AMD - Piledriver chitchat

[SiliconAudio]

That all sounds quite promising, apart from:


March? Well, it's still Q1 I guess but it's not what I'd hoped...

This is just the latest of many slips. We will not see Zen in Q1. We will be lucky to see it Q2. And that's if AMD's investors don't jump ship first. This is terrible news that could be the end of AMD.

I'm an AMD enthusiast, but this is starting to look like writing on the wall to me. I actually think it's possible that Zen will never actually ship.

I skipped a 25% off deal on an 8-core Intel to hold out for Zen in early 2016. I now see I made a huge mistake.

[kalniel]

This is terrible news that could be the end of AMD.

I'm an AMD enthusiast, but this is starting to look like writing on the wall to me. I actually think it's possible that Zen will never actually ship.

That's a very odd opinion to take IMHO.

It's not like Intel chips have progressed all that much in recent years - mine is over 7 years old now and there still isn't any convincing argument to upgrade it!

And you noted yourself back in June that you expected Zen to be Q2 2017.

I bet my lunch it will be delayed until Q1 or Q2 2017. I'm holding out to replace my FX8350 rig with a Summit Ridge system, but I won't be holding my breath.

[CAT-THE-FIFTH]

AMD were originally aiming for a late 2016 launch,but its not surprising as they are dealing with GF,that there would be issues,and this is why they delayed it until early next year.

My viewpoint is that I would rather they delay Zen and make sure it launches in as best a condition they can launch it instead of a half baked one.

They cannot afford to screw the launch up and if they need to delay it even six months,it does not matter - if they couldn't launch this year with KB not out,they delaying it a few months more with KB out,at least gives them a chance to orientate the positioning of the chip in the price range.

My viewpoint is still more like IB to Haswell level performance with probably lower clockspeeds than KB. If they get more like Haswell/Broadwell level IPC I would be pleasantly surprised.

If Zen achieves even the former and is reasonably priced,I would probably quite happily buy it especially as I see them actually have mini-ITX motherboards and one thing I like about AMD is that their chips are compatible accross multiple motherboard generations,so even if the motherboard goes there is a good chance I can get a replacement mb.

[watercooled]

This is just the latest of many slips. We will not see Zen in Q1. We will be lucky to see it Q2. And that's if AMD's investors don't jump ship first. This is terrible news that could be the end of AMD.

I'm an AMD enthusiast, but this is starting to look like writing on the wall to me. I actually think it's possible that Zen will never actually ship.

I skipped a 25% off deal on an 8-core Intel to hold out for Zen in early 2016. I now see I made a huge mistake.

I don't recall AMD promising anything other than Q1 2017 for volume availability of Zen (fabricated rumours claimed otherwise as usual) - they mentioned there might be limited volumes in late 2016 but that could just mean production/engineering samples which people already have hold of.

People have been proclaiming the 'end of AMD' for at least two decades, and every time a release slips by a month or so. Even if they delayed Zen to the second half of 2017 I doubt it would make a jot of difference as that market isn't exactly moving quickly anyway - Intel have Kaby Lake which is basically a clock bump of Skylake as far as the desktop is concerned.

@kalniel: I have a ~6 year old 1055T and have pretty much the same opinion TBH - I really don't have a pressing need to upgrade!

[shaithis]

What I find most weird is how long ago the designer left AMD.....

He start design in 2012, left AMD Sept 2015 and here we are Nov 2016 looking at another 6 months.....before the product start shipping

I am only expecting the originally quoted 40% IPC increase over excavator at this point, which by Q2 2017 is going to look quite "meh" and is going to have to be cheap to entice people to upgrade or OEMs to try and push.

And at this point, there can't be too many people left waiting for an upgrade........they will really need Dell and HP on board to shift them IMO.....and historically, that has been a BIG problem for them.

[watercooled]

The architectural work would have been substantially complete before Keller left. No doubt he's put a lot of work into it but it takes far more than one person to design a modern CPU microarchitecture. Either way, since then will have been steps like implementation, layout and, if necessary, bug fixes. CPU manufacturing is a long pipeline, 2012-2017 is nothing out of the ordinary and the sort of timeframe it takes Intel to iterate their CPU cores (either die shrink or uArch iteration) with a significantly larger workforce and budget, let alone a brand new design and a new process node.

For example, look at the gap between ARM announcing their IP cores and seeing that core shipping in processors - it's usually well over a year and the major partners have likely had access to the design before its public announcement. By that time the architectural work, silicon prototyping and bug fixing has been done by ARM and the foundries, and these are substantially smaller cores.

I don't know where you're getting 6 months from. Like I said before though, I don't think even Q2 2017 would make much of a difference from Q2 2016 - it's not like the market has moved on much. Intel have faced their own "delays" (going by rumour logic) going from tick-tock to tick-tock-tweak and running into the same problems as the rest of the industry with regards to pushing back smaller nodes until they're financially viable.

[CAT-THE-FIFTH]

Looks like the Zen SKU levels have been leaked:

http://wccftech.com/amd-zen-summit-r...etails-leaked/



Looking at the Chinese prices and the relative positioning to the other CPUs,it looks like the lowest end Zen will start at around £200ish.

[kalniel]

7..5..3.. now where have I seen that series numbering before?

[CAT-THE-FIFTH]

AMD seems confident to price the top end SKUs at a reasonable price for them. So hopefully the lower 4C/8T Zen is actually not too bad. Even if it were Haswell levels of single threaded performance,after looking at BF1,I would quite happily get a 4C/8T Zen SR3 chip over a KL Core i5.

[SiliconAudio]

That's a very odd opinion to take IMHO.

It's not like Intel chips have progressed all that much in recent years - mine is over 7 years old now and there still isn't any convincing argument to upgrade it!

And you noted yourself back in June that you expected Zen to be Q2 2017.

It's not so much the delay, rather the fact that they have such big problems this late in the game. I was expecting minor tweaks in Q1 and a ramp up of production ready for launch in Q2. Realistically, AMD have a lot of work to do and it has to be said, recent indications from AMD have had the market expecting a January release.

Again, that's if these rumors are actually true. If they are, I can't help but feel a little pessimistic.

[CAT-THE-FIFTH]

BTW,the chap behind a well know Italian rumours site said EVGA might start making AMD cards a few months ago. This chap works for EVGA:

http://forum.kingpincooling.com/show...88&postcount=5

It makes me wonder if EVGA will make Vega based cards??

[watercooled]

It's not so much the delay, rather the fact that they have such big problems this late in the game.

Maybe I'm missing something, but what 'such big problems' are you referring to? The errata mentioned in the Reddit post? That sort of thing and the workaround are not uncommon in prototype silicon - and just to re-iterate, don't forget this is a completely new CPU core with many firsts for AMD. I agree with CAT - I'd much rather it be few months later than we'd hoped (which is no time in the grand scheme of things) than leave bugs in shipping silicon. You know what the media are like with AMD - they misplace an apostrophe and it's a disaster. But Intel can have silicon bugs e.g. chipset USB3 dying, TSX extensions broken, Skylake freezing under certain workloads, etc and people pretty much ignore it.

[Xlucine]

BTW,the chap behind a well know Italian rumours site said EVGA might start making AMD cards a few months ago. This chap works for EVGA:

http://forum.kingpincooling.com/show...88&postcount=5

It makes me wonder if EVGA will make Vega based cards??

If they make AMD cards it's guaranteed, they'd hardly rehash 3XX series cards

[CAT-THE-FIFTH]

Wow,I actually thought AMD would have lost marketshare after the last set of financials from Nvidia:

http://www.mercuryresearch.com/graphics-pr-2016-q4.pdf

[watercooled]

So... logic dictates that AMD must have increased sales proportionally i.e. the market has picked up and the companies have held fairly steady on market share? That's pretty impressive given AMD have left the high-end segment of the market this time around.

Maybe I've just not been paying attention for a while but it seems like AMD are doing fairly well on the mobile side of things, especially given those numbers are for just discrete GPUs.

[DanceswithUnix]

So... logic dictates that AMD must have increased sales proportionally i.e. the market has picked up and the companies have held fairly steady on market share?

Or the profit per board has increased to outweigh the lost sales, for Nvidia at least.