AMD silently launch M400 series: Rebrands everywhere

http://videocardz.com/59932/amd-launches-radeon-m400-series

 

So I guess 485X is about 970M performance, 490 - 980M and 490X - 10-15% faster than 980M. If only 490X is priced at 970M level, it's not that bad, but once mobile Pascal is out, these will be utterly destroyed.

 

This reminds me of Nvidia's GTX 800 series - they rebranded so much that they skipped the series numbering for the desktop versions...

 

490 and 490X will probably be exclusive to the iMac 5K.

 

This was pretty much confirmed a few months ago and I think that it was from you (in the mobile Polaris thread). What the table was lacking back then was the M480(X) and M490(X). As far as I can see, it's still lacking them, so - nothing to see here, another useless "news flash" topic.

 

No, Polaris 10 was already confirmed for high-end notebooks, but probably the iMac version will be called R9 M495X (OC'ed 490X).

 

That would go against everything AMD has been marketing Polaris for.

 

Man, if AMD is going to combat the GT 1040M (960M performance rumored) with another generation of rebrands, they are freaking hosed.

 

AMD - the rebrand king continues to reign!

 

Over their entire history the two companies are pretty close really.

 

Unfortunately this is true. I mean I understand needing to make a profit off old silicon that's lying around, its the deceptive practice of putting lipstick on a pig that makes me mad. When I ordered my Alienware 18, I had originally configured it with the M290X CF until I came on here after ordering it and realized what it really was. Dealing with Dell for an hour trying to get the order changed to 880Ms and remembering just how terrible their customer service is became the reason I canceled that order and placed one with Xotic for my 9377-S in the first place.

 

Hard to forget the 8800M -> 9800M - GTX 200M fiasco. We literally went 3-4 years with no new high-end chips, and then we were "rewarded" with the GTX 480M.

Yeah, all hands are dirty.

 

Wasnt this because of the limits of sillicon in those days?

 

AMD had 3870, 4850/70 and 5870 back then. That was the only point people gladly bought 5870s, and then again not all of them. There still were the ones buying the burners that 480m and later the 580m were (approaching near 100% failure rate as the history tells).

 

If Fermi had that high of a failure rate OEMs wouldn't have picked up the 680M, much like how they avoid AMD now because of the 7970m. Both companies have put out lemons, don't make it sound like either one is any better than the other because they aren't. If Fermi was really as bad as you say, we would live in an AMD market and we do not. I'd say the 880M was probably around an 80% or so failure rate but as long as Fermi was properly cooled, it didn't have a lot of problems. The fault of Fermi failing was due to inadequate cooling on the part of the OEMs and consumers not taking care of them whereas AMD was purely a manufacturing fault that they took forever to correct. Nvidia should have foreseen the thermal issue with the 480M and AMD should have fixed their manufacturing of the 7970m but neither have ever taken responsibility when they've screwed up just like any corporation.

 

Actually I'm making things equal, because everyone points finger and says - AMD bad. I doubt that lack of AMDs has anything to do with 7970m. The mother of all failure rates the 8xxx series, happened and OEMs were still using, and continued to do so, nGreedia. Is there a laptop with original (not replaced) 580m (675m)? Since you mention that we live in properly cooled times, what makes the 980m failures then?

 

This is bullcrap. AMD never sold as many cards as Nvidia did with the 8xxx series. But AMD had the same issues with the solder and it is universal to the manufacturing of GPU's of those days. But it is even worse that a lot of mobile AMD gpu's still have these issues in for example the Macbook pro 15 retina series. This was universal in the whole GPU industry and is also the cause of the Xbox 360 & PS3 fat failure rates. Fun detail that MS used an AMD card and Sony a PS3 card.

The 3850/3870 also those solder issues in desktops and the 4xxx series weren't perfect as well. In the end I saw no Nvidia card dying after the 8xxx series and I had a LOT of cards and I am quite the heavy user. the 78xx and 79xx AMD mobile gpu's have poor flipchip quality though.

Blame the EU for this with the law for banning solder based on lead.

 

What if I call your post bullcrap? Is this the way to make discussion? The PS3's GPU is made by nGreedia. I'm a heavy user as well and I haven't had a single ATi/AMD fail on me. EVER. So my word against yours. But yes, I know where the problem is, I'm just saying that things are equal and I said it a few years ago, but everyone was laughing. Only now people are starting to realize this. Even with fourth and fifth player the problem would still be there. That's why I'm saying people have to know about failure rates of either brand, but people tend to forget to mention that nGreedias fail and only point at AMD. Lastly is there an official units sold vs failure rates (not estimates) from a reliable source? I don't think so, but in a country like mine where it's still cheaper to repair than buying a new thing, there's something close to it, it's called service centers. Still there are people that couldn't be bothered, or don't have the money, so again, there's no reliable source to get info from.

TL;DR My point always was and would always be that things fail and people should know. The reason I point at nGreedia is because everyone else points at AMD!

 

Hard to take you seriously, when you use immature terms like nGreedia.

Breaking news: AMD doesn't give a damn about you either.

 

About the worst I've seen from nVidia was the massive issues with the 8600M GT cards (bumpgate). Including the RIDICULOUS class-action judgement where plaintiffs lost potentually thousands by being offered a crap $329 retail laptop. No matter if they unit they lost was top of the line and cost thousands.

 

I had an 8700M GT in my Toshiba Satellite X205-s9359... It failed three times before I said forget it and bought my M17xR1 with the GTX 260M SLI. Those cards were put through hell and they both worked up until 2014 when I tried to flash one of them to a Clevo vbios and it bricked. And when I say I put them through hell I mean it. When I was living in Louisiana I had only my laptop for entertainment and I constantly used it on my mattress which blocked all the airflow. The 8xxxM cards were definitely horrible. The 78xx and 79xx desktop cards were nasty too.

 

If you dig WAY back (somewhere around 2014 and before), before they make everything possible to iron that term I'm using now, I was, not to say the only, but among the very few, that wrote the company in question the way it should - lower-case first letter, all capitals afterwards. That's the way it should be written, I pay attention to this stuff (e.g. I always write quattro (not with upper-case first letter) for Audi's all-wheel drive system, or ATi, and etc). I might missed to write something properly here and there, but most of the times I do. BTW you can check here as well, it's a long time since the topic has been locked, or I've edited my post.

Because? You are calling me, or the terms I'm using, immature and your very next sentence is immature. I would love some argumentation.

 

How about we take this down a notch and focus on the M400 instead.

 

Well, I think it's time to start fishing of do sth else instead of "hoping for" AMD. Nvidia just dominate the market(btw nice prices for new series...)...

 

Its not my word agaisnt yours. All 80nm GPU's from that era that just switched for lead based solder to lead free and had to deal with the stress from powering up and down and thus cracking solder balls had to deal with reliability issues. But only Nvidia and MS (xbox 360) got sued for this. While AMD and Sony (PS3) could walk away fairly unscathed from these issues. When I say bullcrap I was aiming that at your statement that Nvidia GPU's after the 80nm series still has issues in terms of reliability while they are considered reliable now.

I had a HD 3850 die by the way because of the same reason a Nvidia 8800GTX would die.

 

Well I had HD 3650 both in my laptop and my PC. Both were alive (the PC is still around, but not used). The Mobility has seen 3 hour sessions (sometimes more) of BioShock @105ºC (because Spain) for an year and was still alive when I sold it. Should've kept it and put it in a frame. It deserved it.

As for the reliability of newer GPUs - that's why I'm vocal about it, there ARE failures. More than anyone is willing to admit. Especially for 980m. For a while now I was wondering if I should make a topic, but I'd guess not many would post... if any. People are sealed shut if their nGreedia fails. And then why AMDs market share is shrinking - lies, lies, lies, either from the EvilCorp (nGreedia, to avoid confusion ) itself or the users (because resale value, unwillingness to admit that their mighty glorious has fallen, whatever).

Ironically a friend's 7970m* has failed yesterday, but we'll fix it and the new chip would have lead balls (pun intended).

*Bought from a forum member and I think that it's well over 2 years, probably 3, and more than an year of this time was running @~85ºC. Not sure at what temps was running at the previous owner.

 

Seriously, do some research before you start throwing out garbage information. Something this basic shouldn't be glossed over.

Goes to show you BGA manufacturing is still heavily flawed. I expect this issues to continue as BGA laptops made in 2014-2015 will start exhibiting this behavior in a year or two.

 

Hahaha this is too funny. If you state something, be ready to back it up. Here it comes.

This is about reballing/reflowing GPU's or other chips.



This is from someone actually repairing pcb's for a living. As you can hear in this video it has NOTHING to do with the solder balls on the BGA package itself. It is the solder on the flipchip. Reballing heats up the chip and also the solder inside the flipchip itself thus letting it run again. This created the whole BGA myth. A friend of my working for Nvidia and being an advisor for super PC's would love to explain it to you in full detail.

Its a combination of multiple factors, bad solder, bad chip design, poor thermals.

Now thermals are not the issue anymore since we have much smaller and more efficient chips now. It shouldnt be a problem with for example the 8800GTX as well. But it was due to the change of solder because of EU regulations. They have fixed that by now. Thus leaving 2 other options open. Bad chips design and poor thermals. With the Macbook Retina's both was the case unfortunately. The chips where vulnerable for this.

To back this fact up even more. The past 4 Nvidia generations run reference at 80C under load. The fan profiles are tuned for fan noise and not low temperature. How can it be that these chips still work without dying like it happened with for example the 8800? Because better solder, better chip design etc. These are BGA chips. The AMD's run even hotter. They still run after years of use. What you state is pure ********. It has NOTHING to do with BGA. Educate yourself in this matter. Posts like this keep the myths alive. Only selective products died on a massive scale because of bad solder and bad chip design and it had nothing to do with BGA. Thermals over 90C could kill a chip premature though if the other factors are poor as well.But it happens on the inside way before the heat does its work on the BGA mount itself.

 

I was talking about your completely incorrect information about the rMBP but ok, glad you see you didn't address that at all in that long tirade rant.

Now, consider this:
Why is if I heat a 6990M or RSX without liquid flux, I get pretty much weak/no improvement and immediate failure after light usage, but if I use liquid flux and drip it under the BGA package, I get a much longer fix (permanent in most my cases). Nothing to do with BGA right? Sure, ok...whatever you say. Putting pressure on the die to apply pressure on the BGA balls to make proper contact aka penny trick has nothing to do with BGA, right? ok....

The presure needed to fix the solder internally under the die itself would be too much, crack the die.

 

When you pressure the die you also pressure the chip itself. And still the failure there is because of the sudden change to lead free solder which was new. Again not a BGA specific issue. THe rest of your claim cannot be backed up in any way with scientific proof. You never documented it so I cannot check your claim.

Oh and the incorrect information about the rMBP? I just didnt name the Nvidia GPU's but that doesnt mean that there werent any AMD laptops with the same issues and that was my point.

 

It is not a sudden change. Lead free has been a thing for the past 10 years, and yet, we have still have problems with lead free solder. The proof not scientific in the slightest, but empirical evidence will do for me and 90% of users out there. Let's see how your beloved BGAlienware holds up 3 years after manufacture.

You were faulting AMD in the rMBP for an issue that doesn't even exist (yet...) with their chips.

 

It has been a sudden change for chips that become this hot.

If my "beloved" Alienware lasts as long as my previous Vaio's then I have nothing to complain. My old Vaio Z1 from 2011 is still rocking on in someone else's posession and has been overclocked, heavily gamed on etc. It was a Nvidia 330 an even older vaio with a Geforce 8400 is also still working fine and I have tons of other machines that still work. Just as most graphic cards keep working even though they are all BGA.

 

It's very unscientific, but putting an old, almost gone, chip with lead balls grants extended use, unlike using lead-free solder where it might not even turn on afterwards. I think that it has a lot to do with flexibility. We are talking about microns here, but they are enough to crack paths/balls/you-name-it. Lead solder is more flexible and allows for reduced stress/bending on the chip. After all there are A TON of things to be taken into account - the chip itself doesn't heat-up equally, because of different placement of components on it (shaders, cache and etc); the board has different flex characteristics, that's why directly soldered chips have better chances to fail especially using lead-free solder; and pretty much everything that heats/expands/bends differently and is around the about to fail chip. Every little difference helps to put more stress on the solder and eventually crack. Again that's very unscientific and is my take on the matter.

 

THis is ofcourse correct. But when resoldering (reballing) a chip. A user is also reflowing the flipchip package itself by default which often fixes the chip for a short period of time while that was actually in modern cases most of the time the problem.

 

@rinneh and I will always have our differences (inherent characteristic of the internet), but let's try and get this topic back on track. Aside from the rebrands, there is no info about the Polaris mobiles yet is there?

 

No new info as far as I know, my opinion about this topic is on the first page and I think that we actually improved it later on

@rinneh - That's why I said that using lead balls further reduces the stress and even almost gone chip can outlive a new machine with bad design, because you are limiting or removing entirely that bending. As @Raidriar mentioned that's the reason why oven backing can only be used that much times, at second or third time there wont be anything to melt and/or to help the process of melting and bonding i.e. flux (yes I'm talking about the flip chip as well, if it goes no amount of lead balls and flux (on the interposer) can get it back, you'll need a new chip). There's also the thing that lead-free soldering happens at higher temps, that puts even more stress to the components and there are a few variations of the lead-free solder depending on the elements used and of course since there are variations, the quality varies as well.