.Tech Report said:
People have been doing frame-to-frame measurement far longer than Tech Report:
http://forum.beyond3d.com/showthread.php?t=57620
and people have been moaning about stuttering for years before that. The IHVs had all the tech to take this seriously and didn't even think to look.
Love_In_Rio
Veteran
wasn´t stuttering problem solved in the last AMD drivers?. Ok, i´ve just read not.
Last edited by a moderator:
I wouldn't be so sure Nvidia got there a while ago. What I would be sure about is if Nvidia knew they were good at something AMD was bad at you'd hear about it. So either AMD drivers were good until now or Nvidia just started paying attention.
I now remember the thread you linked, but I don't remember other discussions about frame to frame latency. The Tech Report has done a better job explaining the situation and coupled with them being a bigger site it gets people to pay attention.
Also, I've been in this industry for a while and I haven't noticed a lot of moaning about stuttering outside of multi-GPU. It's easy to measure average FPS and that's what consumers looked at to make purchase decisions so it's natural driver teams would focus on this metric.
This, exactly. Nvidia is paying attention.
All i want is Nvidia and AMD continue to pay attention to it, because, lets be honest, it can be really game specific and so, driver specific. and can change from a driver to another.
Now, im not sure it is a good idea to only test GPU's with that. or you will need do a new review of the hardware at each driver release. ( a new card could looking bad on a frametimes graph on the first release review and be a beast 1 months after, depending the games and drivers used. and vice-versa). Game developpers have too their work to do on it, we have allready see many situations where games was not really tested on some hardware/brands as they should be, lets hope it dont start to be a programm of TWITMP and Games Evolved. )
Last edited by a moderator:
UniversalTruth
Veteran
Hybrid 7970 -> Radeon HD 8970, Radeon HD 8950, Radeon HD 8870;
Hybrid -> Radeon HD 8870;
Xtreme III -> Radeon HD 8870;
Xtreme 7970 -> Radeon HD 8970, Radeon HD 8950, Radeon HD 8870;
Twin Turbo II -> Radeon HD 8870;
Plus mono -> Radeon HD 8870, Radeon HD 8760;
L2 Plus -> Radeon HD 8760, Radeon HD 8740;
S1 Plus -> Radeon HD 8870, Radeon HD 8760, Radeon HD 8740
Arctic shows the compatibility of its cooling systems with graphics cards AMD Radeon HD 8xxx
Hybrid -> Radeon HD 8870;
Xtreme III -> Radeon HD 8870;
Xtreme 7970 -> Radeon HD 8970, Radeon HD 8950, Radeon HD 8870;
Twin Turbo II -> Radeon HD 8870;
Plus mono -> Radeon HD 8870, Radeon HD 8760;
L2 Plus -> Radeon HD 8760, Radeon HD 8740;
S1 Plus -> Radeon HD 8870, Radeon HD 8760, Radeon HD 8740
Arctic shows the compatibility of its cooling systems with graphics cards AMD Radeon HD 8xxx
iMacmatician
Regular
I'm not exactly sure which one of three threads this report is most appropriate for…
According to Sweclockers: "The next generation GeForce and Radeon postponed to Q4 2013" (original), AMD and NVIDIA have delayed their next-gen GPUs until Q4 2013 at the earliest. AMD might release the 7990 in the meantime.
————————
If true I'm disappointed, I was hoping for at least a midyear release of the "real" 8000 and 700 series. I wonder if there will be a small clock bump to the 8970/8950 around June or so (or whenever the retail versions come out), similar to the 7970 GHz Edition.
According to Sweclockers: "The next generation GeForce and Radeon postponed to Q4 2013" (original), AMD and NVIDIA have delayed their next-gen GPUs until Q4 2013 at the earliest. AMD might release the 7990 in the meantime.
————————
If true I'm disappointed, I was hoping for at least a midyear release of the "real" 8000 and 700 series. I wonder if there will be a small clock bump to the 8970/8950 around June or so (or whenever the retail versions come out), similar to the 7970 GHz Edition.
This would probably belong in the "nonsense" thread.
I can't think of any valid reason for AMD or NVIDIA to do this.
boxleitnerb
Regular
To extend the life of 28nm since 20nm is terrible (regarding yields and wafercosts) and possibly not viable until H2 2014?
Similar threads
