The Spin off Smoothness Thread

[roofsniper]

Like I said, this whole thread is epic fail, in my opinion... Everytime the thread starter hits a brick wall on his arguments, he starts calling people trolls.

Talking about low FPS, you can check this very interesting article from many moons ago.

http://www.firingsquad.com/hardware/$500_gaming_pc_upgrade/page7.asp

The average FPS all looks similar until you see the minimum FPS in the table down the page! I call this real "smoothness" data...

since how most people like you are either avoiding the topics we are saying and then trying to say that we are saying different things i see this thread as pointless now. all it is is people trying to find out how to test for smoothness and then intel fanboys posting stuff that has absolutely nothing to do with the topic. so theres no point to post anymore just to let you know.

[grandmaster]

I've been doing smoothness tests on the console platforms for some time. Here's a comparison of Call of Duty 4 running on Xbox 360 and PlayStation 3, and an HD compilation of the clips here.

The methodology is to capture digitally lossless outputs of the HDMI ports of the consoles (DVI would work equally as well), then count the number of duplicate frames to ascertain true frame rate. The dump is then used to generate a graph which is overlaid over the original video. Here's the same thing done with Tomb Raider Underworld on console.

The only limitations with this methodology come down to the 'limitations' of the capture technology - I can capture 1920x1200 at up to 60fps. That would probably be sufficient for testing out this smoothness theory, but alas I have no access to AMD boards or CPUs. If someone from AMD wants to front up the tech and put these smoothness claims to the test, I'd happily carry out whatever tests are required - I do have a range of graphics cards up to and including GTX295.

[soundood]

did you do the Tomb Raider Underworld on console comparison between 320 and PS3?

[grandmaster]

Yes. I mentioned that video because it demonstrates quite nicely what I think the point is that people are trying to make. A smoother framerate is basically a smoother image on-screen, but crucially, dropped frames mean less feedback from the system from your input, meaning choppier controls and a less satisfying experience.

In the case of Tomb Raider though, PS3 version runs at 1280x720, while the Xbox 360 code runs 1024x576 upscaled to 720p, which explains how it maintains the solid 30fps.

The only other limitation of this technology concerns games where you lose v-lock. I can measure when a game outputs a torn frame, and include that on the graph, but I can't measure where on the screen the tear is. To illustrate a comparison of non v-locked games, here's console FarCry 2. A vertical line represents a torn frame.

But I'm guessing that most people prefer v-lock active. I know I would based on the amount of cash a decent graphics card costs.

The point is that this methodology would work equally well in testing the smoothness claims being attributed to the new AMD chip.

[Apokalipse]

See it's impossible to know really, look how bogged down this side of the forum is with AMD fanboys.

Fanboyism isn't a club exclusive to AMD.


Anyway, I think this picture I made should explain things better:

the fps on the left is probably higher on average. but it's very rough, and you're probably not going to have fun playing through it.
The one on the right, however, is quite smooth, and highly playable.

[roofsniper]

Fanboyism isn't a club exclusive to AMD.


Anyway, I think this picture I made should explain things better:

the fps on the left is probably higher on average. but it's very rough, and you're probably not going to have fun playing through it.
The one on the right, however, is quite smooth, and highly playable.

and thats the idea of how it should be tested. we need 2 systems that get the same average fps and maybe even the same min and max. one needs to be a core 2 duo and one needs to be a phenom system. then you get fraps and make it record the frame times in some game or something(needs to be decided on) and then you get the frame times and calculate the differences between the times and graph it. a perfectly straight line is the best possible situation. i don't have the systems to test so we need to find someone that does.

[charged3800z24]

and thats the idea of how it should be tested. we need 2 systems that get the same average fps and maybe even the same min and max. one needs to be a core 2 duo and one needs to be a phenom system. then you get fraps and make it record the frame times in some game or something(needs to be decided on) and then you get the frame times and calculate the differences between the times and graph it. a perfectly straight line is the best possible situation. i don't have the systems to test so we need to find someone that does.

I think you guys might be on to something... I wonder if Chew* would be interested in doing this.

[roofsniper]

I think you guys might be on to something... I wonder if Chew* would be interested in doing this.

i said it and a few others said it multiple times in the other thread but you could most likely not see it because everyone is arguing for no reason. i would be interested in the results but i can't personally test it.

[charged3800z24]

i would be interested in the results but i can't personally test it.

I know, that's why I was wondering if Chew* would, he has to systems that he was going to do a comparison with. He could add this into the mix.

[roofsniper]

I know, that's why I was wondering if Chew* would, he has to systems that he was going to do a comparison with. He could add this into the mix.

yea xcept he would have to make the systems equal performance. and i bet even if the phenom one is smoother ppl will cry that its a q6600. its worth a try tho if he is up for it.

[grandmaster]

Fanboyism isn't a club exclusive to AMD.


Anyway, I think this picture I made should explain things better:

the fps on the left is probably higher on average. but it's very rough, and you're probably not going to have fun playing through it.
The one on the right, however, is quite smooth, and highly playable.

As I've said, I can run these tests with no problem at any resolution up to 1920x1200. The results require no benchmarking tools and are 100% based upon the actual output of the DVI/HDMI port of the GPU. I've never dealt with AMD before - I work with the console companies for the most part - but if someone wants to put them in touch with me, I can do the comparison and put this to rest once and for all. The alternative is to basically keep on arguing about subjective perceptions and prove nothing.

[soundood]

As I've said, I can run these tests with no problem at any resolution up to 1920x1200. The results require no benchmarking tools and are 100% based upon the actual output of the DVI/HDMI port of the GPU. I've never dealt with AMD before - I work with the console companies for the most part - but if someone wants to put them in touch with me, I can do the comparison and put this to rest once and for all. The alternative is to basically keep on arguing about subjective perceptions and prove nothing.

its funny grandmaster, the last 6 posts just totaly ignore your solution to the whole issue, even to the point where Apokalipse draws a graph, by hand of what you have already done in real time to many games.

this is so funny, talk about seeing only what you want to see

ohhh yeah lets get chew* to do it, and ignore a guy who does this kind of thing for a living, the wonderful chew* can do it so much better than you all and this guy ehh?

this is in no way any disrepect to chew*, no more the attitude here in this post.

the guy can provide a visual test bed for all to see, but doesnt have some kit needed, he has all the graphics cards but needs some AMD boards and Chips.

what we should be doing is trying to 'pull' our contacts here, and getting this guy some bench gear, so he can put this to bed for us all, and PROVE that an AMD gaming system is 'smoother' than Intel.

surely AMD would be happy to bag this claim, with all test for all to see? also done independly, so nothing to take AMD to court for?

and if you missed what the guy is saying, heres his video comparison between Xbox 360 and PS3: http://www.youtube.com/watch?gl=IL&h...0TPY_-s&fmt=22
cmon Macci, and all the insider AMD lads, get this guy some kit, you would think nothing about trashing with some LN2

[roofsniper]

ahhh i never even saw grandmasters post. well if you are up for it go for it. what exactly are you doing tho to test it?

[grandmaster]

The methodology is very simple and is explained in post #152.

To go into more depth.. DVI/HDMI output a pure digital signal. I get a byte-for-byte dump of the port. I can then digitally analyse every single frame, and the software tells me the percentage difference between one frame and the next. Where a frame is identical to the last, that is a dropped frame. A game locked at 30fps (as many console games are) is still outputting 60Hz from its HDMI port, so you would expect every other frame to be 0% different from its predecessor. In the case of Tomb Raider on 360, by and large this is the case.

Due to the competely digital, totally lossless workflow, the results simply can't be argued with and the Tomb Raider video (same gameplay, if not the same actual video being generated) is a clear example of how this can help ascertain what's going on with this particular CPU in that it is showing a rock solid refresh rate (and therefore a similarly solid response from the controls) whereas the PS3 version refreshes very inconsistently in places. I believe this exactly what you guys are saying with regards the AMD chip vs Core 2.

So I'm sitting here with a Core i7 920-based system, a Q6600-based system and I've also got E8400 and Q9300 CPUs along with oodles of DDR2 RAM. I've also got a Radeon 4850, 4870 along with an 8800GT and a GTX295. What I'd need would be an AMD board and the Phenom. If this chip does indeed have the gaming advantage it is alleged to have, my results would be gold dust for AMD, so if they want to put the chip to the test, I'd be willing to do it for free. And yes, I am usually paid for this sort of thing I'd just like to see my tools being used outside the console arena.

[roofsniper]

So I'm sitting here with a Core i7 920-based system, a Q6600-based system and I've also got E8400 and Q9300 CPUs along with oodles of DDR2 RAM. I've also got a Radeon 4850, 4870 along with an 8800GT and a GTX295. What I'd need would be an AMD board and the Phenom. If this chip does indeed have the gaming advantage it is alleged to have, my results would be gold dust for AMD, so if they want to put the chip to the test, I'd be willing to do it for free. And yes, I am usually paid for this sort of thing I'd just like to see my tools being used outside the console arena.

yea thats what i was looking for. a q9300 and a 4870 vs a phenom II and a 4870? the two systems would have to be optimized to have the same average fps and maybe even the same min and max if possible. that should give a good comparison. i don't got a spare phenom II or amd board so im not sure how you would get that. if you think it is worth your time tho maybe try the q9300 vs the i7 920 with the same average fps and maybe min/max. the i7 should be smoother than the q9300 which would show that the average fps and min and max are not the only factors to consider.

[gosh]

So I'm sitting here with a Core i7 920-based system, a Q6600-based system and I've also got E8400 and Q9300 CPUs along with oodles of DDR2 RAM. I've also got a Radeon 4850, 4870 along with an 8800GT and a GTX295.

It would be extremely interesting if you could test those CPU's. And if it is possible to select heavy areas in one game (or three) that is single threaded comparing that with how multithreaded games work.

[demonkevy666]

with a name like grandmaster would you expect anything less

[grandmaster]

That dates back from playing Quake 1 when I used to work in games mags. We'd all vie for victory in lunchtime Quake tournaments and the title of 'the master', but I won so often that the accolade needed an upgrade!

[informal]

Guys,take a look at the first Phenom II AM3 review and the conclusion about the minimum fps(note: min fps is NOT the only thing responsible for "smoothness" of Phenoms,but it plays a part in it):
http://hwt.dk/literaturedetails.aspx?TeaserID=9545

Quoting user Mads321 from news section:

Hi.

Thought it would be ok to open a topic on the new Phenom II AM3 CPU's since the NDA runs out right about now (feb. 9th).

I'm gonna be sneaky and put a link here to my own review of the X3 720. There's 130+ graphs showing everything:

• Stock speed vs. old Phenom and the X4 810
• Clock for clock testing featuring E8400, X2 5000+, Core i7 920 and Phenom 9950 BE. Asll tests done @ 3.0 GHz
• AM2 vs. AM3 performance
• 1333 MHz DDR3 vs. 1600 MHz DDR3 @ AM3
• Overclocking

"It is not hard to see that Core i7 is ahead in most of our tests, but it has not proven itself stronger in gaming, where Phenom II actually was ahead in most tests when it comes to minimum FPS; a point which is vital regarding experiencing lag in games. The theoretical performance of the Core i7 is higher, but the question is if you will ever be able to use this extra power as a private consumer. Of course there will always be applications which favour one over the other, and if someone uses an application which is strongly favoured by one of them, the choice has already been made on those grounds, and then the price is not as important. For others we have to assume that value for money plays a big part."

[Frontl1ne]

grandmaster, I do hope you get in touch with AMD and get a Phenom II kit for testing purposes. Your method appears to be able to silence all this arguing about smoothness once and for all.

[grandmaster]

Well I was hoping that AMD might like to contact me but failing that perhaps someone in the know could PM me contact details for AMD, and indeed Intel. Contact names and email addresses for personnel who actually deal with the enthusiast sites, please!