• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

VRAM - AMD/Nvidia, why does it differ?

VRAM speed and Bus width, it all amount to the same thing, GigaFlops per second on the Memory threading speed.

With a 384Bit Bus the 780ti has about the same memory performance as the 290/X with its 1250Mhz and 512Bit Bus, the 290/X running the same 1750Mhz memory speed it will be churning out some 460 Gflop/s vs 325 on the 780ti, that will make a difference at any resolution.
But yes that difference will increase as the res goes up, but thats also because the 290/X has 64 Raster Units vs 48 on the 780TI.

The 290/X is predetermined to have very high performance, its held back at present with low component settings. there is more to be had out of it.
By comparison the 780TI has just about every little bit already squeezed out of it.

The 290/X is just more GPU tied up, anchored down. Can't wait to see it let loose.

Memory bandwidth is not rated in flops :rolleyes:
 
Theres some differences in the architectures that make it far from straight forward comparing the bandwidth side of it - tho the capacity side I've not been a fan of nVidia always seem to go slightly too lower and I'm guessing thats to push customers into upgrading sooner rather than later :| as much as it is extra VRAM pushes a substantial extra cost onto already expensive cards.
 
This kind of thing happens all the time. I don't know why people get offended by it. But if you are stay clear of their products. That's your right. Nobody is forcing you to buy them.

Not getting offended but you said you agree with their assumption that 3gb is enough for 4k. You might say you don't agree with that assumption, (regarding 4k but just in general for non 4k resolution - in which case fair enough) but they do and that was my point.

That mentions resolutions ? meaning more than 1, what are the other 'HD Resolutions' that more memory is needed for ? as i see 2560x1600 on there but thats the slowest growth, The next highest res is 1920x1200 which is my res, and 2gb is enough , slide lies.

If you read what i wrote i said 1440P and above, so in that respect the slide does not lie and is correct. 1440p is becoming a fast growing resolution and as such you need a 3gb card for that for optimal performance. Nvidia don't have a slide saying 3gb is enough, but they recommend it on their website for 4k.
 
On the vram front could it be that Nvidia feed the chips more voltage at stock which in turn allows higher speeds. This could always be checked to see if indeed Nvidia are using better memory.

They use the same brand and type Memory chips, there aren't many different types to go round

On the 780 and 290/X they use the same 1500Mhz rated Elpida and Hynix IC's, both AMD and Nvidia have a preference to the better Hynix IC's, the problem with that is they are at times in short supply so they only have the Elpida available to them. what you end up with is random 780's and 290's with either Elpida or Hynix.

While the 290/X comes with the same 1500Mhz IC's AMD have under-clocked theirs to 1250Mhz, they did exactly the same thing with early 7970's, later releasing 1500Mhz versions like the 7970 GE as a performance revision.

Once in the users hands those 1250 / 1375 / 1500Mhz 7970's clocked to 1750Mhz, just about all Hynix IC's on the 7950/70 could do that with a touch on the volts. Elpida its a little more difficult.

The 290's are no different, AMD have held them back because they probably want to release a performance revision at a later date.
 
Got the Hynix ICs on my 780 GHZ - have a bit of an odd relationship with clocking them and the core speed for some reason ~1750Mhz seems to be about as high as they go when the core is clocked but they sit happily at ~2000 if I don't touch the core :S
 
They use the same brand and type Memory chips, there aren't many different types to go round

On the 780 and 290/X they use the same 1500Mhz rated Elpida and Hynix IC's, both AMD and Nvidia have a preference to the better Hynix IC's, the problem with that is they are at times in short supply so they only have the Elpida available to them. what you end up with is random 780's and 290's with either Elpida or Hynix.

While the 290/X comes with the same 1500Mhz IC's AMD have under-clocked theirs to 1250Mhz, they did exactly the same thing with early 7970's, later releasing 1500Mhz versions like the 7970 GE as a performance revision.

Once in the users hands those 1250 / 1375 / 1500Mhz 7970's clocked to 1750Mhz, just about all Hynix IC's on the 7950/70 could do that with a touch on the volts. Elpida its a little more difficult.

The 290's are no different, AMD have held them back because they probably want to release a performance revision at a later date.

Do you not think that they are clocked at 1250Mhz because for a lot of cards, anything over results in a black screen? And in fact, it still seems that many have that without moving the memory and GPU/memory on stock clocks. It appears the memory clocks are at the max.
 
Not getting offended but you said you agree with their assumption that 3gb is enough for 4k. You might say you don't agree with that assumption, (regarding 4k but just in general for non 4k resolution - in which case fair enough) but they do and that was my point.

If they believed it was a true market at present, they'd have cards for it. And that costs money.

Trash talking / trolling / whatever you want to call it is free. And it can impact. So, no, I don't think they do see it as a viable market at the moment.

Don't really have anything else to say on the subject. :)
 
If they believed it was a true market at present, they'd have cards for it. And that costs money.

Trash talking / trolling / whatever you want to call it is free. And it can impact. So, no, I don't think they do see it as a viable market at the moment.

Don't really have anything else to say on the subject. :)

Well their PR Campaign and own Website says different. So whilst you can believe that, you're wrong. You don't start a PR campaign and kick up a big fuss about it if they don't consider it important. You really should take a read at the stink they kicked up over 4k. Since it was discovered that the Hawaii 290 is the card for 4k, they've been pretty quiet.
 
Last edited:
Well their PR Campaign and own Website says different. So whilst you can believe that, you're wrong. You don't start a PR campaign and kick up a big fuss about it if they don't consider it important. You really should take a read at the stink they kicked up over 4k. Since it was discovered that the Hawaii 290 is the card for 4k, they've been pretty quiet.

I'm not wrong because they don't have any cards for it. Unless you consider 3-way SLI 780Ti's as an actual feasible option but again, this is likely to be a tiny portion of the market (which is what I have said all along).

I never said it wasn't important to them, as their planning will be looking well ahead into the future. I said that there isn't a market for 4K at the moment. The lack of feasible hardware would prove that.

Everything you've described that has been done by nVidia is cheap for them to do. Website updates and press releases cost nada in the grand scheme of things so it isn't much effort for them to do this. So for you to say that this indicates that there is currently a feasible market for 4K is a fallacy. I think we'll get there eventually but in a generation or two. Or perhaps even longer.

I'm not really that interested currently but 60Hz 27" screens may perk my interest. At a reasonable price point of course. :p
 
Oh it's probably worth stating for clarity that a mass adoption of 4K would be a wonderful thing for AMD and nVidia so it's clearly in both their interests to get the upper hand in the early battles. :)

Doesn't change the fact that the market for it today is minuscule to the point of being practically non-existent.
 
I'm not wrong because they don't have any cards for it. Unless you consider 3-way SLI 780Ti's as an actual feasible option but again, this is likely to be a tiny portion of the market (which is what I have said all along).

If they don't have cards capable for it why launch a PR campaign and plaster it all over Geforce.co.uk/pcper? Baffles me if im honest. :D ;)

This has been my point all along which there is no argument to. You can say its not important to them, but thats not what they're saying with their PR campaign, website and gpu recommendations.

Now this is my final say on the matter, for the sake of both of us. :p :)
 
Not getting offended but you said you agree with their assumption that 3gb is enough for 4k. You might say you don't agree with that assumption, (regarding 4k but just in general for non 4k resolution - in which case fair enough) but they do and that was my point.



If you read what i wrote i said 1440P and above, so in that respect the slide does not lie and is correct. 1440p is becoming a fast growing resolution and as such you need a 3gb card for that for optimal performance. Nvidia don't have a slide saying 3gb is enough, but they recommend it on their website for 4k.

What you say is true, and i only ever said that the slide you posted lies.

What the slide misinformed people into thinking is that 2gb is not enough for HD resolution enthusiasts, but we have 3gb cards buy ours.

Both companies are as bad as each other with regard to smearing the other but thats what they do.
 
Last edited:
If they don't have cards capable for it why launch a PR campaign and plaster it all over Geforce.co.uk/pcper? Baffles me if im honest. :D ;)

I was using cards as plural for the context of the sentence. A 4-way SLI/CF set up would do nicely at 4K but we're talking Kaap and his card collecting brothers only here really. It's not a real 'market'.

The answer is very simple: because it's in their long term interest for 4K adoption as it will drive GPU sales. If you had any experience in project / business planning you'd understand why the information is already out there and ready. They're trying to drive the market towards 4K for reasons which are not altruistic. AMD are the same when they started talking about 4K at the conferences prior to 290 release.

But I'll say again, having information out there and/or talking about it is not equal to there being a feasible market today.

This has been my point all along which there is no argument to. You can say its not important to them, but thats not what they're saying with their PR campaign, website and gpu recommendations.

That's the key point you're not understanding. You're reading what I'm saying and interpreting it as me saying it's not important to them. I never said it wasn't important to them: I said it costs nothing to talk about 4K but it costs a hell of a lot more to get capable cards out. I then said there's next to no market for it at the moment. Market being, people who are going to buy 4K now and the required GPU power to make it work. Maybe this was where the breakdown in communication occurs. My fault, I often assume everyone I talk to knows about the same things I do which is obviously not true. I know nothing about cars for example :p.
 
Last edited:
Got the Hynix ICs on my 780 GHZ - have a bit of an odd relationship with clocking them and the core speed for some reason ~1750Mhz seems to be about as high as they go when the core is clocked but they sit happily at ~2000 if I don't touch the core :S
Samsung IC's on mine and cant eek an awful lot out of them at all. Getting a balance of core and mem oc proved a bit tricky.
 
You have to remember that the 3GB cards are capable of playing at 4K resolutions and because Nvidia promoted them as 4K, doesn't mean you need every detail fully maxed out to be able to say 4K ready. The frame returns on a single Titan on newer AAA titles wouldn't be worth bothering with, so if you are a single carder, you would need to turn settings down, just to get a playable experience.

I don't care either way but Matt seems to be getting a little wound up with Nvidia's advertising.
 
Theoretically does anyone know how many 780Ti's / Titan's / 290X's you need to get playable frames with max settings at 4K in the usual games?

(assuming VRAM was unlimited)

I think it's about 3 but then I started doubting myself :D
 
Theoretically does anyone know how many 780Ti's / Titan's / 290X's you need to get playable frames with max settings at 4K in the usual games?

(assuming VRAM was unlimited)

I think it's about 3 but then I started doubting myself :D

It is 3, as they sent out 3 Titans to all the reviewers and stated they were 4K ready :D
 
Theoretically does anyone know how many 780Ti's / Titan's / 290X's you need to get playable frames with max settings at 4K in the usual games?

(assuming VRAM was unlimited)

I think it's about 3 but then I started doubting myself :D

If you read my earlier posts you would know. According to Geforce.co.uk you need 2 780 SLI 3gb cards for 4k.
 
If you read my earlier posts you would know. According to Geforce.co.uk you need 2 780 SLI 3gb cards for 4k.

I did read it but it doesn't answer my question. Technically need 1 for 4K. Max settings, is a different kettle of fish.

Of course, to play games at 4K you’re going to need a top-end system, one that features SLI GeForce GTX 770s, GTX 780s, or GTX SLI TITANs. Only these configurations have the performance to power games like Assassin's Creed IV: Black Flag, Batman: Arkham Origins, Call of Duty: Ghosts, and Watch_Dogs to their maximum potential.

Doesn't say two there. It says SLI. 3 is still SLI.

Unless I've missed something.
 
Back
Top Bottom