[H] Yet again say SLI is smoother than crossfire

[Lepton87]

Sadly with NV strong brand and marketing NV does not have to release their high-end chips for the consumer market. Why would they if they can sell their second tier chips with much higher gross margins and still outsell AMD's cards. NV sold a hell of a lot more GTX680s then AMD sold 7970s despite being late. People on this forum don't acknowledge the fact that NV offered better value only briefly with limited supply and that was only in the US. In Europe GTX680 was always much more expensive then 7970, even at launch, and yet it sold tons more than 7970. That is the power of brand and marketing. If people continue to buy NV second tier chips for high-end prices we might never see high-end NV chip in a geforce card again. Even worse there is a chance that AMD goes bankrupt then I'm sure that we would never see 500mm2+ chip in a GF card, only sub 300mm2 like now.

 

[Zanovar]

Sadly with NV strong brand and marketing NV does not have to release their high-end chips for the consumer market. Why would they if they can sell their second tier chips with much higher gross margins and still outsell AMD's cards. NV sold a hell of a lot more GTX680s then AMD sold 7970s despite being late. People on this forum don't acknowledge the fact that NV offered better value only briefly with limited supply and that was only in the US. In Europe GTX680 was always much more expensive then 7970, even at launch, and yet it sold tons more than 7970. That is the power of brand and marketing. If people continue to buy NV second tier chips for high-end prices we might never see high-end NV chip in a geforce card again. Even worse there is a chance that AMD goes bankrupt then I'm sure that we would never see 500mm2+ chip in a GF card, only sub 300mm2 like now.

Erm, i think you are meandering a bit mate *laughs*.

 

[Lepton87]

Erm, i think you are meandering a bit mate *laughs*.

Yeah, it's off-topic I admit but I'm just angry that I can't buy a fast single GPU card because there is none and I don't want to deal with multi-gpu again. I never considered AMD to provide that, now I can't even count on NV

 

[Zanovar]

Yeah, it's off-topic I admit but I'm just angry that I can't buy a fast single GPU card because there is none and I don't want to deal with multi-gpu again.

fair enough,i see your point.maybe next round?

 

[Lepton87]

fair enough,i see your point.maybe next round?

I'm afraid I will have to wait for the next process shrink because I don't believe either AMD or NV will make 500mm2+ 28nm cards for their refresh, at least for the consumer market. So I'm in for a really long wait and in the meantime I will have to rely on AMD atrocious quad-fire CF support. I would settle for a single GPU card that once overclocked would trade blows with stock GTX690. Probably too much to ask on 28nm.

 

[supremor]

I think they would have released GK110 gaming parts had they been ready in time but I suppose it is plausible they saw they could match 79xx series Radeons with GK104 and get away with it which would allow them to make and sell more GK110 Tesla cards for much larger margins.

I don't really take interest in sales numbers but if you are correct I'm surprised NV sold that much more. The 79xx's after the first wave of price drops were very good cards for the money. I nearly went with 7950 CF on this rig but I heard of too many people having issues with CF+Eyefinity so I decided to roll the dice on GTX 670 SLI which I have been extremely happy with.
Single cards are irrelevant for triple monitor gaming in my opinion and probably always will be but with that said if I were in the market for a single card I'd be hard pressed to buy 670/680 over a 7950/7970 especially living outside the US where NV parts are always more expensive than respective AMD part's regardless of relative performance.

In any case I think we should stop derailing a microstutter thread :sneaky:

 

[jimhsu]

No SSD on review system. That alone probably explains much of the increased stutter (long frame times) on Skyrim outdoor scenes.

Skyrim and a lot of TES games really, really, want to be installed on a SSD. I get stuttering even with an SSD, but that's running full-bore ENB, highres textures, and more textures on top.

To me these tests essentially suggest that Nvidia somehow preloads better than AMD on certain tests.

 

[Deders]

I agree, we have been talking about microstutter for years.
Now I'm not trying to be nasty or mean or anything, but folks DO try and keep up with this question.
In all the years talking about this, when does anyone remember so much talk about Nvidia being smoother than AMD? Serious question.

I remember around the time ATI's 48X0 series' became very popular and were the cards recommended by most magazines (over in the UK at least), Multigpu setups weren't recommended due to problems people were having with microstutter and less than desirable performance issues.

I personally couldn't see what the fuss was about as I was getting pretty much 90% scaling on my 9800GTX+ SLI setup in most games and I didn't really notice any microstutter until I played Just Cause 2 with the Nvidia exclusive graphical features (which were only calculated on one GPU, meaning it took one card a little longer to render than the other) It took these publications up until the release of the 460 to catch up and say it was a viable option, my guess is that their experience with crossfire put them off without giving SLI a fair chance.

 

[3DVagabond]

Exactly, check 560ti and 5870.
http://www.anandtech.com/bench/Product/511?vs=547 If GF100 failed NV could increase GF104(114)clocks further and make some turbo shenanigans and have a very competitive product performance-wise from their MID-RANGE LINE. They would be competitive until Cayman release. There're further parallels between those chips, both have seriously crippled compute performance, because they are gaming chips, never meant to be their highest tier solution. If 7970 was 80-90% faster then 6970 instead of 45% with release drivers I'm pretty confident that we could now buy GK110 based GeForce cards. AMD was too conservative with 7970 because even 300mm2 NV mid-range chip could beat it at release.
There isn't a big performance gap between 5870 and 560ti. GTX460 was released with very low clocks and its shaders were cut-down. 560Ti shows us what GF104(114) can do when fully unleashed. Locked voltage control, cheap PCB it all screams 300$ card. Do you really think that NV planned their next high-end card to have the same memory bandwidth then its predecessor? Apart from that, NV already released their high-end chip in the form of GK110. So arguing that GK104 was always meant to be their high-end flagship card is from my point of view stupid.

UPDATE:

Yes, that's exactly what I'm saying but it only pertains to gaming performance, let's leave compute performance out of this.
560ti and 5870 there you have it, NV mid-range card pretty match performs on par with AMD high-end card. It already happened in the past so why do you think it's so improbable?

560ti almost ties 6950 which is a tiny little bit faster then 5870. So that's another example of NV mid-range chip performing on par with AMD high-end, the other example is 680 vs 7970. Although with 7970 GHz edition AMD managed to eke out a small win, but that's nothing major. With current drivers vanilla 7970 pretty much ties with 680. Even at release GTX680 was only about 5-8% on average faster then 7970 at 2560x1600 resolution. After OCing both cards 7970 was faster.

So, the next generation midrange (560ti) was as fast (more or less) as the previous generation's high end (5870). We are talking the same generation with the 680 and 7970. The comparison's not valid, I'm afraid.

 

[Lepton87]

So, the next generation midrange (560ti) was as fast (more or less) as the previous generation's high end (5870). We are talking the same generation with the 680 and 7970. The comparison's not valid, I'm afraid.

The comparison is valid, 560Ti was essentially the same chip that was in GTX460. It was just fully unlocked and optimized to draw less power. If they unlocked GF104 and turned up the clocks they would end up with the same performance albeit with more power use. Sadly you're unable to grasp that. They didn't do that because they had functional high-end card even sooner(gtx480) as opposed to a few months ago when GK104 launched. Hell, even the code name suggests it was supposed to be GF104(114) successor.

 

[szvwxcszxc]

Why argue on pointless subjects? You penis will not grow because your card is .00005% smoother. My opinion.

 

[The Alias]

The comparison is valid, 560Ti was essentially the same chip that was in GTX460. It was just fully unlocked and optimized to draw less power. If they unlocked GF104 and turned up the clocks they would end up with the same performance albeit with more power use. Sadly you're unable to grasp that. They didn't do that because they had functional high-end card even sooner(gtx480) as opposed to a few months ago when GK104 launched. Hell, even the code name suggests it was supposed to be GF104(114) successor.

ok regards to the 680, I think the 680 was always meant to be the 680 as the nvidia probably knew the yield rate for gk110 would be horrific . That being said I, like you, believe the 680/670 were cards that cut too many corners specifically in their ability to take voltage beyond 1.175v and their vrm design .

in regards to your claims about the 460, it was never meant to compete with the 5870 as it's transistor layout was horrible . Just because the 460 'could have' competed with the 5870 doesn't take away the fact it didn't . Nvidia rushed fermi gen1's design and paid the price for it with incomplete transistor and driver design . The sad thing is, despite fermi's horrid drivers, rushed design, and being 6 months late they still outsold the radeon 58xx series

 

[The Alias]

Why argue on pointless subjects? You penis will not grow because your card is .00005% smoother. My opinion.

The point of a forum is to express opinion . Why would you want to stifle that ?

 

[KCfromNC]

Is that what you're going to interpret my comments as? Have at it. You'll have fun conversing alone.
I'm saying EXACTLY what my words say. As I usually do.

Thanks for clearing that up for us. I feel better informed about nVidia products and/or services now.

 

[SirPauly]

Why argue on pointless subjects?.

TechReport and HardOCP may differ.

 

[KCfromNC]

Thanks. And that was with 12.3 drivers. So After the 12.11's It must be getting more pronounced if now TR is reporting on it.

How much worse did TR conclude that it has gotten with the 12.11 drivers compared to 12.3? Please point us to the review you're alluding to here. Thanks in advance.

 

[SirPauly]

Maybe not with the 12.3's but clearly feels different than the past investigation using the 12.7 driver set to conclude this:

The 99th percentile frame time, though, captures the impact of the Radeon's frame latency issues and suggests the GTX 660 Ti is easily the superior performer.

 

[Lepton87]

ok regards to the 680, I think the 680 was always meant to be the 680 as the nvidia probably knew the yield rate for gk110 would be horrific . That being said I, like you, believe the 680/670 were cards that cut too many corners specifically in their ability to take voltage beyond 1.175v and their vrm design .

in regards to your claims about the 460, it was never meant to compete with the 5870 as it's transistor layout was horrible.

You're right it wasn't meant to compete with 5870 because GTX480 was already out in the market, but if something was very wrong with GF100 and they couldn't release it, I'm sure they would take GF104 and then clock is as high as possible, unlock all its shaders and sell it as a direct competitor to 5870. The point is that it's not far-fetched that NV second tier card can compete with AMD's best card.

 

[The Alias]

You're right it wasn't meant to compete with 5870 because GTX480 was already out in the market, but if something was very wrong with GF100 and they couldn't release it, I'm sure they would take GF104 and then clock is as high as possible, unlock all its shaders and sell it as a direct competitor to 5870. The point is that it's not far-fetched that NV second tier card can compete with AMD's best card.

you don't understand the gtx 460 couldn't just be unlocked . If they did the card would have been unbearably hot because of the bad/rushed transistor design the 560ti was released because nvidia fixed their transistor design AFTER rushing the fermi gen1 to the market because they were 6 months late

 

[Lepton87]

you don't understand the gtx 460 couldn't just be unlocked . If they did the card would have been unbearably hot because of the bad/rushed transistor design the 560ti was released because nvidia fixed their transistor design AFTER rushing the fermi gen1 to the market because they were 6 months late

How do you know that it couldn't be shipped with all shaders functional?They did that to increase yield and not cannibalize their own GTX480 which was a very hot card and yet they released it. Some GTX460 could even be OCed to 1GHz and thermals weren't a problem, at that clocks it was already faster than stock 5870 even with its cut-down configuration. Someone should compare GTX460 and GTX560(GTX560 has the same configuration as the GTX460) at the same voltage and clocks and see how much they managed to improve thermals. It certainly wasn't a very big chasm you make it out to be.

Just to be clear I did have 5870 at the time because when I bought it neither GTX480 or GTX460 was released and GTX 285 was a pathetic competitor for 5870.

I find it hard to believe that NV purposefully didn't want to have the fastest single GPU on the market.

 

[Vesku]

Maybe not with the 12.3's but clearly feels different than the past investigation using the 12.7 driver set to conclude this:

My Skyrim Fraps runs (outdoors with some combat) at 1080P Ultra+FXAA Radeon 7950 Phenom II 1090T @ 3.6GHz does not match with TechReports experience at 2560x1440. Is it a system difference, resolution difference? HardOCP and Techreport need to do thorough follow ups otherwise it's value to readers is minimal.

 

[blastingcap]

FRAPs is insufficient to tell if something is smooth or not to the end user. I don't know why people keep posting it. Read this for why FRAPs is not enough: http://techreport.com/review/21516/inside-the-second-a-new-look-at-game-benchmarking/11

 

[Vesku]

Yes, I understand that blastingcap. But even within Techreport's FRAPs methodology they haven't been thorough enough, imo. I generally have a positive opinion of Techreport but if they don't do a timely follow up it will be disappointing.

There is another related thread that discusses more thorough testing methodology:

http://forums.anandtech.com/showthread.php?t=2287709

I have to say being sensitive to multi-gpu microstutter I was a bit surprised to see several tech sites discuss single GPU "smoothness". I've had no such experience with this 7950 which peaked my interest enough to see how my FRAPs output would compare.

Again, Techreport and HardOCP need to do thorough follow-ups, mixing of the subjective and objective shouldn't be done lightly.

 

[Skurge]

You're right it wasn't meant to compete with 5870 because GTX480 was already out in the market, but if something was very wrong with GF100 and they couldn't release it, I'm sure they would take GF104 and then clock is as high as possible, unlock all its shaders and sell it as a direct competitor to 5870. The point is that it's not far-fetched that NV second tier card can compete with AMD's best card.

http://www.xbitlabs.com/images/video/palit-geforce-gtx460-sonic-platinum-extreme-oc/p460_power.png

GTX460 clocked to 900mhz uses about as much power as a 5970 and barely keeps up a with a stock 5870. So I don't think it would have been a viable compatitor and today the 5870 is performing well and sometimes beating the 480. So if nvidia tried to take on the 5870 with an overclocked GF104 then it would have been similar to what AMD tried to do with the 2900XT.

 

[Lepton87]

http://www.xbitlabs.com/images/video/palit-geforce-gtx460-sonic-platinum-extreme-oc/p460_power.png

GTX460 clocked to 900mhz uses about as much power as a 5970 and barely keeps up a with a stock 5870. So I don't think it would have been a viable compatitor and today the 5870 is performing well and sometimes beating the 480. So if nvidia tried to take on the 5870 with an overclocked GF104 then it would have been similar to what AMD tried to do with the 2900XT.

But GTX460 does not have a fully functional die. Certainly they could make GF104 compete with 5870 at the expense of power consumption. But the same was true with GTX480, it consumed a hell of a lot more power and was only about 15% faster. It would be a very similar situation that we have now. They took their second tier die (GK104) clocked it as high as they could and make it tahiti competitor and it worked very well seeing NV sales and gross margins. It was all made possible by AMD clocking tahiti way too conservative, not to mention immature drivers, but that is expected for a completely new architecture. Right now most people think that GTX680 is the best single gpu card there is based on release reviews.

The only difference now is that NV has power consumption in check.