• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

ASUS GeForce GTX 670 Direct CU II TOP 2 GB

After your review of the DCUii GTX 680(TOP) I decided to put a preorder on one of those

Now after this review I've started faltering a bit in my faith; if it is the best buy ( obviously).

It all comes down to my preorder on the dcuiiTOP was placed at the price of 3849 Danish Kroner.
That translates to either 3849/7,5 = 513,2 € OR 3849/5,6 = 687,3 $
The cheapest GTX 670dcuiiTOP is priced at 3467 Danish Kroner which means a price of
3467/7,5 = 462,3 € OR 3467/5,6 = 619,1$

The price difference isn't huge - but since I'm having a really hard time deciding between the 2 I thought: hey - why not ask here? :=)

What's your take on the 2 cards when prices are as they are here in DK W1zzard/others?

It should also be mentioned that the gtx680dcuiiTOP is now priced -at the cheapest in DK of 4200Dkr ( 560€ or 750$) as I preordered before they upped their prices by quite a few % (9)
 
..The only time such comments even come out is when poor people don't like to see them leading the industry (same happens when Intel smacks AMD), or AMD/ATI fanboys can't stand seeing them smack the crap out of the competition in gaming product.

All I have to say is put your petty angst aside and maybe you'll realize that in gaming product the value now lies with Nvidia, or, you can stay angry and ignorant, though blissful I doubt. The irony is, the budget shopping consumers with much shallower pockets now have no reason whatsoever to label Nvidia as price gouging elitists, but they still find ways to troll against them with baseless arguments as if they are the evil empire.

You're the one coming across with a wee bit of angst chappy. I can't speak for others but I have no problem with Nvidia. And if there were no yield problems why would their CEO JHH say their was and then secure a deal just recently to get priority supply from TSMC?

And I'll take you up on one statement:
The irony is, the budget shopping consumers with much shallower pockets now have no reason whatsoever to label Nvidia as price gouging elitists

Nvidia hasn't released any budget cards yet. The 670 certainly isn't.

And the general consensus is quite clear that GK104 was meant to be destined for mid range but performed so well in gaming that it was promoted to the top tier. Even the 670 PCB and construction quality should give people a clue as to where it's nature lies.
You say Nvidia aren't price gouging, well they are. They're selling mid range design models at top spec prices to compete against AMD's over inflated prices.

Now both companies have been criticised. There we go.

But back on topic.

The Asus 670 TOP card isn't out yet because it's a very hefty custom tweak. It's coming - be patient. I can't get one (or two :D)... there wont be a water block for it - it'll ruin my aesthetic. :cry:
 
You're the one coming across with a wee bit of angst chappy. I can't speak for others but I have no problem with Nvidia. And if there were no yield problems why would their CEO JHH say their was and then secure a deal just recently to get priority supply from TSMC?

JHH never said they had yield issues. This is the statement from where everything came from.

Decline [of gross margin] in Q1 is expected to be due to the hard disk drive shortage continuing, as well as a shortage of 28nm wafers. We are ramping our Kepler generation very hard, and we could use more wafers. The gross margin decline is contributed almost entirely to the yields of 28nm being lower than expected. That is, I guess, unsurprising at this point,

Lower than expected yields does not equate yield problems at all. It means just that: yields at that point were lower than expected, and thus their Q1 forecast was adjusted to reflect that. At a later point he still said that yields were much better than previous nodes. Notice the plural, better than previous nodes.

All this confusion is very simple to explain IMO. He is talking about the Q1 FY 2013 forecast, about why their forecast changed from that previousy predicted, why the gross margin would be lower than they first "promised". So he clearly states the capacity problem and after that he points out that yields on the new node are lower than expected and of course much lower than 40 nm yields at that point, which is why he says "That is, I guess, unsurprising at this point," right after mentioning lower than expected yields. Yields and especially capacity of 28nm is nowhere near that of 40nm, but their new chips, must be 28nm and so they'll not be able to meet demand as well as in previous quarters and so gross margin will decline. That's it, that's all JHH said, IMO, it's very clear if you take 2 seconds to understand it.

Another misconception is that only Nvidia has yield or capacity issues, and this is said because AMD said they didn't have any yield or capacity problems, but that's entirely relative. Come on, does anyone really and honestly believe that AMD and Nvidia have to meet the same demand? First of all Nvidia snatched the Apple deal this round, that is a HUGE amount of chips that need to be delivered. And they also said Kepler has more OEM wins than any other Nvidia generation before.

And regarding AIBs:

http://www.neoseeker.com/Articles/Hardware/Reviews/Nvidia_GTX_670/16.html

Another aspect AMD touched on was the availability of their Southern Islands cards, which is sort of a red herring. It is true AMD's 7000 series graphics cards are more readily available in the market, but the fact of the matter is this has nothing to do with production. Instead the issue is Kepler based graphics cards are in high demand. We have talked to a few retailers and the word from them is Kepler is selling at nearly a 4 to 1 ratio over Southern Islands. This leaves plenty of volume available for AMD, but makes it appear as if Nvidia is lacking the same volume, when this is not necessarily the case.
 
Hi! Should I just buy the gigabyte windforce 3 now while its in stock, or wait for the Asus GTX 670 DCUII to show up? this review makes me really want the Asus card.
 
Hi! Should I just buy the gigabyte windforce 3 now while its in stock, or wait for the Asus GTX 670 DCUII to show up? this review makes me really want the Asus card.

while i havent tested the gigabyte, i would wait for asus, if it were my money
 
JHH never said they had yield issues.

As CEO of a large company he can't say 'the process is a problem' (shares would plummet) but as intelligent human beings we understand his point. He said this...

The gross margin decline is contributed almost entirely to the yields of 28nm being lower than expected

If I plan for 'X' amount as a business and my gross margin suffers because my expected yield is not as high, then that is a problem/issue.

JHH said himself gross margin decline is due to a shortage of wafers. We've both quoted him. It's certainly not going to destroy NV and it's certainly not as bad as 40nm was at first (picked up pretty well in fact) but if the CEO says:

The gross margin decline is contributed almost entirely to the yields of 28nm being lower than expected

then it's an issue. There is no logical way you can argue against that.

They know how many design wins etc they have so they plan for 'x' amount of wafers. They dont get enough so they have to renegogiate with TSMC to get priority status.

It's obvious they have a problem with yield - this is why they are fixing it with an increased capacity from their latest deal with TSMC.

Jeez.....
 
JHH never said they had yield issues. This is the statement from where everything came from.



Lower than expected yields does not equate yield problems at all. It means just that: yields at that point were lower than expected, and thus their Q1 forecast was adjusted to reflect that. At a later point he still said that yields were much better than previous nodes. Notice the plural, better than previous nodes.

All this confusion is very simple to explain IMO. He is talking about the Q1 FY 2013 forecast, about why their forecast changed from that previousy predicted, why the gross margin would be lower than they first "promised". So he clearly states the capacity problem and after that he points out that yields on the new node are lower than expected and of course much lower than 40 nm yields at that point, which is why he says "That is, I guess, unsurprising at this point," right after mentioning lower than expected yields. Yields and especially capacity of 28nm is nowhere near that of 40nm, but their new chips, must be 28nm and so they'll not be able to meet demand as well as in previous quarters and so gross margin will decline. That's it, that's all JHH said, IMO, it's very clear if you take 2 seconds to understand it.

Another misconception is that only Nvidia has yield or capacity issues, and this is said because AMD said they didn't have any yield or capacity problems, but that's entirely relative. Come on, does anyone really and honestly believe that AMD and Nvidia have to meet the same demand? First of all Nvidia snatched the Apple deal this round, that is a HUGE amount of chips that need to be delivered. And they also said Kepler has more OEM wins than any other Nvidia generation before.

And regarding AIBs:

http://www.neoseeker.com/Articles/Hardware/Reviews/Nvidia_GTX_670/16.html


I take it you read the Q&A..

The forecast isnt just the kepler but all of its entities Tegra, HPC, Mobile & GPU. They had to adjust it cause aside from HPC none of them are doing as forcasted.

Jen-Hsun Huang
We have an understanding about wafer prices from TSMC and we negotiate with them and we now both understand where the pricing is

Jen-Hsun Huang
And so I expect us to be supply constrained from wafers at the wafer level, and we'll continue that way throughout the end of the year


You cant meet demand if you arent producing enough cause you cant buy wafers. He was also question on that but as you know these things are a pitch to investors.

He doesnt want to be downgraded again like in Jan/Feb while the company keeps bleeding money.
 
Last edited:
As CEO of a large company he can't say 'the process is a problem' (shares would plummet) but as intelligent human beings we understand his point. He said this...



If I plan for 'X' amount as a business and my gross margin suffers because my expected yield is not as high, then that is a problem/issue.

JHH said himself gross margin decline is due to a shortage of wafers. We've both quoted him. It's certainly not going to destroy NV and it's certainly not as bad as 40nm was at first (picked up pretty well in fact) but if the CEO says:



then it's an issue. There is no logical way you can argue against that.

They know how many design wins etc they have so they plan for 'x' amount of wafers. They dont get enough so they have to renegogiate with TSMC to get priority status.

It's obvious they have a problem with yield - this is why they are fixing it with an increased capacity from their latest deal with TSMC.

Jeez.....

They are not in a worse situation than other players. It's not an Nvidia issue at all, it's an issue everyone is having. Pay attention and you'll see they simply aren't in any worse situation than others.

Yes if you plan on being able to produce 10 million (random number) cards in a quarter, but you can only get 5 million that's an issue, one that you basically created for yourself. But it doesn't mean you have an issue with production. Not when you were only able to sell 2 million in your previous generation in the same timeframe and when you're outselling your direct competitor in a 4 to 1 ratio. You have no real issue, again in production/allocation, you just aimed too high.

An issue is when your gross margin suffers badly compared to previous exercises, i.e. versus Q1 FY 2012. When your gross margin suffers against your overly optimistic forecast, your only issue is with your investors confidence and maybe an issue with reality (well and SEC if wrongdoing/deception can be demostrated). But no matter how much you expected if you have 50% yields, and everybody else has 50% yields, YOU don't have yield issues. Simply because you expected/wanted/sold to investors a 75% yield figure, based on what your production partner told you would be posible, that doesn't make it your yield issue. Again lower than expected != bad yields. Bad yields is when yields are lower than normal/usual/previous nodes, not when yields fail to be better than they usually are, no matter how much these better than normal yields are expected to be.

They know how many design wins etc they have so they plan for 'x' amount of wafers. They dont get enough so they have to renegogiate with TSMC to get priority status.

And that's a volume problem not a yield problem. They had to renegotiate because they are selling or want to sell 4x as much as AMD does and with current wafer allocation that's imposible. The issue has always been that they need to move many chips and there's simply not enough wafer starts. Plain and simple. And yes in that sutuation if you expected for example 40% yields and you get 37%, your gross margin will definitely suffer and that still does not mean you have yield issues. You have yield issues when you are way below that mark.

I take it you read the Q&A..

The forecast isnt just the kepler but all of its entities Tegra, HPC, Mobile & GPU. They had to adjust it cause aside from HPC none of them are doing as forcasted.

You cant meet demand if you arent producing enough cause you cant buy wafers.

And that again is an issue with volume and not yields, so your post is redundant to what I've been saying. Nvidia is expecting to only have a 30% of its production be 28nm by the end of the next quarter and have doubts of reaching 50% by the end of year. This when their entire future (Kepler, tegra 3+, etc.) depends on the node.

AMD, Qualcomm, everybody is in the exact same situation, but there's only 2 GPU vendors and the last thing AMD wants to say is "we wouldn't be able to meet demand, if not for the fact that there is NO demand for our products, come on ladies and gentlemen, buy our shares". No they say "we are able to meet demand" and it's even essentially true, just not the truth that they really want to be true.
 
Last edited:
And that again is an issue with volume and not yields, so your post is redundant to what I've been saying. Nvidia is expecting to only have a 30% of its production be 28nm by the end of the next quarter and have doubts of reaching 50% by the end of year. This when their entire future (Kepler, tegra 3 , etc.) depends on the node.

AMD, Qualcomm, everybody is in the exact same situation, but there's only 2 GPU vendors and the last thing AMD wants to say is "we wouldn't be able to meet demand, if not for the fact that there is NO demand for our products, come on ladies and gentlemen, buy our shares". No they say "we are able to meet demand" and it's even essentially true, just not the truth that they really want to be true.

I wasnt responding to the yields issue i was responding to the Forecast.

I dont think you can equate those two companies with the volume of 28nm that Nvidia was try'n to allocate which is much smaller. Those two companies cover more products and markets geographicly. Especially Qualcomm.

Jen-Hsun Huang admitted in the Q&A it was poor planning on not allocating the wafers for whatever reason. Tegra 3 has been in the works longer then Kepler so purchase or estimation of purchasing or allocating wafers comparable to the two companies would only mean production not allocation or purchase of said wafers. Would fall in-line as to the supposed talks with Samsung but thats like AMD going to Intel for wafers.

If Jen-Hsun Huang was so adimately 100% sure of the margins if Nvidia had been able to purchase the wafers it needed to comply with said demand. He looked like an amature when JP Morgan asked about it.

Harlan Sur - JP Morgan Chase & Co, Research Division

Great. And then obviously, you mentioned in your prepared remarks about under shipping demand in both the desktop and notebook segments in Q1. Can you just quantify roughly dollar-wise how much more revenue you could have shipped in Q1? And given obviously the constraints here in Q2, what the rough dollar impact on July sales could be as well.

Jen-Hsun Huang

We haven't broken it out, but it's a lot.

It's an earning call and not having potential earning numbers when talking to investors is a no-no

Just seams they dropped the ball on the roll out where it had potential to get it back on track and maybe have them stop loosing money.
 
Sorry guys, didn'T read all the post, but I would like to know if this card has RAM cooling? It looks like a big NO, and this worries me.. Overclocking ram, fine but if not cooled... For this I prefer the Zotac AMP I think (but cooler seems bigger).
 
Well so basically that's what I've been saying. I'm going to try this one to be my last post on the matter since it is way off-topic. And keep in mind that as always this is only my view, my opinion. Thoughtful and as based on evidence as posible, but just an opinion in the end.

Nvidia's only issue is that of over-promising and then under-delivering, due to both external and internal influences/problems. And yields have nothing to do on this, they do not have abnormally lower yields, looking at what evidences I could find.

What must be understood tho, if any conclusions are to be made from what is said in conference calls, is that in the GPU scene we have 2 players and it's very important to know who they are and in what situation they are.

AMD: Aside from the long standing situation that AMD has been living on for the past years, AMD lost the Apple deal that they had in previous generations and they have gained very few other OEM design wins. Not enough to make up the difference anyway. Their message is thus that of abundance. "We can meet demand" == "you can place orders". They want/need new customrs. Don't let the words make believe you that I'm saying this is the doom&gloom for AMD. They are just in a situation where they want more contracts, and their message reflects that.

Nvidia: Got Apple back and have more design wins than ever before. They are not able to meet demand for several reasons and thus their instance is that of keeping the customers and keeping them happy. And so "finding excuses" for underdelivering. Lack of wafer starts is THE reason as well as the fact they were not able to allocate enough of the available wafers, but both of these only translate to "cannot meet demand" and "not completely on their hands", as well as "the situation will hardly change in the near future". And that's not the message that Nvidia wants, so "yieds lower thn expected" is a good "excuse" considering the circumstances, because the promise of higher yields is always there for new nodes (not that it's false that yields are lower than expected, but neither is true the message that has been conveyed from this).

It does not mean anyone is lying, but everyone uses what is best for them. Wafer starts are very low, which is the innevitable truth and yields are lower than expected (which does not equate bad yields or yield issues). AMD is apparently able to meet demand which does not mean that they have good yields and plenty of wafer allocation. Devil is in the details.
 
I wouldnt put too much hope on the Apple contract. Its good they got it back but they were in the tablet business and its there way of putting there foot in the door in a big way. Many forcasters think the tablets are on there way out with lighter slimmer ultrabooks they will be competing for similar clientel. Thats why the new iPad is going to be smaller in size after the slim revision on iPad 3.

Would have been better if they kept Sony Playstation contract for a steady stream of reveneu or able to snatch Nintendo Wii-U or Microsoft Durango from AMD. Now having all 3 console platforms go to AMD it gives them a advantage in the future having a revenue stream even if its light and a closer relationship with game developers. Which I suspect Nvidias announcement of its new program to collect users info is an attempt to counter. That worries me more because Developers will be inclined to streamline their development and aim a set # of audience in the console and develope PC games along side that. The Way Its Ment To Be Played delepore assistance program might vasnish soon after the consoles hit the market or might become a financial liability to Nvidia.


Back on topic...

I called two local retail vendors that had a 670 not OC model too for $489 <- WTF :eek:. Gigabyte and EVGA both out of stock but I wanted to know why the high price.
 
I wouldnt put too much hope on the Apple contract. Its good they got it back but they were in the tablet business and its there way of putting there foot in the door in a big way. Many forcasters think the tablets are on there way out with lighter slimmer ultrabooks they will be competing for similar clientel. Thats why the new iPad is going to be smaller in size after the slim revision on iPad 3.

Would have been better if they kept Sony Playstation contract for a steady stream of reveneu or able to snatch Nintendo Wii-U or Microsoft Durango from AMD. Now having all 3 console platforms go to AMD it gives them a advantage in the future having a revenue stream even if its light and a closer relationship with game developers. Which I suspect Nvidias announcement of its new program to collect users info is an attempt to counter. That worries me more because Developers will be inclined to streamline their development and aim a set # of audience in the console and develope PC games along side that. The Way Its Ment To Be Played delepore assistance program might vasnish soon after the consoles hit the market or might become a financial liability to Nvidia.


Back on topic...

I called two local retail vendors that had a 670 not OC model too for $489 <- WTF :eek:. Gigabyte and EVGA both out of stock but I wanted to know why the high price.

Onle two models left at newegg, hurry!
 
Got an MSI OC version Friday night , should be here Wednesday or Thursday.
They are selling out fast. I don't think there are any of the OC versions left.
 
I'm confused cant you over clock a card yourself?

when the hell does the ASUS DirectCu II TOP come out!?
 
I just want two of these bad boys in my rig right meow!!!!!!!!!!!!!:banghead:
 
I'm confused cant you over clock a card yourself?

when the hell does the ASUS DirectCu II TOP come out!?

I put in an order for the non-TOP model on day 1 an now the retailer put in "coming: over 25 pcs" on current stock , YEY. hopfully it'l be on my doorstep soon :).

And they also put in TPU seal of approval on the TOP model picture ;) Link
 
I put in an order for the non-TOP model on day 1 an now the retailer put in "coming: over 25 pcs" on current stock , YEY. hopfully it'l be on my doorstep soon :).

And they also put in TPU seal of approval on the TOP model picture ;) Link

Holy crap batman! 449 euro ftw!!
 
As to the issue of yield when they can make so many variants off the single node. The question of yield is not the same as "normal" when a customer anticipates say two different decent chips and may be a true gelding off a node. Nvidia can make lot's different variants off one GF104 node, so yes a wafer yield is good, but not offering enough "top-shelf parts", which appears to be the situation.

Jen-Hsun Huang said, "gross margin decline is contributed almost entirely to the yields of 28nm being lower than expected". The high margin chips aren't there, and they need more capacity of the on 28Nn because it's being used for every card they want to release.
 
On the last page:

Not only noise is low but temperatures are great, too. With only 74°C under load the card is quieter than most other GTX 670 cards tested today.

Shouldn't it say cooler instead of quieter?
 
Wow. I said I wasn't replacing my 6950 until later this year... Well. Now I just wanna go to my usual e-tailer and order this thing right away. :eek:
 
Back
Top