New AMD Zen CPU details have emerged

[shambles1980]

i dont see whats hard to believe..
if they hadnt decided to cripple the bulldozers by making them not full cores then none of this would be in question, amd had intel beat at the core 2 quads and duos. and had them beat long before that, then for some reason they decided to waste money on bulldozer which even their previous cpus were faster than because of some stupid idea they had which said cores did not need to be actual cores.
and now they decided just to make a proper cpu again. I have no questions about clock per clock performance being the same or even slightly better it would fit in withe what they are capable of..
Energy efficiency and subsequently temps at those clock speeds may well be a factor to look at when benches come out. But clock for clock.. i don't see why people would doubt it, unless you did not come in to the pc game untill 2nd gen i series cpus.
honestly before bulldozers amd had intel beat, and if amd had given the bulldozers all the components for each core then it would not have been as useless as it turned out to be.
Amd prety much did the same mistake intel did when intell went and made the pentium 4 cpu. Some how though the world ddnt explode and turn against them because they made a terrible cpu that was worse than the pentium 3.
Then they made a pentium D which was just 2 pentium 4's superglued together. Again the world didnt explode.
then they made the core 2 duo and quads, which were good cpu's "based on the same tech as the pentium 3's" . granted the amd offerings usually beat them. but it wa a nice even playing feild. then intell decided to make the 1st gen i seriese cpu's "werent much better to be fair" and then 2nd gen were great..
Amd them made the bulldozers which were a joke. And the world did explode. Which is just stupid because intell had only just done the same thing and no one batted an eye lid at it.

So Its very strange to see so much doubt about amd cpu's when really they only messed up once.. intell have messed up more than that but seem to get away with it.

 

[SeekaX]

It can beat a 1080. That doesn't mean it always will.
In addition, even if this is the lone result that Zen beats Intels Flagship 8 core, the fact it's this comparable to something more than Haswell would suggest and more than likely fall true, that it'll beat any Haswellchip and probably Skylake too. They promised 40% IPC and have shown it's more than that. Even if cherry picked, it's still giving us more than promised.

I'm not saying it'll be the new king and destroy Intel. Just saying it is impressive even if you say it's cherry picked

Their message was very clearly that two 480s beat one 1080, not that they can beat a 1080 in a very small amount of applications. In the CF review Tom did they beat a 1080 in two out of 16 benchmarks, both of them synthetic, so by your own standards irrelevant. The magic of cherrypicking.
Skylake is already a year old, haswell is three years old, so i don't even know how that comparison is relevant. Kaby Lake will release before Zen, Skylake-X/Kabylake-X shortly after Zen and Cannonlake probably half a year after Zen. So i think it's fair to assume that even if AMD take the crown they won't have it for long. Assuming these results are cherrypicked and AMD falls short on the final product and assuming intel provides a strong lineup next year Zen could fall back into the role of i5 competitor/cheap rendering core quite quickly, which still wouldn't be bad though.
I'm not denying that it is impressive either, an AMD comeback in the CPU segment of any kind is impressive considering how far they were behind. I'm just saying that it's too early to get hyped, especially when it's benchmarks from the manufacturer.

 

[Zoot]

Idential Clock speeds, Comparable Configurations....

The devil is in the details, I wonder what they mean by that - it certainly isn't Identical. The cynic in me says they gave the Zen CPU an advantage that it's not going to have in the real world.

Still though, this is extremely hopeful. Color me catiously optimistic, although it'd have been nice to see a single thread benchmark too.

Cue waiting game once more....

 

[AlienALX]

AMD has had decent overclockers in the CPU world for a while now. 5Ghz was often possible on their chips. That's quite hard to achieve even with a 6700K let alone a 6900K, which tops out at around 4.4Ghz. If AMD can release an 8-core CPU on a versatile platform that is as powerful as a 6900K with decent overclocking (4.4Ghz+) for less money than Intel, that's a winner in my books. I'd definitely considering spending more than what I would need to if the price and performance are right. I do hope they have a slightly cut down version, though, that competes with the 5820K and 6850K respectively at a lower price.

What they clock like is a mystery. Back in the early 2000s a P4 could do a rock solid 5ghz under phase. Die shrinks actually hurt the clocks really.

I certainly wouldn't use Vishera as a base for the clocks of Zen though.

 

[demonking]

Idential Clock speeds, Comparable Configurations....

The devil is in the details, I wonder what they mean by that - it certainly isn't Identical. The cynic in me says they gave the Zen CPU an advantage that it's not going to have in the real world.

Still though, this is extremely hopeful. Color me catiously optimistic, although it'd have been nice to see a single thread benchmark too.

Cue waiting game once more....

its probably to do with chipset ect. doubt even the best engineers could get a zen cpu working on the X99 chipset this side of us achieving faster than light travel lol.

Maybe... just maybe AMD have actually sorted things out. I wonder if they thought they could take a step back after Phenom and took too much of one

 

[AngryGoldfish]

What they clock like is a mystery. Back in the early 2000s a P4 could do a rock solid 5ghz under phase. Die shrinks actually hurt the clocks really.

I certainly wouldn't use Vishera as a base for the clocks of Zen though.

I always read that a die shrink would increase core frequency, but it seems that is not always the case. Intel clock speeds have decreased. AMD and nVidia's core clocks have increased with a die shrink, but that'll hit a brink wall, wouldn't it?

Their message was very clearly that two 480s beat one 1080, not that they can beat a 1080 in a very small amount of applications. In the CF review Tom did they beat a 1080 in two out of 16 benchmarks, both of them synthetic, so by your own standards irrelevant. The magic of cherrypicking.
Skylake is already a year old, haswell is three years old, so i don't even know how that comparison is relevant. Kaby Lake will release before Zen, Skylake-X/Kabylake-X shortly after Zen and Cannonlake probably half a year after Zen. So i think it's fair to assume that even if AMD take the crown they won't have it for long. Assuming these results are cherrypicked and AMD falls short on the final product and assuming intel provides a strong lineup next year Zen could fall back into the role of i5 competitor/cheap rendering core quite quickly, which still wouldn't be bad though.
I'm not denying that it is impressive either, an AMD comeback in the CPU segment of any kind is impressive considering how far they were behind. I'm just saying that it's too early to get hyped, especially when it's benchmarks from the manufacturer.

Wouldn't it be considered hype to suggest that Kabylake, Skylake-X and Cannonlake will all beat Zen handily? There is very little evidence to suggest that except the fact that Intel has consistently upgraded their IPC's slightly every generation.

 

[SeekaX]

Wouldn't it be considered hype to suggest that Kabylake, Skylake-X and Cannonlake will all beat Zen handily? There is very little evidence to suggest that except the fact that Intel has consistently upgraded their IPC's slightly every generation.

I don't really think intel is the one who needs to provide evidence that they'll have a solid lineup, they've had a solid lineup for the past four years whilst AMD was in hibernation, which is kind of the reason why the IPC increases were rather small. If we go with the normal 10% increase per generation intel is most likely going to stay on top of things, they have a huge lead right now, no need for 40% increases for them to compete.

 

[AngryGoldfish]

I don't really think intel is the one who needs to provide evidence that they'll have a solid lineup, they've had a solid lineup for the past four years whilst AMD was in hibernation, which is kind of the reason why the IPC increases were rather small. If we go with the normal 10% increase per generation intel is most likely going to stay on top of things, they have a huge lead right now, no need for 40% increases for them to compete.

And "solid" counts as disqualifying (or more fairly, doubting) the power of the competition?

The point I'm trying to make is, Intel has made steady improvements, but they're not dramatic enough to discount the competition and initial testing results. I have doubts that AMD can match a 6900K across the board, but do they really need to? If they over-exaggerate their results, they're still well within the margin of competitiveness. So to suggest that Cannonlake is going to dethrone Zen is similarly speculative. If Zen is comparative to 6900K according to cherry-picked testing, for argument sake the realistic outcome is more like a 6800K, which Cannonlake might be able to match once it's out in late 2017/2018. That's a fair perspective I feel and leaves Zen a chance to regain AMD a market share. Also, the initial launch of Zen isn't the only Zen release. AMD will (or, should) continue to improve on the architecture just as Intel continues to improve on its architectures.

 

[SeekaX]

And "solid" counts as disqualifying (or more fairly, doubting) the power of the competition?

The point I'm trying to make is, Intel has made steady improvements, but they're not dramatic enough to discount the competition and initial testing results. I have doubts that AMD can match a 6900K across the board, but do they really need to? If they over-exaggerate their results, they're still well within the margin of competitiveness. So to suggest that Cannonlake is going to dethrone Zen is similarly speculative. If Zen is comparative to 6900K according to cherry-picked testing, for argument sake the realistic outcome is more like a 6800K, which Cannonlake might be able to match once it's out in late 2017/2018. That's a fair perspective I feel and leaves Zen a chance to regain AMD a market share. Also, the initial launch of Zen isn't the only Zen release. AMD will (or, should) continue to improve on the architecture just as Intel continues to improve on its architectures.

Yes, steady improvements which can be tracked over multiple years are more likely to continue than a large jump in performance making up a 4 year deficit and then some is likely to happen.
I don't doubt that AMD is going to regain marketshare either, i just doubt they'll suddenly be on the same level as intel. If Zen turns out strong they could close the gap completely in a generation or two. Icelake is going to be the interesting one, unless Intel have really been relaxing over the past few years and prepared nothing for AMD's return that could be a big jump.

 

[AngryGoldfish]

Yes, steady improvements which can be tracked over multiple years are more likely to continue than a large jump in performance making up a 4 year deficit and then some is likely to happen.
I don't doubt that AMD is going to regain marketshare either, i just doubt they'll suddenly be on the same level as intel. If Zen turns out strong they could close the gap completely in a generation or two. Icelake is going to be the interesting one, unless Intel have really been relaxing over the past few years and prepared nothing for AMD's return that could be a big jump.

Intel's progress is more likely to continue, but it doesn't mean it is going to dethrone Zen's position as a competitor. As I said, Zen's progress should theoretically continue to grow just as Intel's does. If they grow a little behind Intel, that's OK. Realistically, a 4790K is a 3-year old CPU, but it is still perfectly usable in almost any situation. I doubt AMD will only be able to compete with a 4790K, but even if they did their progress will still be valid and noteworthy, and it'll continue to grow just as Intel has done.

 

[SeekaX]

Intel's progress is more likely to continue, but it doesn't mean it is going to dethrone Zen's position as a competitor. As I said, Zen's progress should theoretically continue to grow just as Intel's does. If they grow a little behind Intel, that's OK. Realistically, a 4790K is a 3-year old CPU, but it is still perfectly usable in almost any situation. I doubt AMD will only be able to compete with a 4790K, but even if they did their progress will still be valid and noteworthy, and it'll continue to grow just as Intel has done.

I never said Zen wouldn't be able to compete anymore, that dethroning comment was aimed at who has the best performing CPU, if Zen turned out to really be better than Broadwell-E. Beating Broadwell-E isn't all that relevant to AMD being able to compete.
Them being able to re enter the mid/high range market is impressive in general considering their R&D budget compared to Intel's, it's roughly 1/10th and that includes their other products. Just in terms of funding Intel could leave AMD in the dust if they chose to do so, hence i think Icelake is going to be interesting, it'll show whether Intel is fine with being on equal footing with AMD or if they want to build a lead again.

 

[AngryGoldfish]

I never said Zen wouldn't be able to compete anymore, that dethroning comment was aimed at who has the best performing CPU, if Zen turned out to really be better than Broadwell-E. Beating Broadwell-E isn't all that relevant to AMD being able to compete.
Them being able to re enter the mid/high range market is impressive in general considering their R&D budget compared to Intel's, it's roughly 1/10th and that includes their other products. Just in terms of funding Intel could leave AMD in the dust if they chose to do so, hence i think Icelake is going to be interesting, it'll show whether Intel is fine with being on equal footing with AMD or if they want to build a lead again.

I remember the hype surrounding Skylake. So many seemed to think it was Intel finally setting a new standard. In the end it was still just a marginal gain over Haswell. The most interesting advancement was from the motherboard chipset. So I don't have high hopes for Intel CPU's. One might say they're a sure thing, but it's still marginal gains to milk the consumer, and that's like buying a new BMW every year when all they've done is refined the transmission and added an inbuilt mp3 player. I like to see advancements that are exciting and fun. It's not just numbers or gaming. It's fun to see silicon progress. So, yeah, I'm interested to see whether Intel decide to truly advance or whether they continue serving slightly less watered down version of Balvenie 21 every year. I have higher hopes for Zen over Intel. If you break it down to entertainment value, AMD are more fun to chat about.

 

[SeekaX]

I remember the hype surrounding Skylake. So many seemed to think it was Intel finally setting a new standard. In the end it was still just a marginal gain over Haswell. The most interesting advancement was from the motherboard chipset. So I don't have high hopes for Intel CPU's. One might say they're a sure thing, but it's still marginal gains to milk the consumer, and that's like buying a new BMW every year when all they've done is refined the transmission and added an inbuilt mp3 player. I like to see advancements that are exciting and fun. It's not just numbers or gaming. It's fun to see silicon progress. So, yeah, I'm interested to see whether Intel decide to truly advance or whether they continue serving slightly less watered down version of Balvenie 21 every year. I have higher hopes for Zen over Intel. If you break it down to entertainment value, AMD are more fun to chat about.

They didn't need to set a new standard, now they do if AMD pushes them. I wouldn't even call it milking the consumer since anyone with half a brain knew that an upgrade wasn't worth it, someone who owns a 2015 BMW is unlikely to buy a 2016 BMW as well. I wonder if that could be bad for Zen sales as well, 8 cores/16 threads are neat and all, but if they aren't utilized you are left with nothing better than an i5, i wouldn't be willing to upgrade my entire setup for 5% more performance in games. Anyone with a sandybridge or newer CPU can probably still comfortably wait till 2018.

 

[AngryGoldfish]

They didn't need to set a new standard, now they do if AMD pushes them. I wouldn't even call it milking the consumer since anyone with half a brain knew that an upgrade wasn't worth it, someone who owns a 2015 BMW is unlikely to buy a 2016 BMW as well. I wonder if that could be bad for Zen sales as well, 8 cores/16 threads are neat and all, but if they aren't utilized you are left with nothing better than an i5, i wouldn't be willing to upgrade my entire setup for 5% more performance in games. Anyone with a sandybridge or newer CPU can probably still comfortably wait till 2018.

For me, an upgrade is because I want to upgrade. My CPU is fine for what I use it for, but I would like a faster SSD, USB 3.1, PCI-e 4.0, etc. on top of the benefits of an improved architecture. I also would love to have an AMD CPU and graphics card. That would be pretty cool. Again, I'm breaking it down to trends as well as everything else.

 

[Bartacus]

Please Santa, let AMD be competitive again, if not just to hopefully drive down x99 pricing. Intel needs a kick in the butt.

 

[Kushiro]

Intels IPC gains per generation has been miniscule over the past years, hardly unbelievable that Zen can compete with it. Its not like people are saying AMD is going to beat it..

 

[NeverBackDown]

Intels IPC gains per generation has been miniscule over the past years, hardly unbelievable that Zen can compete with it. Its not like people are saying AMD is going to beat it..

Well AMD said that indirectly when barely edging out Broad-E. I think most likely scenario is that Flagship Zen will *match* BE. That is it can keep up in some areas but lose in others(probably more lose in others) but while doing this, maintain a lower TDP(rated for 95w vs 140w) and potentially heat output, since less heat, but who knows. For other CPUs like the 6700k, it'll probably match at worst but surpass at best, since skylake has better IPC but less physical cores.

On the gaming front. I don't expect much of a difference though between a 6700k or 68/69xx CPU. That's how I see it at least, feel free to disagree but it is speculation. One thing they need to do is perform best on price, but then again, that's relevant to performance level too.

 

[MacLeod]

I'm getting excited about this. I don't expect an Intel killer but it is looking like Zen could be competitive at least.

 

[SeekaX]

Well AMD said that indirectly when barely edging out Broad-E. I think most likely scenario is that Flagship Zen will *match* BE. That is it can keep up in some areas but lose in others(probably more lose in others) but while doing this, maintain a lower TDP(rated for 95w vs 140w) and potentially heat output, since less heat, but who knows. For other CPUs like the 6700k, it'll probably match at worst but surpass at best, since skylake has better IPC but less physical cores.

On the gaming front. I don't expect much of a difference though between a 6700k or 68/69xx CPU. That's how I see it at least, feel free to disagree but it is speculation. One thing they need to do is perform best on price, but then again, that's relevant to performance level too.

AMD winning on powerconsumption, that's really weird to read. When was the last time that happened?

 

[MacLeod]

AMD winning on powerconsumption, that's really weird to read. When was the last time that happened?

Fermi.