What claim did I make about GT5? I made a bet with ethomaz about it not breaking 10 million by 2013 and I won last I checked.

As for this whole PS4 performance thing, uh everyone in the thread agrees about 9-10x improvement in power, so yea I'm kinda right here. You originally claimed it could be as much as >15x, you were wrong, why is that so hard to admit? geez (head banging does not suffice indeed). Do you not see my discussion with Pemalite? Were both using links to back up our arguements.

What the heck are you talking about?

1) If you compare just PS4's and PS3's GPU, than difference is most certainly more than 10x.

2) You're using FLOPS as only measure of systems capabilities - don't...as I said, shader performance skyrocketed in last gen, but other parameters have not...making overall performance jump around 10x from XBOX1 to 360 (which is insane in just 4 years, and is much more impressive than 10x or slightly more in 8 years)

3) Wiki page states 230 GFLOPS for Cell (that's PS3 CPU) - Cell has 1 PPE and 8 SPEs - each of these performs at 25.6GFLOPS, single precision. 1 SPE is inactive for yield issues, 1 is reserved for OS. So math is (1PPE + 6SPE)*25.6GFLOPS = 179.2GFLOPS available to devs (plus 176GFLOPS from GPU). Of course, unless you're Miyamoto Musashi of PS3 coding, you won't be able to see anyhere near this numbers.

4) Because core speed is completely irrelevant if you don't know other parameters. For example, 360 is running at 550MHz core clock, and it has 48*5 shaders (for total of 240). So in theory it can do 240 * 0.5GHz = 120 Goperations per second. Since each shader can do 1 multiply+1 add at the same time, whole GPU can perform 120*2= 240GFLOPS. That's in theory...360 is built on VLIW5 architecture (just like WiiU, while we're at it), so in most cases only 4 out 5 shaders in group will be working, bringing down that number effectively, and that's just for start.

PS3's GPU (RSX) does not have unified shaders, but instead 24 pixel and 8 vertex shaders, each capable of 10 operations per cycle. RSX runs at 550MHz, so math is: 24 * 0.55GHz * 10 + 8 * 0.55GHz * 10 = 176GFLOPS. This is theory again, because you need perfect alignment of loads on pixel and vertex shaders to achieve theoretical maximum.

PS4's core clock is 800MHz, and it has 1152 unified shaders, so math is 1152 * 0.8GHz * 2 = 1843.2GFLOPS. PS4's GPU is also built on GCN architecture, which has numerous improvements over VLIW5.

Same thing is when calculating theoretical pixel/texel fillrates - clock * number of ROPs (pixel) or clock * number of TMUs (texel) (360 has 16 TMUs and 8 ROPs, PS3 has 24 TMUs and 8 ROPs, PS4 has 72 TMUs and 32 ROPs).

On top of all this, you have memory bandwidth - good example how it affects same GPU at exactly same core clock speed are 3 HD5550 cards with 3 different memory bandwidths (12.8GB/s : 25.6GB/s : 51.2GB/s) - 30fps : 40fps : 49fps

So, as you can see, basing your comparisons on core clock (or GFLOPs for that matter) is...well, wrong. At the end of course, no matter how much theoretical data you have, you need real benchmarks, and that is only real measure of how certain GPUs perform (or compare). In this case...go to 1)

Making me dig....

Also since you're also claiming that I made some false claim about GT5 where I initiated a debate and was wrong, I want you to bring up the post where I made this false claim about GT5.

You can't say that I make claims and never back them up, and then make a claim yourself....and not back it up.

If you're refering to this, first of all this was a prediction, not a claim of some sort which can be backed up with evidence, and it wasn't a bad one. When GT5 did eventually surpass 10 million (which still surprises me with how poorly it opened), I did admit being wrong.

http://gamrconnect.vgchartz.com/post.php?id=5344398

----

Anyway, I did some digging, just for you sweetie, cause man, you have an ego.

This was your first post

http://gamrconnect.vgchartz.com/post.php?id=5244098

I said PS4 is much more then 9x more powerful then the PS3, Captain_Tom said 10-15x, you agreed.

This is a futhur post - http://gamrconnect.vgchartz.com/post.php?id=5249105

Where you literally say:

So there, you used to think PS4 could be 15x as powerful as PS3, I showed you it was closer to 10x, but you were too arrogant to admit that I was right, and you were wrong.

You did it a few times actually here:

http://gamrconnect.vgchartz.com/post.php?id=5249716

Anyway after I called you out on that, instead of backing up your claim like I asked you to do multiple times, you resorted to insulting me, and that's what pissed me off. You have a huge ego, if you can't admit you're wrong, and resort to insults instead.

Throughout the thread you agree'd with Captain Tom, multiple times, and he stated

So, it's a bit of a stretch since you didn't actually quote him on this, but because you agreed with Captain_Tom multiple times, and he said that even 15x more powerful is a conservative estimate, then you basically agreed that PS4 could be more then 15x as powerful as PS3.

----

Now last I checked, you're still running around saying 10x more powerful, not 10-15x more powerful. Your original high estimate was wrong, I showed it to you in that thread to the best of my capabilities, but you instead resorted to insults.

Be a man.

I... I almost believed it.

You're still using pointless GFlop numbers.
The PS3 wasn't hard to program for because it had lots of CPU cores, otherwise this new generation will be harder to program for than the last, which is simply not going to be the case.
Instead the PPE ends up being more of a "managing processor" whilst the SPE's are more for specialised tasks, that's why it's hard to program for, not because of the amount of cores.
It was also hard/impossible to reach the theoretical max because it's stupidly bad at doing any type of meaningful processing outside of linear equations.

I mean really, do you honestly beleive those numbers? That makes the Cell order of magnitudes faster than my own processor a processor that costs more than two PS3's combined, has an extra 6 threads, more and faster cache, faster memory bandwidth, wider, better branch predictor (The Cell doesn't have one.) and is 7 years newer, roughly 10 times the transisters...
http://i.imgur.com/5tuXDda.png

Which then means that the processor in the PS4 is going to be roughly multiples slower than the Cell (Again, not happening.) Because the CPU in the PS4 can't hold a candle to a high-end Hexacore i7.

And that's why, GFlops are pointless.

As for the GPU, yes you were wrong, I've provided plenty of information prior to this post on that point alone.

1) So I'm right when we talk about raw power, PS4 is about 10x as powerful as PS3, not up to 15x, or more powerful?

2) I don't undertand how the jump from XB1 to 360 is only 10x, when the GFLOPS diff is about 15x. I know you said I shouldn't use GFLOPs to measure, since stuff like shaders have improved drastically, but the lack of an improvement in shaders wouldn't make the performance jump smaller.

Either way, not worth arguing, you admit that in 4 years the graphical jump was more insane then current gen jump, and if I were to argue anything (why are we arguing about PS4's performance, we both agree 9/10x or so), that's my arguement. Also I would argue that we see less improvements with similar jumps in power because of dimishinging return. 600-6,000 polygons is noticable, but 60,000 to 600,000 polygons isn't really. PS4/One will look good because they can have so many objects on screen. Killzone for instance had a whole city rendered, well I would add.

3) Okay, makes sense, I read about those 8 CPU's, I didn't realize the 230GFLOPS value accounted for unused cells.

4) I follow you half way, LOL. You say I can't us GFLOPS to measure performance, but here you are calculating GFLOPs.

Where did you find out the number of unified shaders?

Also I'm not basing my comparisions on core clock speed. I said:

I didn't say it to downgrade PS4's performance, just to make a point that not everything on PS4 improved so drastically. I didn't say anything wrong, but people assumed I was implying something by saying that.

----

Basically, what I've learnt from you is that you can't directly measure PS4's performance based off its specs. GFLOPS seems the best way to estimate a number (What's a better way?), but you need a raw benchmark because things like Memory Bandwidth, fill rate, and bus (something I know was PS2's bottleneck) affect performance. With consoles, you won't get that, since PS3/PS4 games run off 1 setting, where PC games can be altered.

Now I have a question, if Memory bandwidth is only a 5x improvement in the PS4 (38.4 -> 192 GB/s), will that hinder and help the PS4's performance? Is there a linear relationship, or exponential?

15x is still wrong, and you used to say 10-15 times, where I told you 15x is too much, 10x is more accurate. Now you only say 10x as powerful, which is not what you were saying at first.

Also you agreed with Captain_Tom, who said over..... So take it or leave it, I'm not oblivious to fact.

---

Now................... Gran Turismo 5. You're claiming I made incorrect statements? Back it up, because honestly if you expect me to back up my arguements, but refuse to back yours up, and then resort to belittling comments, then you have no right to critisize me anymore. Clearly in this entire thread, I have said nothing wrong and I'm tired of your arrogence.

---

To Mods: I'll take your advice, I won't engage with junk like this anymore. I'll be the better man. Lock this thread if you like, but I won't reply to CGI anymore, I'm tired of him trying to demean me and his superiority complex.

Memory bandwidth is difficult.
Memory bandwidth will have a direct relationship with fillrate, however there comes a point where more memory bandwidth just won't improve anything.

Think of it like this, with memory bandwidth if it is too slow, the GPU has to wait on more data to arrive, that's wasted clock cycles and thus you receive a performance penalty.
Where-as if you have memory bandwidth that is too fast, the GPU has all the data it can possibly handle leaving memory bandwidth wasted, this is a waste of money as you could have gone with tech that is cheaper/slower clocked.

Microsoft and Sony would have done the math on this, same with AMD and nVidia and it seems for a GPU that's around the performance of the PS4's the magic number is around 150-200GB/s.

You also need to keep in mind that consoles are only rendering at 1080P or lower, it's not that demanding anymore so you don't need roughly 300GB/s of memory bandwidth like in high-end PC cards.
All in all, when running similar resolutions the memory bandwidth can have orders of magnitude performance benefits, provided the GPU can keep up and use it, that's not to mention that technology such as 3dc texture compression will save on memory bandwidth right out of the gate.

What are PC's rendering at, if not 1080P?

How do you find out about all this stuff? Like PS4 needing 150-200GB/s of Memory Bandwidth?

Either way, in the end, PS4 is about 10x as powerful as PS3. I did learn a lot about why, but I never claimed it was more or less then this (well I think closer to 9x, but 9x, 10x, it's about that).

---

So do you agree with me about how PS4 games will look better? Like I said increasing the number of polygons, even by a factor of 10x, makes a small difference now, compared to N64 times. However with the larger number of shaders, and higher RAM, PS4 and One can render more objects on screen, with better graphics and texture resolution then a PS3. So PS4 games in comparision to PS3 games, won;t look particularly better, but effects (like sand), and the number of ojects on screen will improve, making game worlds look much more full.

Actually, I hear this is how Watch Dogs works, the PS4 and PS3 game are based off the same engine, but the PS4 game has higher texture resolution, more effects, NPC's, cars, object, etc on screen, etc.

PC's aren't limited to 1920x1080 (1080P)
2560x1600, 2560x1440 resolution monitors are available.
You can even throw several together in eyefinity or in surround gaming for 7680x1600 or 7680x1440 or even 3x 1080P screens for 5760x1080.
If you are running say... A couple of Titan GPU's on a crappy 1920x1080 TN monitor or TV, you are wasting performance and money.

And trust me, running games in eyefinity is stupidly demanding, but the experience is downright amazing that consoles will never be able to achieve.
This is eyefinity in action: http://www.youtube.com/watch?v=IyQoFPO8RyA

As for where I learnt this stuff, well. Over the course of roughly 15 years being a hardware enthusiast it just starts to become second nature.
Most of it is from places like Anandtech, reading wikipedia, reading indepth hardware analyses of new architectures etc'.
Most of it is just logic, probably helps that I built some primitive processors back when I was in school too years ago which provided me with some funademental understanding on how it all works.

Also, I never said the PS4 needed 150-200GB/s, I said that's what these Billion dollar companies would have come to the conclusion of for best results, they run all sorts of tools to find out what is holding back architectures and then pass that information onto the general public in various whitepapers and documentation such as the reasonings on why AMD dropped VLIW5 for VLIW4 and then moved onto GCN.

But yes, I agree with you that PS4 games will look better, anyone who disputes that is silly.
As for geometry, it will make a big difference. I'll show you why.


Think of the left part of the picture as the PS3, think of the right hand side as a PS4, that is the difference a little bit of extra geometry does to a scene, it's not to be understated. (That's with everything else the same, I.E. textures, lighting, shadows etc'. - All will be upgraded on the PS4)
Right now games generally use flat ground.
Everything gets dialed up to 11 this next generation, well for consoles at-least, the PC has had all this technology for years, it's just only in the last year or two that games are starting to catch up.