New Zen microarchitecture details

[USER8000]

It's not about the graphics cards at all, but the reliability of the statements / claims AMD marketing has made in the past.

And how is different than what Nvidia has said either.Or should we push every argument about say Tegra into something about the GTX970 lawsuit,async support or something else??

All I see is the same users who argue massively in the graphics card section,now push a discussion about Zen into some AMD vs Nvidia E-PEEN graphics card war.

The worst thing is you don't get all the same arguments were had weeks or months ago in the graphics sections,and now they are being dragged up for a second flogging of the dead horse.

Some of you are way too obssessed about graphics cards,and since the moderation is much more stricter there,you now infect this section with more incessant bickering about graphics.

 

[Arachnotronic]

And how is different than what Nvidia has said either.Or should we push every argument about say Tegra into something about the GTX970 lawsuit,async support or something else??

All I see is the same users who argue massively in the graphics card section,now push a discussion about Zen into some AMD vs Nvidia E-PEEN graphics card war.

The worst thing is you don't get all the same arguments were had weeks or months ago in the graphics sections,and now they are being dragged up for a second flogging of the dead horse.

Some of you are way too obssessed about graphics cards,and since the moderation is much more stricter there,you now infect this section with more incessant bickering about graphics.

I think The Stilt's point stands. Marketing claims should always be taken with a grain of salt from all companies, and given AMD's recent track record, AMD should not be excluded from such skepticism. That's all.

 

[USER8000]

I think The Stilt's point stands. Marketing claims should always be taken with a grain of salt from all companies, and given AMD's recent track record, AMD should not be excluded from such skepticism. That's all.

No its another case,of the same people who made the same arguments weeks and months ago in multiple threads,many which got locked,flogging the same old dead horse weeks and months later in a different section. Thats all.

I came to read about Zen and all I can see is more moaning about Polaris - last time I checked CPUs and GPUs were quite different in design anyway. You might as well compare Apples to Oranges.

If I wanted to read regurgitated arguments about Polaris,there are a 100 million threads in the graphics card section about it.

Its basically thread thrashing,since I see a lack of such things in any Tegra or Intel CPU thread either. Or are you going to start arguing about the PR plunders Nvidia has also had in graphics in any Tegra discussion,or Intel graphics drivers in a discussion about Kaby Lake??

Some of you are way to obssessed about graphics cards.

Plus if you are that dubious about PR,then just wait until the CPUs are released.

But continue on,I will come back in another month and no doubt you will have regurgitated the same argument for the thousandth time.

 

[KTE]

I think The Stilt's point stands. Marketing claims should always be taken with a grain of salt from all companies, and given AMD's recent track record, AMD should not be excluded from such skepticism. That's all.

Most accurate statement so far.

Sent from HTC 10
(Opinions are own)

 

[imported_falacy]

I'm looking forward to seeing what AMD make using Zen. If its CPU performance is +/-10% of current Intel CPUs, that's perfectly fine in my books. It's not like Intel is pumping out new CPUs that are significantly faster or otherwise more improved than their current ones any way.

I think the following Zen APUs would really help the desktop market rebound in sales, provided that OEMs and retailers don't get greedy and over price them (good Lord, have you seen the prices on the junk OEMs are selling these days...).

$99 - 4 Core / 8 Threads, 2.4GHz base, 3.0GHz Turbo, 0MB L3 Cache, 384 Shaders. (Basic)

$129 - 4 Core / 8 Threads, 2.8GHz base, 3.4GHz Turbo, 2MB L3 Cache, 384 Shaders. (Budget)

$169 - 4 Core / 8 Threads, 3.4GHz base, 4.0GHz Turbo, 4MB L3 Cache, 512 Shaders. (Budget Gaming)

$179 - 8 Core / 16 Threads, 2.8GHz base, 3.6GHz Turbo, 16MB L3 Cache, 128 Shaders. (Workstation)

$199 - 8 Core / 16 Threads, 2.8GHz base, 3.6GHz Turbo, 8MB L3 Cache, 512 Shaders. (Workstation Balanced)

$199 - 6 Core / 12 Threads, 3.4GHz base, 4.2GHz Turbo, 6MB L3 Cache, 768 Shaders. (Mainstream)

$249 - 6 Core / 12 Threads, 3.8GHz base, 4.4GHz Turbo, 6MB L3 Cache, 1280 Shaders. (Mainstream Advanced)

$249 - 12 Core / 24 Threads, 3.2GHz base, 3.8GHz Turbo, 24MB L3 Cache, 64 Shaders. (Workstation Advanced)

For all but the the top 10% or so PC users, those APUs would be perfect. Think of the over all savings and the simplicity of systems presented here, without the need for a discrete graphics card, even for "better than console" quality gaming.

Anyhow, it will be neat to see what products AMD comes up with. Hopefully they're interesting!

 

[The Stilt]

And how is different than what Nvidia has said either.Or should we push every argument about say Tegra into something about the GTX970 lawsuit,async support or something else??

All I see is the same users who argue massively in the graphics card section,now push a discussion about Zen into some AMD vs Nvidia E-PEEN graphics card war.

The worst thing is you don't get all the same arguments were had weeks or months ago in the graphics sections,and now they are being dragged up for a second flogging of the dead horse.

Some of you are way too obssessed about graphics cards,and since the moderation is much more stricter there,you now infect this section with more incessant bickering about graphics.

Every single company in the industry has made false or at least questionable statements / claims in the past. AMD however is lately completely unrivaled in making such statements. I know as a fact that some of these statements were made by an individuals who were later dealed with accordingly, so in all of the cases the company as a whole cannot be blaimed. However the sheer number of these "unfortunate events" is so high that no objective person can nor should take any of the statements or claims made by AMD as granted. I hate to admit it, but for me personally it is "guilty until proven innocent" when it comes to AMD

ps. Did anyone notice that AMD decided to use "2x Perf/Watt of previous AMD mainstream GPUs available" in their most recent investor presentation, instead of the previous 2.8x figure? The 2x figure applies on different GPUs than the 2.8x figure did, however I still find it pretty odd.
https://app.scoutfin.com/document.h...b759fd048&-branch-match-id=297670879870932296

 

[FIVR]

What does AMD's choice of wording in "2x" vs "2.8x" (which says essentially the same thing, they just appear to be conservatively rounding down instead of rounding up) in a powerpoint from a presentation about Polaris have to do with Zen?

There are a few people in this thread who should be banned from this subform completely. Literally everything they post is FUD or threadcrap, see above.

 

[The Stilt]

What does AMD's choice of wording in "2x" vs "2.8x" (which says essentially the same thing, they just appear to be conservatively rounding down instead of rounding up) in a powerpoint from a presentation about Polaris have to do with Zen?

There are a few people in this thread who should be banned from this subform completely. Literally everything they post is FUD or threadcrap, see above.

So:

- 80% (relative) difference in the claimed performance per watt is insignificant?
- The history of questionable statements AMD has, doesn't make further claims (which is Zen matching Broadwell-E IPC in this case) questionable at all?

 

[cytg111]

Besides for the epeen points, do anyone really care what amd have said at any point in the past?
We (should) problary care more about the actual product, actual benches, actual silicon.

 

[FIVR]

No why not listen to the FUD producers and ignore benchmarks, technical specs, and articles while we discuss the "Horrible failure" that is the AMD disaster Polaris. I hear that Lisa Su herself has been buying up all the stocks of RX480 and that they will all show up as liabilities in the Q3 earning report.

/s

 

[FIVR]

So:

- 80% (relative) difference in the claimed performance per watt is insignificant?
- The history of questionable statements AMD has, doesn't make further claims (which is Zen matching Broadwell-E IPC in this case) questionable at all?

A) They don't specify which polaris architecture or which last gen card they are talking about. In all likelihood there are GPUs that Polaris is 2x more efficient than, and there are ones it is 2.8x more efficient than others. Did that possibility not cross your mind, or are you automatically assuming that they are lying merely because you've made a judgment call about their company?

B)" Innuendo I CHOOSE YOU!" You have an interesting view of history if you think AMD is somehow unique in having made inaccurate PR statements in the past.

 

[Arachnotronic]

So:

- 80% (relative) difference in the claimed performance per watt is insignificant?
- The history of questionable statements AMD has, doesn't make further claims (which is Zen matching Broadwell-E IPC in this case) questionable at all?

The Stilt, a lot of us appreciate the sense of realism that you are bringing to the discussion. You will no doubt be attacked as those riding the hype train will not like what you have to say, but please keep contributing to these forums. Too many great posters have been chased off because they didn't want to board the hype train and I think a lot of us would miss your insight if you were to be chased off by said train.

 

[Glo.]

The Stilt, a lot of us appreciate the sense of realism that you are bringing to the discussion. You will no doubt be attacked as those riding the hype train will not like what you have to say, but please keep contributing to these forums. Too many great posters have been chased off because they didn't want to board the hype train and I think a lot of us would miss your insight if you were to be chased off by said train.

The problem is that there is no realism currently, but absolute skepticism. It turned 180 degrees from hyping out, to downplaying every possibility, that people do not think for a second that Zen can really be clock for clock on the same level as Broadway-E for example.

However it has good side. If it will turn out to be that good, the impact will be much bigger in positive way, than hype train, blowing expectations out of the reality.

 

[imported_falacy]

However it has good side. If it will turn out to be that good, the impact will be much bigger in positive way, than hype train, blowing expectations out of the reality.

Pretty much. All the world really needs from Zen is for it to not suck as badly as Bulldozer sucked.

Clock for clock, a "construction core" is anywhere from 40-60% worse than an "i-Core", which significant to the point where it can actually be noticed in day to day use, let alone high-end tasks. If Zen can knock that difference down to even 10-15% worse it will move us into a place where realistically, the difference in performance isn't really noticeable in most cases. That in turn will put consumers in a better position, provided Zen products are priced competitively, as they most likely will be.

Then there's the hopeful aspect, where Zen is able to match or slightly beat current gen Intel clock for clock, while providing more threads to boot, all for less money. If that happened, people would basically be getting today's Intel for less money and that's a good thing for consumers. Tomorrow's Intel CPUs won't be 20-30% better than today's, probably more like 2-10%, as we've seen with the generations over time. And then, as I mentioned before, that's not really big enough difference in performance to care.

How many folks even here on this forum are still using their first or second gen i5s or i7s? Quite a few, from what I have read. Why? Because, for the most part, they're still plenty good enough. If Zen can match Ivybridge, but do so with 4 cores, 8 threads and less cost than the i5 of the day, then that's massive upgrade for anyone running anything worse than an Ivybridge i5.

But it's not all about "getting more for less". It's about getting better value for the dollar, because the market actually has some real competition that helps to mitigate price gouging. And it's also about bringing back some innovation to the desktop scene, because there are so many idea that never materialized over the years, the result of which is a pretty boring set of hardware to enthusiastic about. Why can't I buy a motherboard with 8GB of GDDR5 and two DDR4 DIMMs? Because the industry stopped trying to make cool stuff and has instead focused on the mundane goal of making the same old stuff run cooler. *yawn* L3 cache on a Pentium motherboard was cool. On die L2 cache in the Celeron 300a was cool. 64Bit and dual core were cool. A 15W cpu that's about as powerful as a 2010 model? Yeah, colour me, "I don't care".


Trolling is not allowed here
Markfw900

Edit: editing my post now buys you a long vacation

 

[Lepton87]

So:

- 80% (relative) difference in the claimed performance per watt is insignificant?
- The history of questionable statements AMD has, doesn't make further claims (which is Zen matching Broadwell-E IPC in this case) questionable at all?

80% How did you get this number? 2.8 is 40% more than 2 not 80%!

Relative? And here I thought % were always relative, there's a new thing to learn everyday

 

[ashetos]

80% How did you get this number? 2.8 is 40% more than 2 not 80%!

Relative? And here I thought % were always relative, there's a new thing to learn everyday

It depends on whether the comparison is with respect to a starting value or not: https://en.wikipedia.org/wiki/Relative_change_and_difference

There's a new thing to learn everyday

 

[ShintaiDK]

It depends on whether the comparison is with respect to a starting value or not: https://en.wikipedia.org/wiki/Relative_change_and_difference

There's a new thing to learn everyday

Exactly!

 

[DrMrLordX]

Most likely, AMD's Blender didn't allow for 256-bit.

How do you know that? The Stilt found that AVX/AVX2 support made no difference in Blender performance on Intel processors.

I hear that Lisa Su herself has been buying up all the stocks of RX480 and that they will all show up as liabilities in the Q3 earning report.
/s

tsk, you know the Ethereum miners are the ones buying those cards.

 

[Abwx]

- The history of questionable statements AMD has, doesn't make further claims (which is Zen matching Broadwell-E IPC in this case) questionable at all?

What is questionable is your ability to take the numbers for what they are once they are at odd with your IPC "estimations", they could have used Cinebench 11.5 that it would had displayed the same figures, it s just that it would have given more info to their competitor, why should they do so when the latter didnt even disclose their SKL uarch before the last HC, that is, one year after it was released, so much for transparency...

 

[Arachnotronic]

What is questionable is your ability to take the numbers for what they are once they are at odd with your IPC "estimations", they could have used Cinebench 11.5 that it would had displayed the same figures, it s just that it would have given more info to their competitor, why should they do so when the latter didnt even disclose their SKL uarch before the last HC, that is, one year after it was released, so much for transparency...

Intel disclosed the SKL micro architecture back at IDF 2015.

 

[Abwx]

Intel disclosed the SKL micro architecture back at IDF 2015.

That s not the subject but they actually disclosed it only at HC 2016, hence the recent articles on the thing, hardware.fr state that they asked for infos that never come despite promises made one year earlier, so much for your 2015 date.

On rappellera que l'année dernière durant l'IDF, Intel nous avait promis plus de détails sur le sujet, sans jamais nous les donner !

http://www.hardware.fr/news/14761/hot-chips-m1-sve-parker-info-skylake.html

What do you see on the slide below, HC 2015..?..

https://www.computerbase.de/2016-08/kaby-lake-prozessor-intel/


Of course one does need much effort to imagine what did prompt them to do so..

 

[Arachnotronic]

That s not the subject but they actually disclosed it only at HC 2016, hence the recent articles on the thing, hardware.fr state that they asked for infos that never come despite promises made one year earlier, so much for your 2015 date.


http://www.hardware.fr/news/14761/hot-chips-m1-sve-parker-info-skylake.html

What do you see on the slide below, HC 2015..?..

https://www.computerbase.de/2016-08/kaby-lake-prozessor-intel/


Of course one does need much effort to imagine what did prompt them to do so..

Dude...

http://www.overclock.net/t/1570069/idf-pdf-skylake-microarchitecture-details

 

[coffeemonster]

Clock for clock, a "construction core" is anywhere from 40-60% worse than an "i-Core", which significant to the point where it can actually be noticed in day to day use, let alone high-end tasks.

I mostly agree with your post, but I have gone between piledriver/sandy and kaveri/haswell many times on desktops and laptops at work and friends machines. There is no perceivable difference in "day to day use".

 

[Abwx]

Dude...

http://www.overclock.net/t/1570069/idf-pdf-skylake-microarchitecture-details

Nothing that ressemble AMD exhaustive explanation of their uarch, frankly, that s poor, no wonder they got back at HC one year later.

To get back on topic you can see on the slide above that SKL has two 128b/256b FP ports while Zen has four 128b ports, if anything this should be more efficient in legacy SSE, wich is the norm since there s no way that AVX or even FMA can be used for more than little parts of a code.

 

[Arachnotronic]

Nothing that ressemble AMD exhaustive explanation of their uarch, frankly, that s poor, no wonder they got back at HC one year later.

http://www.intel.com/content/www/us...-ia-32-architectures-optimization-manual.html

More detailed info in the optimization guide that's been out since last year.