zen and skylake morphcore

[VanayadGaming]

Hello! take this with a lot of salt,not just a grain, but if the rumours are true it might start some interesting things.

Amd and Intel both seem to be working on something called morph core which is able to take one core and divide it in to several virtual smaller ones if needed. This would drastically improve performance and power consumption.

paper for morph core: http://hps.ece.utexas.edu/people/khubaib/pub/morphcore_micro2012.pdf

Source for article: http://wccftech.com/intel-preparing-dirsuptive-skylake-microarchitecture-morphcore/

[ahhming]

lol

[Unhelpful]

Hello! take this with a lot of salt,not just a grain, 

It's WCCFTech.

 

Way ahead of you.

[VanayadGaming]

Sorry for the title mistake, mobiles are hard.

[X1XNobleX1X]

[soaringchicken]

Cause we all know how reliable that source is

[VanayadGaming]

I really don't get the hate. They predicted a lot of stuff lately.

[Menzenski]

This sort of technology is definitely coming in the future, but I'm not so sure it'll be with skylake.

 

One of my professors worked on a similar project a few years ago, and there are tangible benefits.  It's basically the best of both worlds.  Say you have a general 8 core cpu.  It'll handle parallel workloads very well, but will fall behind in single threaded workloads.  But imagine if those 8 cores can act as two large, powerful cores - you gain the single threaded performance when needed.

[Opcode]

So Hyper-Threading has finally had its own babies? Problem with this is it wont be that beneficial for consumers. AMD is likely to push 8 core 16 thread Zen chips in Q4 2016 (Intel has their 5960X) meanwhile most software and games barely make use of 2-4 cores today.

[HippY]

So Hyper-Threading has finally had its own babies? Problem with this is it wont be that beneficial for consumers. AMD is likely to push 8 core 16 thread Zen chips in Q4 2016 (Intel has their 5960X) meanwhile most software and games barely make use of 2-4 cores today.

yes, but then again we are talking about games, and 5960X are not meant to be the best in gaming.

It will be beneficial for the potential customers, the power users, and the non power users as well down the line if it will be implemented as like hyper threading.

They might just not know about it unlike power users.

[Opcode]

yes, but then again we are talking about games, and 5960X are not meant to be the best in gaming.

It will be beneficial for the potential customers, the power users, and the non power users as well down the line if it will be implemented as like hyper threading.

They might just not know about it unlike power users.

The problem is consumer software is not keeping up with the hardware available today. With 8 cores 16 threads becoming standard pretty much by the time Zen launches there's very little that even more threads can offer consumers. Especially given how crippled them threads become due to further fighting over resources. This approach would be beneficial in a super computer or something designed to leverage massive amounts of threads (possibly servers) tho I just don't see it being beneficial for consumer chips. The 10% performance increase I don't think will even catch Intel's attention. While it's something that may come to be in consumer chips eventually I don't see it happening any time soon. We first need a major overhaul of most software to leverage more hardware. Which is difficult because most software doesn't need a lot of threads or resources.

[NAP51DMustang]

I really don't get the hate. They predicted a lot of stuff lately.

One time Linus called them out on something that WCCF predicted in which they were wrong so now WCCF can't be correct since they were wrong 1 time. And because if Linus says it it must be true.

[HippY]

The problem is consumer software is not keeping up with the hardware available today. With 8 cores 16 threads becoming standard pretty much by the time Zen launches there's very little that even more threads can offer consumers. Especially given how crippled them threads become due to further fighting over resources. This approach would be beneficial in a super computer or something designed to leverage massive amounts of threads (possibly servers) tho I just don't see it being beneficial for consumer chips. The 10% performance increase I don't think will even catch Intel's attention. While it's something that may come to be in consumer chips eventually I don't see it happening any time soon. We first need a major overhaul of most software to leverage more hardware. Which is difficult because most software doesn't need a lot of threads or resources.

Right I see

 

I am writing close-to -hardware codes nowadays (PIC chips in C)so I still have loads to go until I 100% see the programming as a whole

 

but I can put a bitmap of "dickbutt" scrollin through an LED array so I am on the right way

[SpaghettiCarbonara]

Hey guys, look! It's WCCF! Let's all jerk eachother off while we all flap our tongues about how bad WCCF is!

[qwertywarrior]

ur talking about visc ?

http://www.extremetech.com/extreme/192858-visc-cpu-virtual-core-design-emerges-could-this-be-the-conceptual-breakthrough-weve-been-waiting-for

[VanayadGaming]

The thing I see good about this that it could dynamically change based on your needs. You need more single threaded performance it will morph in to 2-3big cores, you need more multi threaded performance it will split In to even more. Also, considering there is a paper behind this I think it could be possible that it will appear. I see this great for consumers as some of you already mentioned not all applications use 8 cores 16 threads or so. If threads hey can only use 2 cores then why not have 2 beefy cores.

[Hatsune Miku 「 」]

So Hyper-Threading has finally had its own babies? Problem with this is it wont be that beneficial for consumers. AMD is likely to push 8 core 16 thread Zen chips in Q4 2016 (Intel has their 5960X) meanwhile most software and games barely make use of 2-4 cores today.

Would buy 4 zen chips if they were 8 core 16 threads and had similar single core performance to a 5960x.

More machines to do computation, data processing, version control and run my private email, custom built search engine, image/file hosting and vpn on.

[TERAFLOP]

ur talking about visc ?

http://www.extremetech.com/extreme/192858-visc-cpu-virtual-core-design-emerges-could-this-be-the-conceptual-breakthrough-weve-been-waiting-for

Yup. this VISC. http://wccftech.com/amd-invest-cpu-ipc-visc-soft-machines/

Which AMD is a chief investor in and Intel isn't. Soft Machines the company behind this core morphing technology has already demonstrated working hardware. A tiny mobile ARM CPU had significantly better single-threaded performance than even Haswell.

 

[azurezeed]

if true skylake will be interesting

[VanayadGaming]

Seeing that I want to see this in action even more !

[vm'N]

One time Linus called them out on something that WCCF predicted in which they were wrong so now WCCF can't be correct since they were wrong 1 time. And because if Linus says it it must be true.

They are generally mistrusted on other hardware forums aswell.

They are known to post ridiculous prediction, based on ridiculous rumours.

[vm'N]

So Hyper-Threading has finally had its own babies? Problem with this is it wont be that beneficial for consumers. AMD is likely to push 8 core 16 thread Zen chips in Q4 2016 (Intel has their 5960X) meanwhile most software and games barely make use of 2-4 cores today.

Many even predicts that in the future (5-7nm) only 5-10% of the die would be for x86 cores.

x86 are very specialized in serialized workloads, therefore they lose their value after a certain quantity.

[patrickjp93]

They are generally mistrusted on other hardware forums aswell.

They are known to post ridiculous prediction, based on ridiculous rumours.

Of late their rumors have ended up right more than 90% of the time. WCCF has some strong anonymous sources for sure, at least nowadays. I was speculating for about one week after TSMC's admitted 16nmFF schedule being derailed that it would be in Nvidia's best interest (making it very likely they would) to find a new foundry. I said it would be most beneficial for them to team up with Intel to get Pascal out on true 14nm 2nd gen. FF just after the 300 series in order to cut the 390X off before too much damage was done, and lo and behold they went with Samsung shortly after I made the argument on WCCF.

 

Sometimes what seems like ridiculous rumor isn't so ridiculous when based in solid reasoning. WCCF thought I was crazy, and yet it happened. Then they continued to say I was a moron for thinking Nvidia would move to and I quote "an amateur foundry like Intel which no one in the industry uses to make their chips, especially when they proved they couldn't make good GPUs with Larabe." The posters on WCCF might be crazy sometimes, but for that website rumor has ended up true far more often than not.

[Ren]

I really don't get the hate. They predicted a lot of stuff lately.

"Predicted"

Even you think that they're making up shit on the fly and are sometimes correct when they occasionally have an actual source.

[PortentousLad]

thank you.