8GB DDR3 + 32MB ESRAM vs 8GB GDDR5

8GB DDR3 + 32MB ESRAM vs 8GB GDDR5

In terms of performance, how will these RAM solutions for the next gen consoles play out over the long run? Can we expect both to be about equal in 2 years time, or will one of these RAM solutions perform noticeably better 2 years from now onwards? Or is it nearly impossible to predict at this point?

32mbs of ram wont do much of anything to help

ESRAM is no ordinary RAM. It is MONSTER RAM.

32mbs of ram wont do much of anything to help

32mbs is so small its the tiniest monster ive seen

Nope. 32 MB of GDDR5 RAM > 32 MB of eSRAM.

eSRAM is a small pool of 109 GB/s bandwidth ram that's a pain in the ass to manage. Nothing more. No secret sauce nonsense. It'll only be used as a framebuffer for most games.

Cerny said he weighed the pro's and con's of both of these kinds of ram and he chose the GDDR5 ram because the ESRAM would be another puzzle for the developers to solve.

That's Cerny having some class.

Plain and simple, eSRAM on the Xbone is only limited to 32 mb in size and the bandwidth in that size is still much smaller than the bandwidth of the GDDR5 RAM that the PS4 has.

The eSRAM was implemented to aid the DDR3's slow bandwidth. Like the Xbox 360's design, some of the elements of the game will be stored in the eSRAM while everything else will go the DDR3 RAM. It's not so much as a puzzle as it is difficult to decide which parts of the game should be stored in the eSRAM and which parts should go to the slow DDR3 RAM.

The "will get better performance after we optimize the APIs in the future" is nothing but false hope that Microsoft is giving the Xboners so that they don't migrate to the PS4.

Do you know what ESRAM is?

By that statement, it seems like you have no idea...

I feel like I just stepped out of a time machine. Where I come from (the future) PS4 and XB1 already released and it was found that 32MB ESRAM wasn't sufficient. Xbone can't run COD or BF4 higher than 720p meanwhile PS4 runs COD at 1080p and BF at 900p while maintaining higher frame rates. Than the XBone versions. Also most XB1 games are 720p while all PS4 games with the exception of BF4 are full 1080p.

Do you know WHY or are you just stating the obvious without being open to discussion? If its the latter, i wont waste my time.

Because the Xbone is weaksauce.

Devs are using the eSRAM as a framebuffer, but 32 mb is too small to run at 1080p. Combine that with DDR3 RAM and a bonaire GPU weaker than the 7770 and you get 10+ lower FPS average, no AA, crappy textures and more.

Ok, its the latter, have a good day...

No. Just no.

Everything you just wrote shows you have no idea how Esram works or what it is.

lol...

The eSRAM is just plain ram that's slightly faster than DDR3 but immensely slower than GDDR5. It's nothing special.

ESRAM has much lower latency and higher bandwidth than GDDR5.

Yeah, that lower latency helped immensely with 720p with no AA and lower framerates.

The latency figures are mostly irrelevant, as stated by Cerny and shown by real-world results. For all intents and purposes, GDDR5 latency is superb and anything lower than that has no benefit. Secondly, eSRAM's 200+ GB/s bandwidth is theoretical and it's taking into account the possibility that the eSRAM has a bidirectional bus, which it doesn't. Real-world bandwidth is at 130 GB/s in a highly-controlled setting and ~90 GB/s in practical scenarios (When you're running a real game) compared to the PS4's 176 GB/s theoretical peak and 172 GB/s in real-world performance.

You have no idea confirmed

More asshurt.

How about refuting what I said. eSRAM is overglorified mid-speed RAM.

Cows taught me nothing in the known universe is better than 8GB GDDR5 RAM so I choose that one

Cows are exalting SONY because 8 GB GDDR5 RAM was the best outcome that could have happened in this console war. Imagine having RAM that's almost twice as fast as the Xbone's eSRAM but with the size of 8 GB.

The Xbone is at a big disadvantage.

Why waste time when you have shown that you will just ignore any logical argument?

It terms for memory performance there is no real difference between DDR3+ ESRAM vs GDDR5. What makes the difference is the hardware's processing power.

The main reason why the X1 cant handle the same resolutions as the PS4 is the gpu's processing power is a 3rd slower and only has only 16 ROPS which translates how well the gpu can write the render data to memory.

Isn't having esram a slight boost but also a massive annoyance to devs?

Correct me if i'm mistaken.

Its more than a slight boost, but when you have outdated drivers.......its a major annoyance.

XB1 launch games have been plagued with last minute changes to hardware speed and outdated drivers.

As the dev tools get updated and optimized, XB1 will be running 1080p with no problem. Whether the framerates are as fast as PS4 is another issue.

This is false the fact that X1's gpu is on average 30% slower and only has 16 ROPS vs PS4's 32. means that the X1 wont be able to render the same graphics and resolution along with equal FPS. ROPS on the gpu determines how well a a gpu can translate render data. the ESRAM is not the magic pill, while it can move more data then GDDR5, and read and write 4x faster then GDDR5 its a memory bus buffer. Its not meant for solely graphics work.

RAM doesn't do much when the GPU is roughly 50% weaker.

its not 50% is around 30% slower.

@Thunder7151: 8GB DDR3 + 32MB ESRAM > 8GB GDDR5 the war has been won

I know that......which is why i said running 1080p, and didnt not say equal graphically. The raw specs of both are enough to run 1080p and 60fps. Not in all games obviously, but the argument that the XB1 isnt capable because it is weak is silly. Any multiplat that runs at 1080p on PS4 has that same potential on XB1 when properly optimized. Whether XB1 has the same effects such as particles, filtering, shadows, textures, etc is another story.

32MB ESRAM serves as a bottle neck more than anything else.

The unified pool of GDDR5 RAM is the winner, MS didn't choose ESRAM and DDR3 over GDDR5 for performance, it's because they didn't think it'd be possible to get 8GB of GDDR5 down on the chipset and because of the cost plus the likelihood of the shorter time to shrink the die size.

MS are probably fairly happy with their choice, the price will drop faster so they'll get better margins and they're apprently already making a profit on the X1. Gamers and devs however may be less thrilled.

Either way it'll be irrelevant, the weak ass CPUs used will be the first bottleneck imo but at least Sony's got GPU compute to fall back on and a beefier GPU to take the load.

W

T

F

?

Then explain why multiplatforms , so early in this gen having better resolutions/graphics on PS4 systems than X1s ? Where is this monster RAM to save the day !

It doesn't matter if anyone has 100% understanding of how ESRAM works or 0% understanding.

The fact remains. The X-Bone consistently has less resolution AND frame-rates than its PS4 where multi-plats are concerned.

The proof is in the pudding. There may be some sort of boost that the 32 extra MB offer, but it clearly does not put it over or even at the same level as what the PS4 is able to output.

Just accept it and let it go. There's nothing wrong with having a console that performs a little lower than the other. If you're in it for the exclusives, then enjoy the exclusives. Trying to argue against EVIDENCE is like trying to fly without wings. You will meet the ground relatively quickly, and create a messy scene for every innocent bystander to clean up.

Why limit to "8GB DDR3 + 32MB ESRAM vs 8GB GDDR5"?

You should have added iBuyPower's $499 Steam box (yet another AMD Gaming Evolved based X86 PC box) e.g. 8 GB DDR3 + 2 GB GDDR5.

At 1080p, most games are not ROPS limited i.e. games such as COD Ghost and Battlefield 4 are mostly CU bounded. A CU includes ALUs and TMUs. ROPS includes memory writes.

1.3 TFLOPS GCN (prototype-7850) with 12 CUs + 32 ROPS was proven to be inferior to 1.76 TFLOPS retail GCN (7850) with 16 CUs + 32 ROPS. Both prototype-7850 and retail 7850 has the same 153.6 GB/s memory bandwidth.

Xbox One with 32 ROPS would not change the current situation.

We'll put you out of your misery, just hold on

ROPS make a difference and so affect the gpu's ability in the end results..

ROPS are responsible for final pixel output, having 16 ROPs definitely puts the Xbox One at a disadvantage many pc gpu's and PS4 have 32,, The difference in raw shader performance (12 CUs vs 18 CUs) can definitely be a problem in games that run more complex lighting routines and other shader intensive jobs on each pixel, but all of the resolution differences between Xbox One and PS4 games at launch are likely the result of being ROP bound on the X1 along with less processing power. This is probably why Microsoft claimed it saw a bigger increase in realized performance from increasing the GPU clock from 800MHz to 853MHz. The ROPs operate at GPU clock, so an increase in GPU clock in a ROP bound scenario would increase performance more than adding more CU's.

The ps4 has faster memory and a faster gpu period. Just look at the multiplats the ps4 clearly has almost all games running 1080p vs 720p to 900 on the x1.

No, 7970 has the same 32 ROPS as PS4's 32 ROPS and 7970 delivers superior results i.e. more CU, more TMU, more memory bandwidth.

Note that CU includes both ALUs (for shaders) and TMU (for texture related workloads).

BattleField 4 and COD Ghost's frame rate can be estimated via CU count which roughly matches it's Radeon HD SKU counterpart.

The reason why the 7970 has only 32 ROPS is because its main resolution target is 19**x1**** ranges. And to the fact that AMD decoupled the ROPS and increased the memory bus enough to feed the ROPS. the R290 has 64 ROPS which the resolution target is beyond those type of resolutions. Yet 290's are not massively stronger then 7970's (only 20-30% faster). Have no idea why you are ignoring the fact that ROP count does affect the end results dependent on targeted resolutions. We all know CU/TMU and memory bandwidth does affect a gpu's ability to render, as does ROPS.

ROPS along with the gpu coreaffects the gpu's pixel fill rates, and the amount is set according the size of the bus width, the memory type and the memory clock. ROPS affect the gpu's ability to perform the transactions between the relevant buffers in the local memory. Higher the pixel rate, higher is the screen resolution of the GPU. The X1's 16 ROP and weaker gpu are the reasons why 1080p/60 fps with equal quality as PS4 will not happen.

Then factor the PS4 has a 150% more powerful GPU also.

You basically just said you have to nurff a game on x1 in order to have a intense game run 1080p if the PS4 is runing that resolution, so how is that "same potential" with XB1?

Hell PS3 and 360 games can run 1080p if you nurf enough of what is going on enuogh, you aren't making a very good argument here at all.

This whole "ESRAM" hype is more pathetic than the "cloud processing" nonsense. You Xbots are gullible as **** to think 32mb of any RAM is going to make a noticeable difference when it comes to anything. It's all marketing hype for you little fanboys who know nothing about how computers work to have hope that the Xbone is somehow going to magically catch up to the PS4 graphically when the PS4's video card and 8gb GDDR5 RAM take a fat dump all over the Xbone's weak VC+DDR3 RAM.

But hey, the Xbone will always be better than the PS4 at degrading your cable/satellite's picture quality when you watch TV through it. So at least you can hold onto that fact.

how could you come to this conclusion? in fact it's more than 50% stronger since kinect saps 10% of the gpu. although the xbone's upclock somewhat mitigates that, but it's still a bit more than 50%. that's without factoring the rops and compute advantages.

also, xbone doesn't even have a cpu advantage either, seems like the os and kinect really eat into that as well. either one more core is taken up by the os, or the ps4 is higher clocked, it's probably the former though considering ps4's size.

combining all the ps4's advantages, bandwidth being the biggest, it isn't quite farfetched to say that ps4 is close to double xbone in a lot of situations. the xbone is just too bottlenecked, is focused on kinect and a bloated os, and is just flat out weaker.

not to mention the nasty sharpening upscale and crushed blacks, which 360 had the latter problem as well.

shit, nvm the gpu is around 33% slower, i mixed up words.

Right now? Yes, absolutely.

Down the road? Who knows, its all about the game and/or the engine.

PS3 was also the more powerful console but we rarely saw that in action.

Im just saying that raw comparisons are misleading. We all know the PS4 is more powerful, it would be silly to deny.

All im saying is the difference isnt as huge as cows are saying, or as small as lems are saying. Its up to the devs to determine.