For people who can't comprehend the differences between Ram

#31HalecticPosted 9/25/2013 8:29:49 PM
assassin10133 posted...
Umm, actually i'm pretty sure that math is only correct if you have to be accessing the entire block of RAM all at the same time, and I'm also pretty sure there is no way in hell a game code would need to be accessing 8GB of ram in a single read/write cycle which is what that math would imply. If that would be required then yes, huge advantage PS4, but never going to happen in the real world. It's also worth noting the Mark Cerny himself admitted they considered going with a similar setup and could have had a 1tb/sec bandwidth (I think that's the speed he referenced) but opted for the single large pool of fast ram because it was much easier to develop for and it's one of the main things devs wanted, so that makes me think there has to be some merit to the eSRAM setup if Sony considered it too. Cerny never said they didn't take that route because it offered worse performance, but that it was harder to utilize which is obvious, devs have to try to keep their read/write accesses to 32mb or smaller to be able to match what PS4 offers.



Developers have to jump through hoops to get the eSRAM+DDR3 to get almost the same performance as the unified 8gig GDDR5, but the 8gig of unified ram will still outperform it.
---
Digimon World Dawn FC 000111545835
#32XBOXvsPCPosted 9/25/2013 8:40:58 PM
XBOXvsPC posted...
"PS4 has 256 32MB blocks running at 176 GB/s." - that's not how it works kid.
All these blocks together achieve 176BG/s, not each one.

Jesus...it's really embarrassing
#33assassin10133Posted 9/25/2013 8:51:39 PM
Halectic posted...
assassin10133 posted...
Umm, actually i'm pretty sure that math is only correct if you have to be accessing the entire block of RAM all at the same time, and I'm also pretty sure there is no way in hell a game code would need to be accessing 8GB of ram in a single read/write cycle which is what that math would imply. If that would be required then yes, huge advantage PS4, but never going to happen in the real world. It's also worth noting the Mark Cerny himself admitted they considered going with a similar setup and could have had a 1tb/sec bandwidth (I think that's the speed he referenced) but opted for the single large pool of fast ram because it was much easier to develop for and it's one of the main things devs wanted, so that makes me think there has to be some merit to the eSRAM setup if Sony considered it too. Cerny never said they didn't take that route because it offered worse performance, but that it was harder to utilize which is obvious, devs have to try to keep their read/write accesses to 32mb or smaller to be able to match what PS4 offers.



Developers have to jump through hoops to get the eSRAM+DDR3 to get almost the same performance as the unified 8gig GDDR5, but the 8gig of unified ram will still outperform it.


Yes they will and I agree the 8GB GDDR5 setup is much better for gaming, not disputing that. But if devs do take the time to get that performance out the eSRAM in real live game code there won't be a huge difference in performance.
#34pigboyPosted 9/26/2013 6:58:00 PM
TurtlePowah posted...
There is this strange rumor that Xbox One will have better ram because of the ESRAM speeds of 204GB/s because of this article "http://www.eurogamer.net/articles/digitalfoundry-vs-the-xbox-one-architects".

So let us see ^_^, let us input the numbers.

The smallest number we are dealing here with is 32MB so to avoid confusion for you people not good at math we will only use 32MB increments.

Xbox One has 256 32MB blocks running at 68GB/s has 1 32MB block running at 204GB/s.

PS4 has 256 32MB blocks running at 176GB/s.

So what happens if PS4 256 blocks and Xbox Ones 257 blocks run for 1 minute? 1 hour? How much data gets transferred?

These numbers represent operations achieved with max through put to give a idea of how much more "data" will be going through the chips.

PS4 @ 1 min - 2,703,360
One @ 1 min - 1,056,720

PS4 @ 1 hour - 162,201,600
One @ 1 hour - 63,403,200

That's PS4 Ram running 255.8% times faster then Xbox Ones Ram while utilizing its "ESRAM".

So will people please stop acting like 1/257th of Xbox Ones ram will make the system faster.... Please?


This info truely helps those people without computer knowledge. Anyone with a good understanding of computer hardware & programming knows that a tiny 32MB eSRAM is too little to do any real performance difference for the Xbone. Your simple math just make it so easy to understand the result.

Another problem of eSRAM not only it is so small to the point of non-sufficient, but in order to use it the developer will require extra programming and it will require extra CPU cycle to direct & manage the data from eSRAM to the Xbone super-slow DDR3 RAM.

With current generation HD graphic from 720p & 1080p, the usefulness of the tiny 32MB eSRAM to the game developer is almost to non-existence.

With the PS4 memory is 255.8% faster than the Xbone even with the eSRAM included and much easier to use by the programmers. We can see the PS4 is many times better the Xbone easily regardless to the claims of many Xbox fans. It is just simple logic & basic math any person can easily understand & even do it themselves.
#35Foxx3kPosted 9/26/2013 7:18:35 PM
I can't help but feel that you don't understand what RAM does.
---
[LanParty nF4 Ultra-D] [AMD64 3700+ San Diego] [2x 1gb Corsair XMS 3-3-3-8] [2x 250gb Barracuda] [Soundblaster Audigy 2 ZS] [X850XTPE]
#36SilentS89Posted 9/26/2013 9:13:50 PM(edited)
pigboy posted...
It is just simple logic & basic math any person can easily understand & even do it themselves.


The battle cry around here is "math is morally wrong, don't trust logic, only rely on official spokespersons for information, Game devs don't count either."
#3795_EclipsePosted 9/26/2013 9:18:35 PM
SilentS89 posted...
pigboy posted...
It is just simple logic & basic math any person can easily understand & even do it themselves.


The battle cry around here is "math is morally wrong, don't trust logic, only rely on official spokespersons for information, Game devs don't count either."


Most people think that the magical crop fairy makes the other 8000+MB of ram go the same rate as the 32 MB of ESram.
---
Favorite game to date - Xenogears \/-/-/-/\
If you believe used games are the industries problem, then you sir or ma'am are a fool.
#38kakarot-uchihaPosted 9/26/2013 10:23:40 PM
not one opponent refuted it? tc got a lot of insults, 4 pages in and no one did a little bit of math to prove him wrong, he could be, maybe he did embarass himself. why didnt you show us?

he used max performance numbers we were given
he said 'max' throughput possible, as in the entirety of their ram bouncing off the rev-limiter. so yeah he said that, all parts of ram running highest possible speed. 'thats not how ram works' or 'not 32 mb blocks' wasn't helpful and doesn't prove it wrong. he clearly said using 32mb increments for illustration purposes.
please give some real world numbers if you think either rams tires wont always be able to hook up.

tc please explain your math, cause at 176gbs i get 10,560,000,000,000 in 1 minute, and i thought the gbs was the max transfer. and i thought we got this by multiplying the component speed by the component space. its possible i understood your post and completely missed the point. also just out of curiosity is it bytes or bits/s cause its never spelled out, pure factoid only,wont change the math, if its in bytes thats just an even more impressive transfer.
im gonna go look around.
#39THE_PS1_PATRIOTPosted 9/26/2013 10:46:38 PM
Alot of dummies in here. Get General RAAM in to explain.
---
PLAYSTATION PATRIOT
#40Cowboy082288Posted 9/26/2013 11:22:29 PM
kakarot-uchiha posted...
not one opponent refuted it? tc got a lot of insults, 4 pages in and no one did a little bit of math to prove him wrong, he could be, maybe he did embarass himself. why didnt you show us?

he used max performance numbers we were given
he said 'max' throughput possible, as in the entirety of their ram bouncing off the rev-limiter. so yeah he said that, all parts of ram running highest possible speed. 'thats not how ram works' or 'not 32 mb blocks' wasn't helpful and doesn't prove it wrong. he clearly said using 32mb increments for illustration purposes.
please give some real world numbers if you think either rams tires wont always be able to hook up.

tc please explain your math, cause at 176gbs i get 10,560,000,000,000 in 1 minute, and i thought the gbs was the max transfer. and i thought we got this by multiplying the component speed by the component space. its possible i understood your post and completely missed the point. also just out of curiosity is it bytes or bits/s cause its never spelled out, pure factoid only,wont change the math, if its in bytes thats just an even more impressive transfer.
im gonna go look around.


Hmm interesting... yes I see your point.... but I don't know. Personally I still think Ford is better than Chevy.
---
PSN/XBL/Steam - cowboyoni