Where are the 4770's?

[brblx]

why is the 4770 a perfect HTPC card? you're not rendering 3D while you're playing movies.

correct me if i'm wrong, but wouldn't an integrated chipset or low-end dedicated card with mpeg acceleration work just as well?

 

[Stoneburner]

Originally posted by: brblx
why is the 4770 a perfect HTPC card? you're not rendering 3D while you're playing movies.

correct me if i'm wrong, but wouldn't an integrated chipset or low-end dedicated card with mpeg acceleration work just as well?

I like to game on my HTPC as well Empire total war is fun on a plasma screeen. Also, the 4X00 series from ATI supports LPCM sound passthrough which basically alleviates some sound over hdmi concerns. For a regular HTPC purely for TV, integrated is fine.

 

[Keysplayr]

Originally posted by: SlowSpyder

Originally posted by: Wreckage

Originally posted by: TheMeanestGuest
I don't think he was referring to specifically now Wreckage, I'm pretty sure he meant in the past. I admit I could be wrong, but it seems like the most probable conclusion to me.

Actually I think he was just trying to start a flame, but that's ok.

Right now you found GTX295's in stock, but they really haven't been around the last few months, and you know that. Over the last few months they were hard to impossible to fid for sale. Was it because they were paper launched, or because they were just so good that everyone wanted on and Nvidia just couldn't keep up with production?

I'm willing to bet I'll be able to find 4770's in stock here and there, that doesn't mean that them being in stock was the norm since launch.

AMD releases statement on 40nm supply problems.

 

[Eureka]

Did anybody really anticipate such a huge demand for the 4770s? Why are they so popular anyway? It stands between the 4830 and 4850 and costs more than the 4850. People worry way too much about heat and power.

 

[Keysplayr]

Originally posted by: Eureka
Did anybody really anticipate such a huge demand for the 4770s? Why are they so popular anyway? It stands between the 4830 and 4850 and costs more than the 4850. People worry way too much about heat and power.

Some might say that you don't worry about it enough.

 

[Eureka]

Originally posted by: Keysplayr

Originally posted by: Eureka
Did anybody really anticipate such a huge demand for the 4770s? Why are they so popular anyway? It stands between the 4830 and 4850 and costs more than the 4850. People worry way too much about heat and power.

Some might say that you don't worry about it enough.

Touché.

How much more energy efficient is the 4770 over the 4850 anyway? Really worth the 20% premium?

Edit: Whoa what happened here? Forum stroke?

 

[exar333]

Originally posted by: dawgtuff
It looks like the availability of these cards are drying up. NewEgg only has 2 and Mwave has one. Also, Mwave had the reference card sold out with a June 15th ETA, but now it's not listed. What's up?

Nvidia bought them all up to prevent them from hitting the streets?

(JOKE!)

 

[Idontcare]

Originally posted by: bgeh
20% for a chip that small (assuming designers have built in the usual redundancies) is frankly just ugly (TSMC's 40nm process)

20% is kinda typical though for risk-production on a qual device for a new node. Sure the top dog guys like Intel probably never start risk-production with yields that low, but for the rest of the industry this is pretty much SOP at this stage in the process maturity of a new node.

The concerns for both TSMC and their customers doesn't come from the magnitude of near time-zero yield but rather in any misses to quarterly milestones for improving yields. A miss in the yield improvement timetable is the concern as that impacts everyone's projected cost structure and earnings.

Originally posted by: Zstream
It comes down to TSMC stating they can do 40nm, unfortunately it looks like very little testing took place. Nearly impossible leakage just now shows itself. I would imagine they will need a few new set of tools or some crazy engineering/desings are needed.

Undoubtedly the leakage situation has always been there, it just didn't get tweaked and iterated on with enough learning cycles to close the gap to the spec in time for risk-production startup. I've seen this before with Iddq. With so many aspects of development happening in parallel it is not uncommon for some specs to lag others in terms of hitting their milestones, other squeakier wheels get the grease (wafers, resources) causing problems down the road as resource-starved metrics then become the squeaky problems due to knock-on effects.

So its not a case of a leakage issue just now showing itself as if it went uncharacterized and unmonitored to date, but rather this is a case of the planned/expected timely resolution of a known issue (but binnable, so it merely effects yield/cost and not reliability or qual) not happening to schedule.

In yield resolution and node ramps there are standard escalation procedures for handling these disconnects, the line will be put down temporarily while hot lots containing engineering test splits are zipped thru the fab to more rapidly identify an acceptable solution. It is costly, which is why they don't do it unless absolutely necessary, but this is SOP for crisis management in the fab.