AMD CEO Dr Lisa Su talks about Zen+ processor development

[HEXUS]

And AMD releases the WHQL certified Radeon Software Crimson Edition 15.12 driver.

Read more.

[Marenghi]

It has the same resolved and known issues listed in the release notes.

not quite, the 2 new resolved issues are at the top of the list and the known issues list has tripled

[lumireleon]

I can't wait to see drop in CPU prices......NVIDIA & INTEL will run for the little cash.

[aniilv]

at this rate by the time amd releases consumer priced products (i5 level, not enthusiast level) intel will be a step ahead as usual

[canopus72]

Bearing in mind bintel's pathetic performance over the last few generations, it is impossible to understand why bintel fanboys are brainwashed to believe bintel cpu's will be one step ahead. TBH, fail-lake only showed 3% IPC improvement over has-been (and has-been showed only 5% IPC improvement over ivybridge). Bearing in mind IPC gains, Zen will be circa 8% to 12% faster in IPC gains over fail-lake. So bintel will be at least one generation behind Zen. By the time bintel catch up with Zen, AMD will have already released the Zen+ (which will have 10% IPC gain over Zen). The questions to ask is if/when bintel catch up with Zen, will next gen bintel cpu's still be plagued by problems that hit the 4790K (poor power delivery and over-heating) and the 6600K and 6700K bending and warping under the feather weight of even AIO cooling plates because bintel skimped on QC and opted for the thinner/cheaper silicon wafers yet had the bare faced audacity to charge in excess of £300 for their lame quad core 6700K?.

[canopus72]

analv - you are talking absolute rubbish.

[Corky34]

analv - you are talking absolute rubbish.

Says the person who just entered into a verbal diatribe against Intel, complete with unamusing puns, factual inaccuracies, and wishful thinking.

[eugenius]

Forget about zen+, let's see zen benchmarks and a zen APU with at least 4gb of hbm2 memory onboard, under 65W

[shaithis]

Bearing in mind bintel's pathetic performance over the last few generations, it is impossible to understand why bintel fanboys are brainwashed to believe bintel cpu's will be one step ahead. TBH, fail-lake only showed 3% IPC improvement over has-been (and has-been showed only 5% IPC improvement over ivybridge). Bearing in mind IPC gains, Zen will be circa 8% to 12% faster in IPC gains over fail-lake. So bintel will be at least one generation behind Zen. By the time bintel catch up with Zen, AMD will have already released the Zen+ (which will have 10% IPC gain over Zen). The questions to ask is if/when bintel catch up with Zen, will next gen bintel cpu's still be plagued by problems that hit the 4790K (poor power delivery and over-heating) and the 6600K and 6700K bending and warping under the feather weight of even AIO cooling plates because bintel skimped on QC and opted for the thinner/cheaper silicon wafers yet had the bare faced audacity to charge in excess of £300 for their lame quad core 6700K?.

All those problems, all those TINY 5% increases.....yet Intel have still stayed a country-mile in front of AMD over those years....says something, doesn't it?

There is a reason many people use Intel CPUs over the cheaper AMD counterparts.....

As for Intel catching up with Zen.....you must show me the benchmarks of Zen, I seem to not be able to find them......

[DanceswithUnix]

All those problems, all those TINY 5% increases.....yet Intel have still stayed a country-mile in front of AMD over those years....says something, doesn't it?

That's the annoying bit. If you look at FX Piledriver benchmarks, you notice two things:

1/ Compared to the i7 of the day, it wasn't actually all that far behind (about 20% apart from a few outliers like video encoding benchmarks where it won).
2/ The old Fx8350 is still faster than any modern FM2+ device despite AMD also making tiny 5% per generation (and refresh) updates.

Makes me wonder what they heck they are thinking/doing.

Still, meh, ARM is where the interesting stuff seems to be happening these days.

[Dazz]

All we can do is wait and see, but Intel has been doing very minor changes to their processors over the last few generations, anyone on 2nd generation i5 or i7 havn't been in a rush to upgrade because the performance increase is so small it does not warrant to replace RAM, motherboard and processor. We can only hope that Zen lives upto the hype and make the market competitive, maybe Intel will start to release something more than quad core and finally phase out dual cores altogether!

[SiliconAudio]

The way I see it, AMD has let the cat out of the bag with all these Zen press releases and given the might of Intel plenty of time to put up a competitor that will crush it. Intel having so much more R&D resource and its own fabs, means they will probably drop a bomb just as Zen goes to market. AMD have all their eggs in the Zen basket and if Intel crashes their party (which they surely will), it will be curtains for AMD. Either AMD will fold at that point or be forever relegated to budget low end parts. They have already admitted they can't survive as a low-cost parts provider.

[DanceswithUnix]

The way I see it, AMD has let the cat out of the bag with all these Zen press releases and given the might of Intel plenty of time to put up a competitor that will crush it. Intel having so much more R&D resource and its own fabs, means they will probably drop a bomb just as Zen goes to market. AMD have all their eggs in the Zen basket and if Intel crashes their party (which they surely will), it will be curtains for AMD. Either AMD will fold at that point or be forever relegated to budget low end parts. They have already admitted they can't survive as a low-cost parts provider.

Oh come on it isn't as bad as all that. They let the cat out of the bag with Bulldozer and they are still around to tell the tale.

Besides, all they need is something good enough that the minority of people who are happy to buy AMD have something they can usefully buy. Even if Zen is a twice as fast as an i7, consumes only half a Watt and also gives superb footrubs the majority of punters will still buy Intel muttering "ah yes, but just wait for the next Intel generation that will to backrubs as well as footrubs" until Intel catch up.

That's just how people work. I see the same with cars. Doesn't matter how good a car Alfa Romeo make, people read the good reviews, see the awards won, coo at how pretty it looks, marvel at how advanced the engine technology is, and then go buy a BMW anyway because they feel more secure doing that. And yet, with cars VW have just proved that you can easily pee your reputation right up the wall in one moment of stupidity, yet Intel seem to be immune to that having produced all sorts of buggy CPUs and chipsets along with failures like Larrabee and Itanium without the slightest dent to prestige.

The way I see it, AMD will produce something around the performance of Haswell, hopefully throw in some more cores so they at least win in some benchmarks, the haters will hate & the bargain hunters will buy it as always, and AMD will still be around.

[Corky34]

The way I see it, AMD has let the cat out of the bag with all these Zen press releases and given the might of Intel plenty of time to put up a competitor that will crush it. Intel having so much more R&D resource and its own fabs, means they will probably drop a bomb just as Zen goes to market. AMD have all their eggs in the Zen basket and if Intel crashes their party (which they surely will), it will be curtains for AMD. Either AMD will fold at that point or be forever relegated to budget low end parts. They have already admitted they can't survive as a low-cost parts provider.

I highly doubt Intel wouldn't have known that AMD was working on a new design, the press releases about Zen would only have helped AMD's stock prices, it also takes a long time from design to finished product so if Intel wanted to put up a competitor that crushed Zen they would have to do it with their existing microarchitecture, the most they could do is tweak their existing design.

It's true Intel have much more R&D resources but that doesn't always equal good results, NetBurst being a prime example.

AMD also don't have all their eggs in one basket, far from it (IMO) as unlike Intel they have their GPU division with Arctic Islands due in 2016.

[CAT-THE-FIFTH]

If you look at the Piledriver to Excavator,there is actually a decent IPC increase. An Excavator based FX CPU would have been not too bad at all - however,Intel has had two node shrinks since then and AMD only has had access to one and even that could not hit the kind of clockspeeds AMD wanted,so the IPC increase would have been nullified by the reductions in clockspeed.

You only need to look at Steamroller vs Piledriver:

http://uk.hardware.info/reviews/5156...-vs-piledriver
http://semiaccurate.com/2014/01/23/k...ck-comparison/

It was about a 10% to 11% IPC increase,and Excavator was another 4% to 15% above that:

http://cdn.wccftech.com/wp-content/u...vator-Core.jpg

Thats probably a good 20% increase overall from Piledriver,and its come with massive reductions in power consumption of the cores and even a large area reduction in core size.

If compare Piledriver and Bulldozer it was around 8% for applications and around 13.5% for games:

http://www.hardware.fr/articles/880-...ver-4-ghz.html

So going from Bulldozer to Excavator there is probably around a 30% to 35% IPC increase overall.

To put it in context,if a Excavator based FX CPU was released now,it would probably be close to SB or IB levels of performance in single threaded benchmarks.

[preter_s]

The most important factor here is manufacturing process. With 14nm now available to AMD (courtesy of Samsung and Global Foundries), Intel's one key advantage is gone. Consider also that Samsung is pushing ahead to be first with 10nm in 2016, while Intel doesn't even have any plans yet. Whatever one's allegiance, I am glad Intel's monopolistic money grabbing ways are now seriously under threat. For that alone, I wish AMD success indeed.