Originally Posted by RussianSensation
Hey Groove, what are you going to do when AMD stops buying you NV GPUs for dissecting them and testing their driver issues if Intel and NV merge
and subsequently bankrupt AMD? You'll be jobless! This is comedy gold.
Check out the credibility of ABT's journalism on the HD7970 pre-launch article:
AMD’s Upcoming HD 7970 Exposed – a Short-lived Video card?
"Most telling is that its performance is not that great except in DX11 games and especially in heavily tessellated scenarios – only about 25% faster than a GTX 580 overall."
It's just too bad the author of that article didn't scream how GTX680 was also underwhelming from NV and later on how GTX680 actually went to lose the single GPU performance crown.
I guess facts don't count at ABT or a highly disappointing videocard generation only applies to 1 camp and the other is forgiven because "they purposely delayed a 7 billion transistor chip" (despite making their corporate clients wait until November 2012 to start shipping those easy to manufacture K20/Xs).
I don't think anyone personally accused you of anything in this thread. If you glance at ABT posts, apparently AMD/ATI cards stuttered for a decade now and TR is the first site to expose this. I guess it's convenient to ignore all the previous TR testing where NV cards had worse stuttering?
I don't recall NV fanboys at ABT saying how GTX570 was a stutter-fest for BF3 or that GTX670/680 stuttered more than HD7900 in BF3. Again, using the same website that did HD7950 vs. 660Ti testing.
Remember all those threads on AnandTech every week from AMD users about how NV's GTX570/670 cards stutter in modern games....Ya, there were none because most of us realize all cards exhibit some form of stutter in the real world and NV is not immune to this depending on the game you are playing. Since it's not possible to include 100 games in a review, the choice of games yo u have in a review can change the conclusion if you only focus on the stuttering and not FPS. You can pick 10 games where NV would stutter the most. Although based on the forums, apparently NV users could live with that BF3 stutter or GTX570 users with MP3 stutter, since they didn't say a word all this time. Maybe the micro-stutter captured on a high speed camera at 120/240 fps is actually unnoticeable in the real world by 99% of gamers? Maybe this is a current driver issue that simply needs to be resolved and it is good that TR investigated this? When people start going on a tangent that AMD cards stuttered for years, their credibility/objectivity is somewhat undermined, wouldn't you say? Just a thought.
If NV users never noticed the stuttering in BF3, they sure gained some amazing eye sight that all of a sudden that they noticed stuttering on AMD cards all the time, in fact for a decade now.
Originally Posted by RussianSensation
I never disagreed with Tom's Hardware, HardOCPs or TechReport's findings that SLI felt smoother than CF. I think we kinda knew that a while back which is why you can see why so many here recommended GTX670/680 for SLI but HD7950/7970 for single GPUs with mods/MSAA/high res screens.
The point I am making is that NV cards had stutter, even single GPUs, if you see my post above. Insinuating that it was an AMD/ATI wide issue that went unnoticed for years is simply not telling the truth.
Ya, but it applies not just to GTX680, but GTX500 series too. That's my point. NV cards also stutter and it has little to do with "Kepler" drivers. According to the same methodology, GTX500 cards were stuttering like crazy in Batman and BF3 compared to HD7850/7870 or even HD6870.
Like I said you can pick 10 games that stutter less on NV or AMD, or it could just happen accidentally where 5 out of 7 games stutter more on 1 brand than the other. That's how games always worked, which is why you buy the GPU for the games you want to play. It's not "AMD cards stutter more". You can't just base that off 7 games and only from testing 2 GPUs (7950 / 660Ti). If you check TR's 7970Ghz review, 680 stuttered more there. So what is it? AMD cards stutter more or less? The answer is it depends entirely the games you are playing. What would you rather play Crysis 1 with frames going from 50-60 fps or Crysis 1 at a constant 30 fps? Would you sacrifice a bit of screen tearing/smoothness for faster frames in a FPS or would you rather play locked in at console frame rates? It depends on each gamer but it's not inconceivable that you'd have more micro-stutter at 240 fps camera in a GPU that fluctuates between 50-60 fps in Crysis 1. Would you then conclude playing at 30 fps in a shooter is more enjoyable than at near 60 fps with occasional drops to high 40s-low 50s?
None of this still explains the frames per second results for MOH:W or Skyrim at 2560x1440 w/ AA, which is a separate outlier with the review.
I was not surprised by these results when I read the reviews back then.
The 570 had only 1.25GB of VRAM, and the 560 Ti had only 1GB. Running out of VRAM usually causes stutter.
That's why I didn't say anything about it. All it served was to remind me that 1GB was getting old, really.
EDIT - I do realize that the 6870 had only 1GB of VRAM, but perhaps it had better memory management (as was the case with HD 4870 512MB versus the other 512MB cards from Nvidia at the time - in general, it appeared like as if the Radeon cards used slightly less memory than the Geforce cards), or it could have very well been the case with single-GPU microstuttering that was hardly an issue for the pre-"Never Settle" drivers and more of an issue with the older Fermi cards. I do remember wanting more research to be done on this.
Also, I already realized that it wasn't always the 570 that sucked:
(at the launch of 7970 GE):
It would've been nice if TR included GTX 580 to know for sure if the stutter didn't have anything to do with the game using more than 1.25GB of VRAM. In the majority of cases for me, running out of VRAM causes stuttering/hitching.