AMD CPU speculation... and expert conjecture

[Ranth]

Juan what I don't get is that you seem (atleast to me) say that the principle of locality will fix all problems with thermal-density. Then could you explain to me why we haven't seen any high end server apus? You said the resarch was done several years ago, if this "principle of locality" was the holy grail of performance per mm^2 that you make it look like, why haven't we seen it ramping up earlier? Why is it that there afaik is 3-5 years until we see such product?

It might be my lack of patience to search through the net, but would you give me an article with Intel dropping the xeon + phi for an apu like alternative? I can see AMD focus on the APU since they can't compete with intel in the high-end server space, and Nvidia too, since they are in the same boat as AMD (HOWEVER i Don't know enough on this subject, but I to see nvidias golden egg before I believe it).

Also I don't claim that I have more knowledge than the engineers, that is just you calling me out for being stupid, and from my point of view that is an attempt to try to get me to shut up... Also I'm not saying you are wrong I just do not believe you, because of lack of evidence/proof/counter argument to the points above.


The real important stuff:

I am on these forums to become smarter(!), but as it is with the internet I do NOT trust it, so if something sounds far fetched (Which the idea of the 'principle of locality', negating the effects of thermal density does) I tend to try an argue against it. I would love if you could try and explain how it works, BECAUSE I want to become smarter, telling me that some weird quantum effect causes tdp to not matter, without explaining why, doesn't do me anything...

So what I would love for you to do is:
Explain how this "principle of locality" works, and HOW it negates thermal density <-- (That is what I do not get)

AND i DO NOT want the "well obviously this is teh best since all the clever people are doing it" No I want to know why.

 

[8350rocks]

Sp yeah... Richland is 11% faster than Trinity in Skyrim, and Kaveri having a lead of less than 9% Against Trinity is LAUGHABLE.

The design improvements for Kaveri increase multi-core "IPC" not really single core "IPC". Having dual decoders does nothing for a single core. AMD is going after efficiency and multi-thread performance.

Intel is going to lead in single core "IPC" as they have a very wide 7 port execution engine (including 4 ALU).

What was I saying 10-15 pages ago or more? This end result is precisely what I predicted it would be, and it still will not replace the 8350 in my PC.

 

[noob2222]

@ juan

All of them? Dont even attempt to call me a liar.

How do you get 34% improvement in RE 5?

You threw out this benchmark, thats how. Along with all the others that dont show what your trying to prove.

 

[8350rocks]

Juan what I don't get is that you seem (atleast to me) say that the principle of locality will fix all problems with thermal-density. Then could you explain to me why we haven't seen any high end server apus? You said the resarch was done several years ago, if this "principle of locality" was the holy grail of performance per mm^2 that you make it look like, why haven't we seen it ramping up earlier? Why is it that there afaik is 3-5 years until we see such product?

It might be my lack of patience to search through the net, but would you give me an article with Intel dropping the xeon + phi for an apu like alternative? I can see AMD focus on the APU since they can't compete with intel in the high-end server space, and Nvidia too, since they are in the same boat as AMD (HOWEVER i Don't know enough on this subject, but I to see nvidias golden egg before I believe it).

Also I don't claim that I have more knowledge than the engineers, that is just you calling me out for being stupid, and from my point of view that is an attempt to try to get me to shut up... Also I'm not saying you are wrong I just do not believe you, because of lack of evidence/proof/counter argument to the points above.


The real important stuff:

I am on these forums to become smarter(!), but as it is with the internet I do NOT trust it, so if something sounds far fetched (Which the idea of the 'principle of locality', negating the effects of thermal density does) I tend to try an argue against it. I would love if you could try and explain how it works, BECAUSE I want to become smarter, telling me that some weird quantum effect causes tdp to not matter, without explaining why, doesn't do me anything...

So what I would love for you to do is:
Explain how this "principle of locality" works, and HOW it negates thermal density <-- (That is what I do not get)

AND i DO NOT want the "well obviously this is teh best since all the clever people are doing it" No I want to know why.

That's because magic has no place in physics, and thermodynamics cannot be overcome with a hand wave and an assurance.

 

[tracker45]

So what that Kaveri is weak .


Mantle will make up for the losses.

 

[The Q6660 Inside]

So... look at what I stumbled upon... ASRock came out with a MoBo that supports 64Gb of RAM for AM3+...

http://www.asrock.com/mb/AMD/Fatal1ty%20990FX%20Killer/

Looks like a "Killer" deal. Bahahaha! Looks like I found a MoBo for my primary Rig.

Hasn't this been out for like 2 months?.. It looks like they are trying to kill the Crosshair V once in and for all by offering an updated Extreme 4 for the people who dont want to dish out $200 on a red/black mobo, although I love staring at my friends CHVFZ rig and personally I think it looks better, has moar SATA ports and has DAT PUURFECT UEFI... If only they could solve the sleep and fast boot issues...

 

[jdwii]

APU's are not about gaming, has that particular user been on Amd's site? Or is that person only using certain information to support his claims.

 

[Rum]

Looks like DX12 is focusing on fixing Driver latency related issues; don't I love being right?

http://www.fudzilla.com/home/item/33638-maxwell-is-not-directx-12-next-compatible

From what we heard, DirectX Next actually fixes a lot of latency related issues that are present in DirectX 11 and earlier versions. The new DirectX should have lower driver latency something that developers have complained for quite some time but we are not aware of any major feature set that will come with the DirectX Next.

Well if it turns out to be true then Mantle served us all greatly in that it pushed M$ to actually work on and fix their crappy API! But Mantle will still serve a purpose because it will be multi platform compatible.

"Multi-platform", meaning PC only, limited to just GCN based AMD GPU's.

"Multi-platform", meaning Multi OS, Not limited to just GCN based AMD GPU's, I have read in multiple articles they would let other vendors use the API in the future they just have to be willing to play ball. It's just terrible that AMD is wanting to push the gaming industry forward out of the stone age isn't it? -_-...

 

[palladin9479]

Juan what I don't get is that you seem (atleast to me) say that the principle of locality will fix all problems with thermal-density. Then could you explain to me why we haven't seen any high end server apus? You said the resarch was done several years ago, if this "principle of locality" was the holy grail of performance per mm^2 that you make it look like, why haven't we seen it ramping up earlier? Why is it that there afaik is 3-5 years until we see such product?

It might be my lack of patience to search through the net, but would you give me an article with Intel dropping the xeon + phi for an apu like alternative? I can see AMD focus on the APU since they can't compete with intel in the high-end server space, and Nvidia too, since they are in the same boat as AMD (HOWEVER i Don't know enough on this subject, but I to see nvidias golden egg before I believe it).

Also I don't claim that I have more knowledge than the engineers, that is just you calling me out for being stupid, and from my point of view that is an attempt to try to get me to shut up... Also I'm not saying you are wrong I just do not believe you, because of lack of evidence/proof/counter argument to the points above.


The real important stuff:

I am on these forums to become smarter(!), but as it is with the internet I do NOT trust it, so if something sounds far fetched (Which the idea of the 'principle of locality', negating the effects of thermal density does) I tend to try an argue against it. I would love if you could try and explain how it works, BECAUSE I want to become smarter, telling me that some weird quantum effect causes tdp to not matter, without explaining why, doesn't do me anything...

So what I would love for you to do is:
Explain how this "principle of locality" works, and HOW it negates thermal density <-- (That is what I do not get)

AND i DO NOT want the "well obviously this is teh best since all the clever people are doing it" No I want to know why.

What he's attempting to discuss is the fact that HPC systems are I/O chocked most of the time. You can build a multi-million USD 64 socket solution but it won't do you any good if you can't get the data to the CPU's fast enough. L1 cache has the absolute lowest latency and highest bandwidth of any memory implementation, so in theory if you could fit your entire data set into the L1 caches of all 64 sockets worth of CPU's (256~512 CPU's assuming 4~8 core per CPU) then your CPU's could perform at full potential. The problem any sane engineer runs into fast is space, cache is expensive space wise which is why we use dense system memory instead and try to prefetch / predict what's needed several steps ahead of time. Heat dissipation also quickly becomes an issue as you gotta remove all that waste heat created during this processing. The higher the thermal density the harder it becomes to cool it, that's why newer Intel CPU's are slamming into OC walls.

So if you had infinite die space (money) and a superconducting CPU (no thermal waste heat) then the fastest implementation would be a purely local single chip implementation. Of course that's not currently possible, it's one of those concepts like the Alcubierre drive, 100% possible in theory but 0% possible in actual practice. Still worth researching as there are always bonus side technology that develops even though the central concept is unreachable.

 

[Cazalan]

Possibly the most promising benchmark yet for Kaveri.

Translation:

Consumption playing BF3.
Watts, less is best.

 

[anxiousinfusion]

Possibly the most promising benchmark yet for Kaveri.

Translation:

Consumption playing BF3.
Watts, less is best.

Is that total system draw or just the APU?

 

[tracker45]

It's out !

Our lives now have meaning.

http://www.dabs.com/products/amd-a10-7850k-3-7ghz-kaveri-black-edition-95V8.html?refs=56020000&src=3

 

[The Q6660 Inside]

inb4AMDkaverisucksonhardforum

 

[andy1998]

http://www.newegg.com/Product/Product.aspx?Item=N82E16819113359 woah $190 on newegg...

 

[Master-flaw]

It'll be $160 in a couple months...or at least I hope.
Edit:
It's hard to justify the $50 price jump over the 6800K....The thing should've been priced at $150...

 

[anxiousinfusion]

http://www.newegg.com/Product/Product.aspx?Item=N82E16819113359 woah $190 on newegg...

Aaaaand.... I bought it anyways. Everyone point and laugh.

 

[etayorius]

190? NewEgg must be crazy... Kaveri is not that great to justify the price.

Heck i am starting to think that AMD price of 170 is rather high, they should had set it no higher than 160, at that price point it would have been a sweet deal.


Why laugh? if you decided Kaveri was worth the price, as long as you`re happy with your purchase no one should point and laugh, maybe you will put it to good use.

 

[de5_Roy]

ugh.. $190. not good. this is starting to look like bd launch.
even if kaveri offered core i5 level performance, it should be undercutting the core i5 cpus, not the other way around. worse, it's official msrp is $173-something, just $6-7 less than core i5 3350p. the msrp has to drop first.

anyway, as we just saw (and as i predicted), people will still buy it. i appreciate you guys being the early adopters!

@cazalan: very promising indeed! i'm looking forward to the reviews.

edit2:
a few core i5 cpus are hovering around $190
http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&N=100007671%2050001157%20600005579&IsNodeId=1&bop=And&CoreMin=undefined&CoreMax=undefined&Order=PRICE&PageSize=100

the younger kaveri has an even bigger markup
http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&N=100006519%2050001028%2040000343%20600484725&IsNodeId=1&Manufactory=1028&name=Socket%20FM2%2b
priced at $170 from edit4:[strike]$120[/strike] $152 msrp (according to at, a8 7600 is @$120), makes 7850k seem the better buy of the two. [strike]worthless garbage[/strike] centurion prices seem up as well..
canada doesn't look to be faring better
http://www.newegg.ca/Product/ProductList.aspx?Submit=ENE&N=100007670%20600484725&IsNodeId=1&name=Socket%20FM2%2b

 

[griptwister]

Hmmmm... I think we can all agree, Newegg must be on crack because they need to lower the price.

I've been thinking, what if these APUs are actually competitive with the i5 and you all are wrong? What if HSA does take off??? Lol, I'm happy with what I'm seeing from these APUs. We haven't even seen the A10-7850K compared to a i5 yet... so hold your horses. I swear, some of you guys just hop on the bandwagon just for the sake of being afraid to be wrong. I'm expecting AMD to release a real banger CPU after Kaveri. I think I'll pick up one of them MSI A88X G45 Motherboards. I think it'd look pretty sick in my rig. Get the MoBo now and wait for Kav to drop in price.

@de5_Roy, I know, all the brave soldiers buying the product before reviews!!!

@The Q6660 Inside, The ASRock Killer series has some pretty sick features on it. Probably more useful to me than a CHVFZ. 64gb ramsupportliekOMG Probably will use 16Gb tops though. lol

 

[de5_Roy]

@griptwister: the apus were always competitive against core i5 depending on what you wanted to do with them. they had the best performing igpu for a long time until irs pro. but they always had the best performing igpu for price, overpriced iris pro only made them seem better. at their previous price points, the apus were almost perfect until fx6300 prices dropped. the apus are still competitive in laptops in terms of performance. if hsa does take off and benefits amd, all the better.

if you can get a microcenter combo, that might be better. i checked mc website, kaveri hasn't shown up there yet... but mc might be where the first cheap kaveri shows up.

 

[griptwister]

de5_Roy, for sure, Microcenter has all the deals. Also looks more professional than Frys. Frys looks like freaking Toys R Us for tech geeks. Nothing wrong with that... except prices... and a boat load of open box items.

 

[deltapapazulu]

I do hope AMD Pull a rabbit out of there hat with this revision of the architectural silicone,
It will be interesting, Its always good to have competition in any sort of market.
we all know AMD has been a bit behind in the last few years. I can only hope this little baby in some ways will keep closer on par with Intel this time.
But we will see once the test or engineering samples are ready for testing.

The only time AMD has ever been competitive with Intel in mid to high range CPUs was during Intel's Pentium D fiasco of '05 and '06. From introduction of Core 2 onward, AMD has had nothing in answer to Intel. And before 2005 the same is true.

 

[de5_Roy]

oh, almost forgot: the first wave of reviews from bigger sites will be bound by review guides, so don't expect a lot of in-depth stuff outside the permitted conditions. hell, if amd hides something like having true audio logic blocks inside bonaire cards (or puts nda over it's publicity), you might never even know. the real good stuff will come out after the first wave. the only thing that may come out early if something glaring like bd's power efficiency (unlikely this time) or haswell's overclocking or splitting off reviews. other than those, expect almost every one to more or less say the same things. fanboy sites are exceptions, ofc. they'll award gold, platinum, unobtainium (this is what amdzone will give while bursting into tears of joy(ful pride)) must buys (makes you wonder if they ever check prices), absolutely must have..(..to buy) (really makes you wonder...), product of the year 2014 (in january!!), king of performance for price, great value et cetra.

 

[FALC0N]

I do hope AMD Pull a rabbit out of there hat with this revision of the architectural silicone,
It will be interesting, Its always good to have competition in any sort of market.
we all know AMD has been a bit behind in the last few years. I can only hope this little baby in some ways will keep closer on par with Intel this time.
But we will see once the test or engineering samples are ready for testing.

The only time AMD has ever been competitive with Intel in mid to high range CPUs was during Intel's Pentium D fiasco of '05 and '06. From introduction of Core 2 onward, AMD has had nothing in answer to Intel. And before 2005 the same is true.

You obviously don't remember this period well. AMD was VERY competitive with Intel on the mid to high end from the release of the K7 in 1999 until the release of conroe in 2006.

 

[Cazalan]

Early adopters always pay some premium. You can't expect the best prices on launch day. The initial volumes look low so Newegg (limit 1) can justify the price. They have the 7700K at limit 2.

And the first A10-7850K Geekbench 3.

http://browser.primatelabs.com/geekbench3/326781