GeForce Titan coming end of February

Page 64 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
K

Kippa

Senior member
Dec 12, 2011
392
1
81
BallaTheFeared said:
At this point given the rumored price I'm more interested in just seeing this monster, I will envy anyone who got the wifes OK on a purchase while I begin my task of locating 670s or 7950's who have lost their homes recently :)
Click to expand...

Polygamy, get more than more than one wife. The you can have all the Titans that you wish for. :D Trust me to come up with a cunning plan like that.

Out of interest is there any news on how the Titan will fare OpenCl wise. If it isn't muted like the previous card then I would consider getting it.
 
O

OCGuy

Lifer
Jul 12, 2000
27,224
37
91
RussianSensation said:
It's supposed to be paper launch Feb 18th, reviews published 18th or 19th (not sure which day) and retail availability Feb 24-26th. NV apparently pushed the launch forward by 1 week which is how we ended up at Feb 18th. Feb 20th is the widely rumoured PS4 announcement.
Click to expand...

I really doubt it has anything to do with the PS4 announcement, it is already known, and so it is already priced into AMD's stock.

Believe me, people who care about a PS4 are not the same people as us here at midnight refreshing tech sites for reviews for a single card that costs 3x a console.
 
3

3DVagabond

Lifer
Aug 10, 2009
11,951
204
106
Kippa said:
Polygamy, get more than more than one wife. The you can have all the Titans that you wish for. :D Trust me to come up with a cunning plan like that.

Out of interest is there any news on how the Titan will fare OpenCl wise. If it isn't muted like the previous card then I would consider getting it.
Click to expand...

HD7970vsK20.jpg

Shouldn't be much different than K20, I wouldn't think.
 
Lonbjerg

Lonbjerg

Diamond Member
Dec 6, 2009
4,419
0
0
Funny how OpenCL is used as an "metric"...sinc NVIDIA's OpenCL implementation is so "good"...

Hint: Google CUDA.
 
Atreidin

Atreidin

Senior member
Mar 31, 2011
464
27
86
Lonbjerg said:
Funny how OpenCL is used as an "metric"...sinc NVIDIA's OpenCL implementation is so "good"...

Hint: Google CUDA.
Click to expand...

I'm not sure what you're bitching about. Somebody asked what the OpenCL performance might be like, and somebody answered.
 
K

Kippa

Senior member
Dec 12, 2011
392
1
81
I am curious. The with new Titan card, will performance be theoretically hinderd if you have a PCIe 16x2 slot instead of a PCIe 16x3? My mobo has the old PCIe 16x2 slots, just wondering if I would get a large peformance hit with the new card.

As for the performance, yes I wanted to specifically know about OpenCl and I know CUDA is good on Nvidia cards. :)
 
M

Masahiro

Member
Oct 25, 2011
87
0
66
Kippa said:
I am curious. The with new Titan card, will performance be theoretically hinderd if you have a PCIe 16x2 slot instead of a PCIe 16x3? My mobo has the old PCIe 16x2 slots, just wondering if I would get a large peformance hit with the new card.

As for the performance, yes I wanted to specifically know about OpenCl and I know CUDA is good on Nvidia cards. :)
Click to expand...

Well since the 690 approaches saturating PCI 2.0, and the Titan is rumored to be a little slower, I'd guess you're fine
 
RussianSensation

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
PCIe 2.0 hit is probably going to be 1-2%.
http://www.techpowerup.com/reviews/Intel/Ivy_Bridge_PCI-Express_Scaling/23.html

Masahiro said:
Well since the 690 approaches saturating PCI 2.0, and the Titan is rumored to be a little slower, I'd guess you're fine
Click to expand...

Are you sure? Is there a review that measured this? GTX680 still delivers 94% of the performance of at PCIe 1.1 x16. I think it's going to be a long time before a single GPU runs into a serious bottleneck at PCIe 2.0 x16.

This thread has more views than HD7950's Owner's Thread since summer of 2012. NV seems to have underestimated the demand for $900 single-GPU cards.
 
Last edited:
B

boxleitnerb

Platinum Member
Nov 1, 2011
2,605
6
81
Question about the PCIe bandwidth impact:

Is it due to the fact that there are 2 GPUs or even two cards in different slots or is it about GPU power only, and agnostic to the GPU configuration?
 
O

OCGuy

Lifer
Jul 12, 2000
27,224
37
91
RussianSensation said:
PCIe 2.0 hit is probably going to be 1-2%.
http://www.techpowerup.com/reviews/Intel/Ivy_Bridge_PCI-Express_Scaling/23.html



What? GTX680 still delivers 94% of the performance of at PCIe 1.1 x16. It's going to be a long time before a single GPU runs into a serious bottleneck at PCIe 2.0 x16.

This thread has more views than HD7950's Owner's Thread since summer of 2012. NV seems to have underestimated the demand for $900 single-GPU cards.
Click to expand...

They had no problem selling the GTX280 for over $600, until AMD came in an made them drop the price. That was during the middle of the biggest US economic crash in my lifetime.

These are going to be extremely limited, and they know that enough people who want them actually have the money to get them.
 
Lonbjerg

Lonbjerg

Diamond Member
Dec 6, 2009
4,419
0
0
Atreidin said:
I'm not sure what you're bitching about. Somebody asked what the OpenCL performance might be like, and somebody answered.
Click to expand...


I'm laughing at OpenCL performance being used as a "metric"...those that know, know...those that don't, don't...but they sure are funny to watch.
 
SolMiester

SolMiester

Diamond Member
Dec 19, 2004
5,330
17
76
3DVagabond said:
Interesting that it's OK to call the 680 bandwidth limited now, when it's been pointed out before as a reason to buy Tahiti it didn't matter then. :confused:
Click to expand...

Jesus 3D, it's been known since launch the 680 was bandwidth limited proven when overclocking....grandstanding someone's quote as gospel doesn't help your cause regarding agendas.....
 
S

ShadowOfMyself

Diamond Member
Jun 22, 2006
4,227
2
0
Lonbjerg said:
I'm laughing at OpenCL performance being used as a "metric"...those that know, know...those that don't, don't...but they sure are funny to watch.
Click to expand...

Of course youre laughing, because AMD is better at it... What a surprise
We all know if things change within the next year, youll suddenly hype OpenCL as the greatest thing since the wheel

And the reason this thread is so popular, aside from the ridiculous price, is probably due to the 28nm generation being generally considered "underwhelming" (even though the 7970 GHZ is now almost twice as fast as the 6970), and people being so hungry for a true high end card

Too bad youll have to sell a kidney to get it
 
I

Ibra

Member
Oct 17, 2012
184
0
0
6 GB is useless? But HD 7970 has only 1 GB extra memory and it's superior for high res gaming than GTX 680... Not. When AMD release card with 8 GB memory it become major fact why their card is better that Titan. But it's not going to happen because AMD will go bankrupt.
 
Lonbjerg

Lonbjerg

Diamond Member
Dec 6, 2009
4,419
0
0
ShadowOfMyself said:
Of course youre laughing, because AMD is better at it... What a surprise
We all know if things change within the next year, youll suddenly hype OpenCL as the greatest thing since the wheel

And the reason this thread is so popular, aside from the ridiculous price, is probably due to the 28nm generation being generally considered "underwhelming" (even though the 7970 GHZ is now almost twice as fast as the 6970), and people being so hungry for a true high end card

Too bad youll have to sell a kidney to get it
Click to expand...

You know how little NVIDIA care about OpenCL?
Look into their implementation...and how..."crude" it is ^^
CUDA gets all the attention....and perks...OpenCL is just a checkbox feature in their (NVIDIA) eyes.
 
MrK6

MrK6

Diamond Member
Aug 9, 2004
4,458
4
81
Lonbjerg said:
You know how little NVIDIA care about OpenCL?
Look into their implementation...and how..."crude" it is ^^
CUDA gets all the attention....and perks...OpenCL is just a checkbox feature in their (NVIDIA) eyes.
Click to expand...
Explains nvidia's poor performance with it then. Maybe they should focus on it.
 
SirPauly

SirPauly

Diamond Member
Apr 28, 2009
5,187
1
0
Imho,

I think Khronos is important to nVidia and the President of Khronos and the chair of the OpenCL Workgroup is a nVidia employee!


More performance from OpenCL content would be welcomed!
 
H

HurleyBird

Platinum Member
Apr 22, 2003
2,818
1,553
136
CUDA can get closer to the metal and is only optimized for Nvidia GPUS, of course it's going to be faster. AMD's GPUs would run GPGPU faster too if they had a CUDA alternative that only targeted their GPUs.

That being said, while Nvidia gets a moderate speedup from running CUDA code, it's not going to make up for cases where GCN is several times, if not orders of magnitude, faster than Kepler in openCL. At least, that's what seems to be the case in this study:

http://arxiv.org/ftp/arxiv/papers/1005/1005.2581.pdf

Regardless, I would take those K20 results with a grain of salt because they are saying that GK110 is significantly worse than Gk104 in a lot of the results:

http://clbenchmark.com/compare.jsp?config_0=14378297&config_1=11905948
 
Last edited:
Status
Not open for further replies.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom