GDDR 5 vs DDR3

@blackace: That’s what’s important.

YES!

But tech talk is interesting sometimes and might teach you new things.

Don’t get me wrong. I wasn’t trying to say that Xbox One was better or PS4 was better. I was just asking what people thought and if latency would be such a serious factor. Most people didn’t even take that in mind and just spitted out something completely irrelevant, notably someone saying that DDR5 is faster while DDR5 doesn’t exist yet. GDDR5 !!!!!!!!!! Anyway, you have got the important factor at hand. Both consoles will play awesome games. :)

eh, GDDR5 is probably overall 'superior' for a games console, but we have not and we will probably never see the difference in reality.

This is some funny sh** you are not a fanboy yet every post you make is skewed to kiss MS ass and ESRAM.

Is funny that you say the xbox one has 192GB/s bandwidth,2 days ago you claimed 218Gb/s,and 200GB,when i told you and i even quote MS engineer on it saying it was capable realistic of only 140 to 150 GB/s a second.

But let me tell you this the xbox one can has 500Gb/s bandwidth way more than the 7970,it means total sh** because it can't fill that bandwidth it has power relative of that of a 7770,which has 72GB's bandwidth,in fact the full 7790 has 96Gb/s,and is stronger than the xbox one is,the xbox one having more bandwidth than the PS4 means little,because it doesn't have the power to do sh** with that bandwidth.

The PS4 use more than 4GB for games dude,see you are totally clueless.

And yeah the so call developer who claimed that the xbox one had more bandwidth that MS thought was MS it sell,it wasn't a developer is was MS who told that to DF and they pretended it was a developer,it was latter confirm when questions were ask and DF responded saying that they didn't know something an that they would have to as MS about it..lol

But once again the xbox one can has 500GB it mean little,is like having a 4 lane race track were you could take any car without having to wait for a hole to pass,but your car has 85 horse power the other car has 135 horse power and more torque and actually racing tires.

@tormentos: WTF ?? So now I can go around and say everything Nvidia is saying about their new Tegra K chip is BS because it sells. In that case don’t buy any product from now on..... It’s probably still not faster than PS4 so why do you cry ???

I love how I said Theoretically 218GB/s

Meaning Freaking Theoretically.

Stop saying things half. You are A FOOL.

Yes the PS4 uses 5GB on games. But there aren’t any games higher than 4GB. So I said 3GB and I said I was being generous.

I hope you do not mind but I added numbers to your quote so you could see what I was responding to easier.

1) 2133 Mhz DDR3 is 1.5 - 1.65v or there abouts, different modules have different specs but if you look at newegg (or ebuyer if you are UK based) you can see the voltages there.

GDDR5 is about the same in terms of voltage but because it runs at a higher clock speed it does draw more power than DDR3 of the same capacity and voltage.

2) DDR3 is 64 bit/channel and GDDR5 is 32bit/channel. The memory controller on a CPU will obviously be configured for DDR3 but there is no reason that a CPU cannot use a GDDR5 configured memory controller, there is very little difference to be honest.

3) This is true but usually the limiting factor for bus width is not how small they can make the memory controller it is how small they can make the physical connections to the socket the GPU/CPU/APU sits in.

4) Not really. GDDR5 is not used in PC systems as standard memory because it is a) expensive and b) overkill. There are very few CPU bound processes that you use on a standard PC that would take advantage of GDDR5.

5) Timings and clock speed are related, hence my car analogy. If you have loose timings but a high clock speed the net effect is the same as tight timings and a slow clock speed.

6) True

7) According to Hynix from their specification documents the following are the timings.

DDR3 document Here (download) pg 31

GDDR5 document Here (opens in browser) pg 131 - 135

As you can see in this case the GDDR5 actually has a lower latency than the DDR3. I do not know the exact timings used in the Xbox One or the PS4 but the latency is going to be pretty much the same in both consoles so neither one has an advantage when it comes to latency.

The advantage the GDDR5 gives the PS4 is higher sustained bandwidth, the Xbox 1 can burst small amounts of data at high bandwidth than the PS4 if you combine the ESRAM and the DDR3 but once you have used what is stored in the ESRAM and are required to stream directly from the main memory pool your bandwidth drops. Managing the ESRAM is key to maximising Xbox 1 performance.

The other problem the Xbox 1 has is that it has a less powerful GPU, so even if you were to manage the ESRAM perfectly to maximise the Xbox 1 performance you are still going to be at a lower performance level than the PS4 which is also easier to extract performance from.

With regards to your PRT (tiled resources) talk, the thing is both consoles have it, it is not unique to the Xbox 1. The idea is to enable the use of higher resolution textures whiling reducing the high speed memory requirements. On a PC you can store the main textures in your main pool of system ram and then just send the small bits of the textures that you are drawing on the screen to the GPU for it to use them.

On the consoles the main pool of system ram is also the graphics ram, sure on the X1 you can send the cropped texture data to the ESRAM but is that the best use of the ESRAM? While it might work really well there might be other things a developer wants to use the ESRAM for that are even more bandwidth constrained. What happens if the scene you are drawing actually requires a large chunk of your high res texture so even after you have reduced it down it still requires more than 32 MB of ram? This is why the ESRAM is going to be a headache for devs until the tools improve and the devs learn what to use the ESRAM for and what not to.

PRT will never work with cloud gaming because the latency to send data to and from the internet is too high, you will have unloaded/ low res textures suddenly changing to their nice, crisp high res textures and it will break immersion. 16MB is not a lot of data but the issue is the math required for PRT to work requires latency measured in nanoseconds not milliseconds.

To render a scene at 30FPS requires that the work for each frame takes no longer than 33.3ms, if you want to use cloud based PRT and your ping is 100ms then by the time you get the texture back from the server the texture data is at least 3 frames old, perhaps longer depending on server side latency, download speed and all sorts of other factors that affect the internet.

In the end the PS4 has better graphics hardware. It can push more pixels, deal with better quality textures and throw more or higher quality effects around the place, it is upto devs to use what it is capable of to create breathtaking games and just because the PS4 has an advantage does not mean the Xbox 1 will have ugly games. Behind the numbers there is a lot of subjectivity in graphics because of the artistic nature of the medium.

Exactly theory mean sh** practical use is what matter and you keep throwing a false number again and again,is not 192GB,is 140 to 150 GB/s that is all,and it comes from MS should even that number should be taken with a grain of salt.

No you are the fool and a blind fanboy to,hell look at btk2k2 post GDDR5 and DDR3 have about the same latency,but speed in GGRD5 is way faster,the first post i quote from you had endless crap about how the CPU in the PS4 will be hurt by latency,remember the test i post which you refuse to admit.?

Yeah the PS4 not only has different memory controllers from those on PC,it also has a true HSA design with hUMA,with GDDR5 apparently rather than been an issue to the PS4,who knows if it is the reason why the PS4 CPU even at lower clock speed beat the xbox one CPU.

Your theories are mostly based on fantasies and wishful thinking,time for excuses is over the PS4 is stronger and will remain that way the whole generation,no drive from MS or anything will change that.

The memory design of the X1 takes up more space, is alot slower moving data bigger than those 32 MB of ESRAM, and is hard to utilize. Because of the high transistor count of the ESRAM and the Data Move Engines, the GPU had to be smaller. Also, having a more powerfull GPU on that memory setup would probably be pointless. Most GPU's now a days that is around 2 Tflops and up uses GDDR5.

Sony had made up they're mind using a unified pool of 4GB of GDDR5 a few years back, but it was in the last minute that 8GB was possible.

The problem is that the ESRAM adds another variable to the equation. From a programming standpoint, having less intermediaries is always better. Considering that 100% hardware utilization is near impossible, having more hardware components with separate sets of code for each hardware will only serve to reduce your efficiency.

The 192Gb/s bandwidth is faster than any another bandwidth Sony has, but IMO, it's not really a plus point. The higher bandwidth is mitigated by the fact that a separate set of code has to be written to make use of the ESRAM.

Like a gaming PC, most of PS4's memory bandwidth is allocated towards the GPU.

@tormentos: LOL at least he explains things. An I can say that he definitely knows what he is talking about. Unlike you, who just keep on telling me I’m talking BS but not explaining anything.

You should stop your sh!t because you aren’t proving anything. You are an idiot.

You keep saying, I’m talking BS while you are saying Direct X is an API while Direct 3D is the API. I mean really ?

At least I would take btk2k2 post and use that info in the future.

You are Pathetic.

Get a life.

Stop trying downplay Xbox One because it’s getting old and extremely ridiculous.

I don’t want to be childish.

Grow Up. And accept that both consoles are good Next Gen consoles.

It’s all in your head.

Ps4 ram is massively better.

On top of PS4 also having DDR3 ram also

@btk2k2: I can’t disagree on most of the things you have posted because you have proof and you seem to know what you are talking about.

But I don’t know if this is relevant but doesn’t each specific ram you buy have a certain CAS latency. Meaning you can buy a GDDR5 with lower latency and higher latency. Lower latency being more expensive but only by a small amount due to it not being that important overall. . I saw this on sites where you can order PC parts.

I don’t know if I just understood it wrong.

So how is that theory of the PS4 GPU getting hold by the PS4 CPU when the PS4 one perform better than the xbox one,while GDDR5 has the same or even batter latency than DDR3.?

Haaha

See your whole points were all base on sh** repeat crap that has been shut down long ago,ESRAM,DME all that crap serve for sony 1080 vs 720p the time for excuse is over.

Funny how you even claimed the PS4 have diminishing returns,and how you try to prove without success that the PS4 would stop giving because it would reach its peak and it would give no more,like if that wasn't the case with the xbox one,how you try to downplay 1.84TF while at the same time cheer leading for the damn xbox one who doesn't even have now 1.2TF usable for games.

Your theories are all pull from internet crap that has been debunked,just like the 192GB/s or 218GB/s all theory non practical,yet you keep trowing the number around as if it really mean anything,.

So what new theory you have to downplay the PS4,while making the xbox one seem better than it is.?

See and there you go trying to spin his damn info,yes the same apply to DDR3 not all have the same CAS,but you will assume that MS has the DDR3 with the best CAS while sony will have the GDDR5 with the worse right.?

So the new spin after been owned is that sony may be using GDDR5 with worse CAS than MS..lol

And while you keep trowing all this crap,there are test showing the PS4 CPU performing better,so this all is meaning less since it shows that GDDR5 latency is not affecting the PS4 CPU in any way,all the contrary the PS4 CPU is performing better while been lower clocked...hehehehe

You can buy memory with slightly different memory timings and I do not know what is in the Xbox One or the PS4. They are probably middle of the road in terms of latency and the tRC is going to be about the same for both types of memory. A few ns +/- is not going to give you a noticeable advantage either way or the other.

When buying DDR3 for a PC a lower latency module is likely to achieve better overclocking results vs a higher latency module of the same speed. DDR1600 7-7-7 will more likely reach DDR1866 9-9-9 than DDR1600 9-9-9 even though the performance difference at stock between DDR1600 7-7-7 and DDR16 9-9-9 is almost 0.

the G stands for Standart memory type on any graphics card above 50 dollars

ddr3 is only for cpu

not even a cheap 300 dollar PC works with shitty ddr3 on graphics

@tormentos: GTFO.

Get a life. Other than trying to break people down. because you are failing and no one gives a crap.

lol, fail

@btk2k2: Yeah, I thought so.

Can’t the RAM be a little bit more modified on PS4 and Xbox One compared to the normal RAM.

Because compared to a normal PC it’s not doing much.

That would require the memory chips to be custom made for the PS4 and the Xbox 1 which would add a significant amount to the cost of those chips. Sony and MS would both be better off buying faster memory modules instead of modifying slower modules as faster modules would have more tangible benefits.

You poor sad lemming that epic meltdown...

Any more theories on how the xbox one will over perform while the PS4 will under perform.?

@tormentos: Okay.....

I’m fine..........

Are you ?