AMD: Speculation, Rumors, and Discussion (Archive)

[CSI PC]

That is another facepalm....
How much is that eating into their margins for the entry 4GB model, that are already tight at this price, and if true and word gets further out it could see 8GB model sales stall (relative to 4GB model) on the product that actually is priced with 8GB component, exacerbating AMD's product margin.
So now the engineering bad news is followed by some business (not for consumer but for the manufacturer) bad news.
I am feeling for AMD at the moment, really seems there is a disconnect within the divisions and teams.
Cheers

 

[dskneo]

isn't the final bandwidth determined by the number of gddr5 chips? how are they maintaining 256gb/s with only half the chips active?
I'm missing something.

also, I saw the overclocking stream yesterday, I could have sworn he mentioned the gddr5 was pulling power exclusively from pci-e and not from the 6pin as was said in the previous page.

 

[AlNom]

Wouldn't they just be limiting addressing ( per chip)?

 

[Grall]

I don't believe in magic so I doubt I driver update can fix that.

The voltage regulator controller needs to reduce the duty cycles for the phase(s) hooked up to PCIe power, and increase duty cycles for 6-pin molex phases. If that can be done (and the controller AMD is using is supposedly one of the best in the business), then it's doable.

Of course, that means pushing more power draw onto the 6-pin connector, so that would bring that one (further, possibly) out of spec. So not exactly a superduper solution on the whole. Best would be to magically grow an 8-pin connector instead for 150W headroom on the molex connector, but that one is not doable in a software patch for sure!

 

[mczak]

Probably because some of the first buyers of the 4GB started to report, that there are indeed 8GB of Samsungs 8Gbps GDDR5 on them (only a BIOS flash needed to activate the full 8GB?).

I would think it is far simpler. AMD wanted to have some 4GB cards to launch (so that this variant wasn't just a paper launch), and they didn't build any (for whatever reason, you could just literally throw 4 gbit chips into your assembly line instead of the 8 gbit ones). And they just "downmodded" a very limited number of 8GB cards down to 4GB, they're sold out now and that's the end of it (only very few retailers seemed to got them in the first place).

 

[xEx]

The voltage regulator controller needs to reduce the duty cycles for the phase(s) hooked up to PCIe power, and increase duty cycles for 6-pin molex phases. If that can be done (and the controller AMD is using is supposedly one of the best in the business), then it's doable.

Of course, that means pushing more power draw onto the 6-pin connector, so that would bring that one (further, possibly) out of spec. So not exactly a superduper solution on the whole. Best would be to magically grow an 8-pin connector instead for 150W headroom on the molex connector, but that one is not doable in a software patch for sure!

The thing is that even if you grow that 8 pin with black magic you will still have problems seems 4 of the phases are bounded to the PCE slot instead of the 6 pin. Idk how AMD will change that with just a driver update(without throttle the card to 380 perf lvls) but we will see tomorrow.

 

[ninelven]

Of course, that means pushing more power draw onto the 6-pin connector, so that would bring that one (further, possibly) out of spec. So not exactly a superduper solution on the whole.

Still the best option overall (if AMD does not intend to recall the cards). That way the only thing that can really be negatively affected is the PSU itself, and most should have some form of protection to prevent anything disastrous. And single rail PSUs should be just fine. OFC, they definitely should not keep producing an out of spec card...

 

[Grall]

could be that te demand in the UK is really low? in the US there have been no stock since july 2.

Please note that july 2 was only 2 days ago, and those days were a weekend, and today is a national holiday in the U.S. (in fact THE national holiday, traditionally celebrated by murricans all over the country sticking fireworks up their bums and setting them off.)

So realistically, not much stock is going to come in until at least tomorrow at the very earliest...

The thing is that even if you grow that 8 pin with black magic you will still have problems seems 4 of the phases are bounded to the PCE slot instead of the 6 pin.

It's a curious engineering decision for sure, but the phases are built out of heavy-duty components and good for quite a bit of power. It might be doable as long as the VRMs stay below 100C... *shrug*

 

[CSI PC]

I would think it is far simpler. AMD wanted to have some 4GB cards to launch (so that this variant wasn't just a paper launch), and they didn't build any (for whatever reason, you could just literally throw 4 gbit chips into your assembly line instead of the 8 gbit ones). And they just "downmodded" a very limited number of 8GB cards down to 4GB, they're sold out now and that's the end of it (only very few retailers seemed to got them in the first place).

Yeah that is a good point and possibility.
They could had done that to say they had hard launched the 4GB 480, and so emphasise it did exist at $199, even if it is not technically hard launching for a further 4 weeks.
Would offset why they would sell the 4GB model with its margins further eaten up by using more expensive memory short term and for a smaller initial launch batch.
Cheers

 

[xEx]

Please note that july 2 was only 2 days ago, and those days were a weekend, and today is a national holiday in the U.S. (in fact THE national holiday, traditionally celebrated by murricans all over the country sticking fireworks up their bums and setting them off.)

4th of july is actually a very big day for online shopping. There are lots of offers is almost all online stores so it is actually weird to not have stock for these day.

 

[xEx]

I think AMD recall cards in the USA to be modify(maybe they didn't in Europe because it would be too expensive?) but thats just a wild guess from my part. we will see how this situation develops in the future. Also customs 480 will start to be available on 18th.

 

[Grall]

4th of july is actually a very big day for online shopping.

So what? You can shop online even if there's no product in stock, unlike at a physical retailer. Your order will be shipped when the product has been re-stocked again.

My point was, if the product was out of stock before a weekend, it's not going to be back in stock again while it's still essentially a weekend (holiday). Even if people want to shop online.

 

[xEx]

So what? You can shop online even if there's no product in stock, unlike at a physical retailer. Your order will be shipped when the product has been re-stocked again.

My point was, if the product was out of stock before a weekend, it's not going to be back in stock again while it's still essentially a weekend (holiday). Even if people want to shop online.

No you can't pre order in newegg or other stores at the moment(just cheked) / actually you can in B&H but idk how much volume they actually have... By missing the holiday weekend you are missing lots of potential sells.

 

[Gipsel]

The thing is that even if you grow that 8 pin with black magic you will still have problems seems 4 of the phases are bounded to the PCE slot instead of the 6 pin. Idk how AMD will change that with just a driver update(without throttle the card to 380 perf lvls) but we will see tomorrow.

As already said, one can program the VRM Controller to bias the balancing of the phases. You can instruct the VRM Controller to draw more from specific phases than from other ones. And that is not a guess, it's an example given in the data sheet for the 4phase version of the controller (didn't find the complete datasheet for the 6phase version in public sources, but it supports the same features according to the first 2 pages which are accessible):

And there also may be the option to simply switch off phase 5 and 6 at normal loads, if they are connected to the slot and not the 6pin plug (it's somewhat of a 50:50 chance right now if they didn't put some thought in it).

By the way, there appear to be some contact pads which allow the rerouting of the 12V to some of the phases if one wants to take out a soldering iron (or one just connects the 12V from the slot and the 6pin plug as they are both available on these switching pads, but then don't forget to connect the 6pin plug or everything will be drawn from the slot in that case). And it could be a solution for later batches of cards if they choose a different configuration at the assembly line.

 

[Geeforcer]

I would think it is far simpler. AMD wanted to have some 4GB cards to launch (so that this variant wasn't just a paper launch), and they didn't build any (for whatever reason, you could just literally throw 4 gbit chips into your assembly line instead of the 8 gbit ones). And they just "downmodded" a very limited number of 8GB cards down to 4GB, they're sold out now and that's the end of it (only very few retailers seemed to got them in the first place).

So is everyone sure that they are looking at 1GB chips in 4GB version, and not, say, 512MB?

If it true that "4GB" version does indeed come with 8GB, at least initially, it would seem that hitting "$200!" marketing bullet point was more important than margins.

 

[mczak]

So is everyone sure that they are looking at 1GB chips in 4GB version, and not, say, 512MB?

If it true that "4GB" version does indeed come with 8GB, at least initially, it would seem that hitting "$200!" marketing bullet point was more important than margins.

I don't know about "everyone" but some had pics posted in the forums. These cards used all the same memory chips. Moreover, even the packaging was exactly the same - just a 4GB sticker on top...

 

[xEx]

more than 2 year in the making, still come out like this. I can't imagine any reason for this than how low the resources was to the engineering team. I think it would be much better to AMD to focus in few product, few well made products at time. Like a 480 alone and then few months after a 460 and then focus everything in Vega and so on. Its better to have few well made product than a whole line up of almost well made products.(Im talking from the perspective that AMD have really low resources so it would be better to focus it on few product to maximize the engineering)

 

[silent_guy]

If AMD concluded that a very temporary, very minor margin hit on a few 4GB parts was an acceptable trade-off for meeting their $200 marketing claims, then why not?

 

[revan]

They (AMD) where
a) a little lazy
b) or they didn't had enough working force
c) or is a bug regarding power-tuning
d) or they wanted as many cards as possible at launch and pumped a lot of voltage in its (...you know, just in case ) ...

And it backfired

Anyway, yet another undervolting study (for the rescue):
http://www.legitreviews.com/amd-radeon-rx-480-undervolting-performance_183699

All is not lost but they should have been more careful... Not only now but many times in the Past... I remember the 290/290X cooler-gate, now we have a "power-gate" on the table that could have been avoided... Pitty....

 

[dskneo]

The thing is that even if you grow that 8 pin with black magic you will still have problems seems 4 of the phases are bounded to the PCE slot instead of the 6 pin. Idk how AMD will change that with just a driver update(without throttle the card to 380 perf lvls) but we will see tomorrow.

Bullzoid measured the 6 Pin to have 3 grounds instead of 2 + 1 sensing. It allows the same current as the 8 pin.