AMD CPUs, SoC Rumors and Speculations Temp. thread 2

Page 64 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.


Pessimist or realist? That is the question 😛
 


GPUs are irrelevant. Since the 14LPP node is optimized up to 2.5GHz, you will not see any regression for cards that barely hit 1.4GHz, but you will see regressions on any CPU that targets >3GHz clocks.

Same situation with Carrizo, which was fabricated on a 28GFA process that did target lower clocks than Kaveri --I provided the relevant slide figure before--. At the lower end, Carrizo mobile models clocked higher than Kaveri mobile equivalents, but Carrizo for desktop was canceled because it couldn't hit 3.7GHz base to compete with top Kaveri model.
 


There is no a simple answer to that. It depends on how complex/costly is the microarchitecture, which is the foundry cost, the sales volume, the return tax,...
 
Hey folks... fancy some comedic reading?

http://www.pcadvisor.co.uk/new-product/pc-components/amd-zen-processor-release-date-price-specs-features-3643552/

Apparently 8 core Zen is going to cost £180 (including vat).... so much misinformation in that article it's unreal.

I'm also fairly sure they've linked to proven fake slides from WTFBBQTech...
 


I will take 3.9
 


Don't worry. the more hardcore fans will eat it up in clicks and spread the fudd.

Always fun awaiting new product launches.
 
"The new architecture release date hasn't been officially confirmed but reliable sources sites such as WCCFtech (...)"

bender_youre_serious.gif

EDIT: Ok, I'll try to be a bit more constructive...

*If* those prices are somewhat believable, then it will be a disaster. The i5 6600K is £225 with tax ("VAT incl") and if the top end part is £180, then it does NOT smell good.

Cheers!
 


Yes they link to one of the fake Zen slides. It is more easy to mention the pair of things that they got right than list all the mistakes and nonsense that they published. As you mention the 8-core Zen going to cost ~$200 (without VAT) is impressive, until I noticed that they claim the 32-core server Zen will cost ~$440. Their pricing of the server chip must be wrong by a factor of 12x or so. I was going to take that as their flagship mistake until I read this masterpiece:

The 14nm Zen processor is set to have a 40% boost in performance in its clocks, whilst having up to 15% improvement over the previous Excavator architecture.
Click to expand...
 

*sigh*

Sadly, there is basically zero media accountability in the internet age...
 


Well to be fair we're looking for honest leaks which is like trying to find an honest thief. 😉
 
http://forums.anandtech.com/showthread.php?p=38363321#post38363321
The most exciting part is core clock. The 8c/95W variant's base clock is 2.8GHz, all core boost is 3.05GHz and maximum boost is 3.2GHz.
The 4c/65W part's clock is the same. (I would expect 3.5GHz base clock for a retail 4c/95W variant.)
Idle clock is exciting as well. AM4 versions can lower the clock to 550 MHz in idle which is a very nice level from an AMD CPU. Idle wattage is 5W for 8c version and 2.5W for 4c version.
The SP3 versions have even lower idle clock: it's only 400MHz. Regarding the boost clocks the 32c/180W version has a 2.9GHz boost clock and the 24c/150W version has a 2.75GHz boost clock.
Click to expand...
 
If reading comprehension is not failing me, in that post (with the copious amount of salt it requires) is stated they are ES CPUs.

Base of 2.8 for *all* cores active while being ES is not that bad. so, a base of ~3.2Ghz is not that out of the picture. Anything above that starts escaping reality though.

Again, that is taking into consideration it is a forum post.

Cheers!

EDIT: Changed word.
 
There are many posts and news about am4 release, experts said that it will be released on 4Q 2016 - 1Q 2017. There will be more than one variant, so it's supposed that will be a powerfull FX 8k and 9k series, up to 8 cores - 16threads at 4,2 - 4,5GHz (turbo) running under Zen, and using high quantity of cache (8mb/core L3) and using HBM (high bandwich memory) technology over this cache, (as we saw in Fury graphics) for those who don't know what means or what HBM do, i'll try to explain simply. Fury has 4GB RAM, and it accomplish a bandwich of 4096bit (this means 1bit per mega), at 1000MHz clock speed it gives an awesome performance. Ok, lets suppose we have an 8MBL3/core cache (8x8 = 64Mb) and lets suppose that we can fastly clear and write this cache at 64bit bandwich. We will have the first all-64bit working processor capable to pass 64 bit data through data bus (everyone) at native 64bit bandwich and (if it continues like FX- series) it may be capable to "eat" 64GB RAM DDR4 (i hope at more of 2400MHz native).
Compared with 1151's intel i7 6700, that have less memory bandwich than old FX-8370 (25,4 vs 29,9 GB/s), and 32bit cache. Lets be open-mind and think what intel did. intel forced DDR4 to appear, but their processors are only capable to handle DDR4 at 2133MHz (remember that DDR3 max speed is 2400MHz). So, as we saw in AM3 APUs, that are running over DDR3 - 2133 native, we should get a native DDR4 3000 on AMD AM4 7'gen.

I'm really hiped since i had an amd 8730, then an i7 4790k and saw the only difference at pro-user level is PRICE (some temperature and consumption too, but who matters? LoL) So i want to sell or give that i7 to a familiar and get AM4 flagship, and i want it as soon as possible! AMD please, don't tease us too much 😛
 


I was searching info since a week ago. Just search amd am4 on google, then go to search settings - date - last week, and keep reading for hours. We definitely have to wait to see real Zen working. First mobile APUs, then desktop APUs and CPUs.
But i dont think as some people said here, that flagships will be 3,5 GHz (turbo) so AMD always granted nice GHz and bet for 64bit architectures. We can see that intel and AMD, they are doing a pulse of force on this way, giving nice hardware at (more or less) nice price. Both will stuck in 4,4 - 4,5 max GHz this year, maybe next 2017 will bring us 5GHz on stable (without need of liquid) taking a bit more TPD and PC (thermal and power consumption).
If we base our reasons on which is not possible to put in-processor modules nearest than 14nm, the war will go on optimize software and bring more caché quantity (i'll not surprised to see 1GB cache in next future) and using better caching technollogies (software again after all, based on low level architecture).

What can we expect?
Intel brings someone "new", then AMD doing the same as intel. AMD brings some more "new", intel doing the same as AMD do, like always
(2 examples: AMD created 64bit atchitecture, intel said that it was no important and laugh. 2 years later inter ran to bring 64 bit architecture by needings, then intel made good multi-threading architecture. AMD start doing the same to be on the way, and bring competition in the market.
More recent: Intel brings low TDP and power consumption to get (quite less, as i tested) performance, but high end cpus at all (intel apus are not valuable with <mod edit> intel hd graphics, as for gaming or more usability, Radeon wins 10 times) Then AMD is running to bring lower TDP and PC as intel did (amd is getting really late on this point) then AMD throws HBM, now intel will bring HBM too).

So it's in CPU the real future? Faster SSD, better graphics cards (lower TDP and PC too bringing same performance), better ram... but CPUs need to allow more MHz as native (must set up to 3200).
There are no other ways to grow nowadays

 
Your arguments have little consistency, Don.

For example: "so AMD always granted nice GHz and bet for 64bit architectures".

AMD will only go as far as the process node used allows them to. Take a look at Carrizo and Kaveri, for example. By just changing/tweaking the process to be more dense, they improved efficiency a lot, but had to sacrifice speed for it. Carrizo, being clock per clock superior to Trinity, is slower in CPU only tasks because of the speed difference.

Also, all companies invest in new technology that might or might not live long enough to be remembered. Case in point, the 64bit thing was a success for AMD thanks to Microsoft in big part and then because of the server needs. Intel pushed his own version (EMT64) but had little success in the Windows world. Now, that being said, do you remember "3DNow"?

I really think you should give this thread a good read from the start. It will beat googling a lot of stuff, because most of the important information is already here.

Cheers!
 


Not EMT64, that is what Intel calls their x86-64. You mean IA64, or the Itanium.

And it failed because it had to emulate x86 code since it is a pure 64bit CPU and in doing so it was slower at x86 code than a normal x86 CPU.

AMD64 succeeded because it was a fully x86 CPU with 64bit extensions that allowed it to perform the same in x86 mode as a non AMD64 CPU while giving the option for 64bit.

To top it off Microsoft also launched their first consumer 64bit, XP 64, so there was a OS to tweak and test with. It was a horrible OS, bad driver support and lots of crashing, and the best benefit I remember was in Far Cry 64bit where they added a few more plants and birds vs the 32bit version. Still it helped to push 64bit faster than before, Itanium launched before a normalized 64bit consumer OS was around.
 


As an architecture, Itanium is far better then X86-64. due to being a clean break from X86. As you noted though, the loss in performance executing X86 code did the architecture in. Today, now that Win32 isn't as big an issue, Itanium would be a far better option, were it not killed by software lock in.
 


No, Jimmy. I did mean "EMT64". It's true that it's similar, but it has key differences with AMD64 (or however you want to call it) that compilers had to be aware when you compiled. I don't want to get technical, but I do remember in the early days of AMD64 in Linux and it was a PITA for Intel people, haha. If the binary was AMD64, it used to fail if not compiled for compatibility.

Cheers!
 


Yes there were differences but I wouldn't say it had little success as compared to Itanium. EMT64 is still what is used and Intel now calls it "Intel 64".

Either way 64bit is not an advantage to either company. Both have full support of 64bit and that is now a discussion of the past.

What matters now is the efficiency of the process, I am sure a lot of people hope Samsungs 14nm LPP is better for CPUs than it seems to be for GPUs, and the uArchs performance with some instructions, if used at all.
 


People in that forum is expecting ~200MHz extra for the final steeping. This means they expect 3.0GHz for final silicon, which is very close to the 3.0GHz I wrote time ago. 😀
 


The Q4 launch was disproved by AMD people as being invention from the media. Early 2017 seems correct.



Zen will run barely above 3GHz. I am rather sure that 4.5GHz will be not achieved even using watter cooling. There is on HBM on Zen, neither in the CPUs, nor in the consumer APUs. The only APU with HBM is the HPC APU expected for 2018.



The differences between both are many more.
 


AMD didn't always granted GHz and it is not the case with Zen. AMD must be doing a pulse force with Intel. Intel is not. AMD is not in Intel radar. Intel is worried by real competition coming from Apple, IBM, Sun, and ARM server guys.

AMD Zen will not hit 4GHz. And Intel will not hit 5GHz next year. The GHz race finished time ago.



1GB cache how? L2 is not the same than L4, adding private caches is not the same than one giant shared cache...



HP and Intel invented a revolutionary 64-bit architecture first. Threy failed, because was too experimental. Then AMD did the easy thing and developed AMD64, just a 64bit extension of the old x86-32. The AMD64 architecture is considered one of the worst and ugliest 64bit architecture in use.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom