NVIDIA announces Titan X, a GPU featuring 12GB framebuffer and 8 billion transistors

[sam_738844]

You have no idea what you are saying or how these silicon chips are developed or manufactured. Use google and enlighten yourself before commenting like a troll.

Graphics chips or processors which belongs to a specific series are all the same with minute difference. The only difference is that the execution units or specific features are disabled on chips that goes into lower end parts.

And the chips that goes into higher end processor or graphic card are the best performing lot in the same batch. Or without any features disabled.
For eg an i5 4590 is same chip used in i5 4690 or 4690k. Intel just choosed to clock them lower for a price difference. An i5 is the same chip as i7 with just hyperthreading disabled in the same series like devil canyon.

This certainly dont mean that cost are increased for intel for manufacturing an i7 over i5 or i3. It just means that they demand a premium for the same chip.

Same goes for nvidia. The higher end card in the same series doesn't mean the additional cost of more than 2times.

Don't make yourself foolish by beleiving whatever your friends or shopkeepers have to say to sell the product. And before commenting do some research.

google yourself and find the result as a blabbering baboon or another reddit fanboy. You dont work in nvidia, you dont work in intel. You dont know a single crap about their Marketing policies or why they quote a premium in their OTS products. Technology comes with a price. Enabling and disabling cores is nothing new. Read more and find that it's come to light that the GTX 980 sports 2MB of L2 cache, the 970 is limited to 1.75MB. The GTX 980 features 64 ROPs, but the 970 is limited to 56 thanks to a partially-disabled cluster. In addition, the 970's memory is segmented into 3.5GB and 500MB chunks with 196 GB/s and 28 GB/s of bandwidth, respectively. Since these chunks can not be striped, the maximum usable memory bandwidth for the 3.5GB chunk is is 196 GB/s.

Why on earth a full blown chip should cost almost the same as a disabled one?

Read this too form tom's

Unlike GM107, the GM204 GPU features four Graphics Processor Clusters (GPCs) instead of one. That means it benefits from four times the number of raster engines. Of course, high-end graphics cards require a beefier back-end to handle all of that data throughput, and the GeForce GTX 980 utilizes four render back-ends capable of handling 16 full-color ROP operations per clock, adding up to 64. Four 64-bit memory controllers create an aggregate 256-bit bus. By the way, you may have noticed that the GeForce GTX 970's 13 SMMs don't divide equally into four GPCs. Nvidia says that there is no predefined recipe of SMMs per GPC in the 970, and each GPU may be configured differently.

There is no age for education, there is no shame in reading more and learning.

Quoting from Anand's

As expected, along with the reduction in SMMs clockspeed is also reduced slightly for GTX 970. It ships at a base clockspeed of 1050MHz, with a boost clockspeed of 1178MHz. This puts the theoretical performance difference between it and the GTX 980 at about 85% of the ROP performance or about 79% of the shading/texturing/geometry performance. Given that the GTX 970 is unlikely to be ROP bound with so many ROPs, the real world performance difference should much more closely track the 79% value, meaning there is a significant performance delta between the GTX 980 and GTX 970.

With your logic tomorrow I buy a honda with a partially disabled subaru engine which magically costs less and the latter is way more expensive just because it has some shafts or whatever enabled?

 

[gagan_kumar]

google yourself and find the result as a blabbering baboon or another reddit fanboy. You dont work in nvidia, you dont work in intel. You dont know a single crap about their Marketing policies or why they quote a premium in their OTS products. Technology comes with a price. Enabling and disabling cores is nothing new. Read more and find that it's come to light that the GTX 980 sports 2MB of L2 cache, the 970 is limited to 1.75MB. The GTX 980 features 64 ROPs, but the 970 is limited to 56 thanks to a partially-disabled cluster. In addition, the 970's memory is segmented into 3.5GB and 500MB chunks with 196 GB/s and 28 GB/s of bandwidth, respectively. Since these chunks can not be striped, the maximum usable memory bandwidth for the 3.5GB chunk is is 196 GB/s.

Why on earth a full blown chip should cost almost the same as a disabled one?

Read this too form tom's




There is no age for education, there is no shame in reading more and learning.

Quoting from Anand's




With your logic tomorrow I buy a honda with a partially disabled subaru engine which magically costs less and the latter is way more expensive just because it has some shafts or whatever enabled?

actually intel does sell cpus with core disabled from their higher model....

 

[sam_738844]

actually intel does sell cpus with core disabled from their higher model....

of course they do, thats the point. The reason you are paying more IS necessarily to have more working cores, more IPC.

 

[REDHOTIRON2004]

of course they do, thats the point. The reason you are paying more IS necessarily to have more working cores, more IPC.

Listen sam no point in getting hyper. Your previous reply was too big to quote. If you dont know something then accept it and move forward.

The whole point of me giving you the intel example was to make you understand that the cost for the companies doesn't increases that exponentially from a lower to upper end chip in the same series and sometimes it doesn't increase at all, that nvidia would start asking for more than 2 times premium on the same silicon.
That was the point all through. You mentioned in your post that they include some new technology and different silicon etc2 that increase there cost. Which is not the case here.
Just to increase your understanding(if you want to understand) an intel i3 or i5 or i7 in the same series are basically the same chips with features disabled or enabled according to the chip designation. Like for eg: i3 have hyperthreading enabled while i5 have 4 discrete cores enabled with hyperthreading disabled. In i7 hyperthreading is enabled on all 4cores making it 8 cores. But the chip remains the same.

Either understand the facts. Or stop babling about the justification of your favorite companies in asking exuberant prices for the same silicon.

There is no point in justifying what they are asking for when the real point of discussion was what is the additional cost incurred in making the new(in your words) chip which is nominal in comparison to what is being asked.

You can give whatever you want to give those companies for your imaginary high end blue chip that your friends or sales man told you. I don't care. But thats not the point here.

 

[sam_738844]

Listen sam no point in getting hyper. Your previous reply was too big to quote. If you dont know something then accept it and move forward.

The whole point of me giving you the intel example was to make you understand that the cost for the companies doesn't increases that exponentially from a lower to upper end chip in the same series and sometimes it doesn't increase at all, that nvidia would start asking for more than 2 times premium on the same silicon.
That was the point all through. You mentioned in your post that they include some new technology and different silicon etc2 that increase there cost. Which is not the case here.
Just to increase your understanding(if you want to understand) an intel i3 or i5 or i7 in the same series are basically the same chips with features disabled or enabled according to the chip designation. Like for eg: i3 have hyperthreading enabled while i5 have 4 discrete cores enabled with hyperthreading disabled. In i7 hyperthreading is enabled on all 4cores making it 8 cores. But the chip remains the same.

Either understand the facts. Or stop babling about the justification of your favorite companies in asking exuberant prices for the same silicon.

There is no point in justifying what they are asking for when the real point of discussion was what is the additional cost incurred in making the new(in your words) chip which is nominal in comparison to what is being asked.

You can give whatever you want to give those companies for your imaginary high end blue chip that your friends or sales man told you. I don't care. But thats not the point here.

Stop refering stuff which is irrelevant here. The discussion was about GPUs, stick to that. It doesnt matter whether its a same chip, different chip, same tech different tech, node shrink or whichever potential optimization or enhancements a chip maker incorporates to make their higher-end products targeted to a specific market segment. The point is, there is a discernible and distinguishable performance contrast, and it calls for a price.

Nothing is free, whether the price delta is justifiable/exorbitant is altogether a different argument, that applies to the price point of the baseline itself, translating to another intel-amd-nvidia price-performance for value war over and over again. The purpose of posting all these is to foreground the fact that extra performance directly translates to extra price. Period.

I dont need friends or salesmen to convince me otherwise.

Meanwhile after GDC...

NVIDIA GeForce GTX Titan X Performance Benchmarks Unveiled - Massive Performance Uplift Versus GTX 980, 2/3/4-Way SLI Results

 

[warfreak]

Meanwhile after GDC...

NVIDIA GeForce GTX Titan X Performance Benchmarks Unveiled - Massive Performance Uplift Versus GTX 980, 2/3/4-Way SLI Results

Wow! more than 200% for SLI and more than 300% & 400% at 3-way and 4-way respectively? Sounds too good to be true!

AFAIK SLI only provided 150% performance gain at 200% price. 3-way and 4-way SLI actually degraded performance in some cases. If these numbers are true, then we will finally be able to see some real performance benefits of SLI.

I hope it provides same results in real-world gameplay tests.

 

[REDHOTIRON2004]

30% performance bar is a lot in silicon performance-space. To yield that amount of performance in same node, the engineering behind cores and memory doesnt come cheap. You are buying a piece of eletronics not Annapurna Aaata from grocery shop where twice the money is...twice the amount.

The cost behind the silicon that goes into same series higher end and lower end chip doesn't change or change nominally. Even if nvidia is selling the same silicon in lower end designation then also they are making a profit irrespective of engineering etc that had gone into it. Your examples about the atta etc are foolish.

of course they do, thats the point. The reason you are paying more IS necessarily to have more working cores, more IPC.

Your are now jumping on another point that refers to premium being asked for the same thing. And its not about the cost as you mentioned in your first post. But, surely its better than your annapurna atta example.lol

Stop refering stuff which is irrelevant here. The discussion was about GPUs, stick to that. It doesnt matter whether its a same chip, different chip, same tech different tech, node shrink or whichever potential optimization or enhancements a chip maker incorporates to make their higher-end products targeted to a specific market segment. The point is, there is a discernible and distinguishable performance contrast, and it calls for a price.

Nothing is free, whether the price delta is justifiable/exorbitant is altogether a different argument, that applies to the price point of the baseline itself, translating to another intel-amd-nvidia price-performance for value war over and over again. The purpose of posting all these is to foreground the fact that extra performance directly translates to extra price. Period.

I dont need friends or salesmen to convince me otherwise.
[/url]

You are jumping all over the place now. And listening to your following comments, I too agree with you that you actually don't need your friends or salesman to make you a fool. You are better at doing that yourself.

Wow! more than 200% for SLI and more than 300% & 400% at 3-way and 4-way respectively? Sounds too good to be true!

AFAIK SLI only provided 150% performance gain at 200% price. 3-way and 4-way SLI actually degraded performance in some cases. If these numbers are true, then we will finally be able to see some real performance benefits of SLI.

I hope it provides same results in real-world gameplay tests.

Ok, looks interesting. But, I doubt about the sli performace. A 200% increase is impossible IMHO. As there is always going to be some redundancy or communication gap between the two gpus working together vs a single gpu.

Also, synthetic benchmarks are indicative only. Real world performace is always much different. Also, the scaling is also going to be different in different games.
Still, I would say that nvidia might have improved/scaled there sli performace quite nicely if those benchmarks are true.

 

[gagan_kumar]

somewhere i heard we can use gpus form different comapnies working together also like amd + nviidia , is it so?

 

[Desmond]

somewhere i heard we can use gpus form different comapnies working together also like amd + nviidia , is it so?

DirectX 12 will support that. That's what Microsoft claims.

 

[sam_738844]

The cost behind the silicon that goes into same series higher end and lower end chip doesn't change or change nominally. Even if nvidia is selling the same silicon in lower end designation then also they are making a profit irrespective of engineering etc that had gone into it. Your examples about the atta etc are foolish.



Your are now jumping on another point that refers to premium being asked for the same thing. And its not about the cost as you mentioned in your first post. But, surely its better than your annapurna atta example.lol



You are jumping all over the place now. And listening to your following comments, I too agree with you that you actually don't need your friends or salesman to make you a fool. You are better at doing that yourself.



Ok, looks interesting. But, I doubt about the sli performace. A 200% increase is impossible IMHO. As there is always going to be some redundancy or communication gap between the two gpus working together vs a single gpu.

Also, synthetic benchmarks are indicative only. Real world performace is always much different. Also, the scaling is also going to be different in different games.
Still, I would say that nvidia might have improved/scaled there sli performace quite nicely if those benchmarks are true.

You are fun at parties aren't you? i5-4590K, i7-4790K, 2MB more cache, hyper threading, higher clock and boost. 100$ price premium, i dont know whether its justified, i dont care, but i know its not free.
8+% Faster in single core performance.
7% Faster in dual
15% Faster in multicore performance.
+42% faster OC multi-core speed.
+41% geekbench
+18% Stronger in Single Threading
+52% Stronger in Multithreading.

Based on Passmark, Passmark ST and Geekbench

Capisce?

 

[warfreak]

I need to make my point before this thread converts into a flamewar and gets locked by mods.

The sole reason nVidia price a premium is because they have no competition up there. nvidia has been ruling the top-end, enthusiast market for quite a while now. That is the reason it has been able to be less aggressively priced.

Take a look at the mid range and AMD starts showing its teeth. Hence the 970 is priced such.

I admit it is unfair on nvidia's part to have such a huge gap between the prices of 970 and 980 but we also have to understand that the system integration process becomes more and more complicated with each new iteration with the fabrication process getting smaller and smaller. It's not just about a piece of silicon, its about how it is designed to store billions of transistors in such a constricted space.

 

[sam_738844]

^^finally someone with sense. Peace out and apologies for clobbering.

 

[sam_738844]

First Reviews are out.

*www.digit.in/forum/graphic-cards/117078-gpu-news-channel-146.html#post2210551

 

[Zangetsu]

will this be the final segment of GPU to play everything in Ultra Settings for a long time

 

[sam_738844]

will this be the final segment of GPU to play everything in Ultra Settings for a long time

R9-390X is around the corner

 

[Zangetsu]

R9-390X is around the corner

Hmm lets see who beats who

 

[sam_738844]

Hmm lets see who beats who

if R9-390X fails to beat titan X by a good margin, then its pretty much game over for AMD again in the next Fiscal. They have played lowball so damn low that it had adversley affected their growth margin in previous quarter. With all those good cards, they only have claimed 24% market share, they have cut down price too much to even for AMD standards,pretty bad. Rest is dominated by nvidia overpriced or not.