"Vishera" End Of The Line for AMD FX CPUs: Roadmap

[HumanSmoke]

I think you've missed my point. I know they can't make CPUs run at 20GHz. I'm talking about what could have been.

That was my point. It couldn't have been- it would be physically impossible...unless you think that a single core with the energy budget of a POWER7 module was feasible for a desktop CPU. Theoretically I think it's closer to Galaxy Quest or Farscape than actual real life..
If you're musing on what might have been, there are plenty of "what if" scenarios that actually could have happened and would have substantially more impact on the industry:
1. Bob Noyce doesn't invest and supply start up capital for W. Jerry Sanders III. Without Noyce's investment, other backers shy away (as it was, Sanders only made the investor deadline with five minutes and $5K to spare). AMD kaput before it starts, IBM's second source for 8088 processors likely falls to National Semi, Motorola, or Zilog.
2. Gary Kildall actually gives a shit about running a company and keeps his appointment with IBM's reps rather than disappear to fly his plane. IBM choose CP/M for the Model 5150...Bill gates and MS-DOS don't get a look in.
3. Jim Harris, Bill Murto, and Rod Canion decide to go with the option of sinking their money into a Mexican restaurant. Compaq doesn't happen, the IBM ROM-BIOS isn't reverse engineered, and the IBM PC clone business either doesn't happen or is stalled past the tipping point where anyone can undercut big blue. More to the point, IBM would have then realized that personal computing's growth warranted more attention/budget that was being lavished on it's mainframe and minicomputer business.

These things all could have happened quite easily, just as Hewlett-Packard could have listened to Steve Wozniak when he approached them about building a personal computer

Better artificial intelligence would have probably been one of them.

Actual intelligence (i.e. the brain) uses parallelization. Speed is pretty much a constant AFAIK limited by chemical and electromagnetic action. Boosting the latter seems to lead to erratic behaviour (analogous to cache misses ?), losing parallelization (lowering core count ?) leads to Alzheimer's and a new found love of reality TV.

 

[qubit]

I'm not really sure what you're arguing about? I was just musing dude.

 

[HumanSmoke]

I'm not really sure what you're arguing about? I was just musing dude.

Ah, okay. For musing, 20GHz seemed quite conservative. How about 1024 cores @ 1THz with a TDP of 5 watts and an MSRP of $9.99 ?

 

[qubit]

2THz

 

[Steevo]

Soon we will be limited by the speed of electricity through the semiconductor traces and wire. Then we are on to optical multiplier processors or to emi processors. Then possibly to quantum processors or quantum bit check emi processors.

 

[HumanSmoke]

Soon we will be limited by the speed of electricity through the semiconductor traces and wire. Then we are on to optical multiplier processors or to emi processors. Then possibly to quantum processors or quantum bit check emi processors.

I hope you get due recognition when AMD incorporate all this into their 2014-15 roadmap PPS next month

/Waits for WCCF to repackage this as front page article

 

[TheoneandonlyMrK]

I hope you get due recognition when AMD incorporate all this into their 2014-15 roadmap PPS next month

/Waits for WCCF to repackage this as front page article

yeah using old style switch ,logic on graphene they could have 20Ghz or more in the bag without tickleing quantum's tum

 

[BiggieShady]

Soon we will be limited by the speed of electricity through the semiconductor traces and wire.

Speed of electricity is actually speed of light (the one we are always limited by) - interestingly enough, movement and speed of electrons in (semi)conductor has nothing to do with this - it's the disturbance in the electromagnetic field that travels. Use the force.

 

[Ravenas]

"AMD will continue to supply AM3+ and AMD FX processors for the foreseeable future, as per AMD's official roadmap update at APU'13 [above]. Recently, AMD launched the FX-9000 series, AMD's fastest desktop processor to date. As AMD's business continues to evolve, AMD will focus on the areas of growth including support for the desktop PC enthusiast leveraging AMD's world-class processor design IP, including heterogeneous compute. AMD's FX branded products will continue to evolve and we look forward to sharing those updates in the future," said James Prior, an AMD manager of APU/CPU product reviews, in a conversation with Gamer’s Nexus web-site.

The roadmap supplied by the OP seems like a stretch of the imagination. FX processors are a core business of AMD.

 

[Steevo]

Speed of electricity is actually speed of light (the one we are always limited by) - interestingly enough, movement and speed of electrons in (semi)conductor has nothing to do with this - it's the disturbance in the electromagnetic field that travels. Use the force.

http://en.wikipedia.org/wiki/Speed_of_electricity

Well understood years ago. But a trace operating at 70% of the speed of light.

209 854 721 m / s

At 5Ghz switching rate means it can only travel.

0.0419709442 m / s

and that is 1.6524 inch per second, so this is the longest any trace can be assuming everything works perfectly, and you have to be able to read and write data out of caches at this incredible rate too unless you want a significantly higher percentage of time in wait states.......that causes timing issues at speed too.

So we are getting close to how fast we can make processors switch unless they start learning about capacitive roll off for every switch and transistor, and the logic to do that is cost prohibitive. Intel was the first to find this theoretical limit when trying to reach the absurd speeds they thought the P4 possible, they then wrote papers on it and it as well as the poor performance and other issues were the reason they moved to a shorter pipeline and higher IPC instead of higher speed. A horse AMD now seems stuck beating aimlessly.

 

[xorbe]

and that is 1.6524 inch per second

Close, 1.6525 inches per clock cycle. Less when you consider the propagation through transistors.

 

[TheHunter]

All they need is make those APUs 8core and all will be good.

I see this newest is already a steamroller

 

[NeoXF]

FX APUs anyone? (As in beyond "A10s").

 

[eidairaman1]

FX APUs anyone? (As in beyond "A10s").

the 8 core apus will probably be A10s, A8s will be Hex Cores and A6s will be Quads and A4s will be duals.

it would only make sense if they lined them up according to the number- so A10 are ten core, etc etc

 

[TRWOV]

I suppose 6 core APUs are a strong posibility once AMD moves to 28/22nm.

 

[NeoXF]

What I meant was, APUs with more cores for software/workloads that just won't scale well with HSA, L3 and maybe eDRAM (L4?) as well and higher TDPs, since I imagine this won't probably happen on 28nm, 20nm is a big contender, especially since AMD already announced it's TDPs for the stock family of APUs will top out at 65W. And I don't know about you guys, but 95-100W sounds about right to me, hell, looking at Intel, 130-150W for a HPC targeted platform sounds just about right.

6M/12C Excavator w/ L3, triple/quad DDR4 IMC and 16 PI GCN compute units @ 20nm @ 125W TDP please?

In any case I don't think AMD will release a 6-core Kaveri anytime soon, unless they change up the roadmap (wouldn't make sense w/ 20nm 65W Carizzo replacing 28nm 95W Kaveri I guess). But I think we can expect a Kaveri clock refresh later next year tho.

 

[TheoneandonlyMrK]

Wheres Bta , guy needs a head sort.

AMD retorted to this tale of woe not long after it was released stating.

FX not dead and not being sidelined or forgotten and will progress into the future.

AM3+ wasnt mentioned positively or neg but likely wont get beyond 2014

and that the slide in question is Bs ,not Amd's

also its one year view doesnt state what comes after FX because they are not discussing that at this time, not that when the chart ends so does FX thats just not a fact.



8 core APU's or 6 even wont arrive until TSV 3d chips become easy(and cheap) to make(2015-16) and you can bank that opinion as fact. until then server scrap parts will make up future FX chips, also bankable and imho before 2015 am3+ WILL see steamroller cores(im banking this one though it is just my opinion based on future server upgrade options for the big data crew)

 

[TRWOV]

Wheres Bta , guy needs a head sort.

AMD retorted to this tale of woe not long after it was released stating.

FX not dead and not being sidelined or forgotten and will progress into the future.

AM3+ wasnt mentioned positively or neg but likely wont get beyond 2014

and that the slide in question is Bs ,not Amd's

also its one year view doesnt state what comes after FX because they are not discussing that at this time, not that when the chart ends so does FX thats just not a fact.

Huh? Nobody said the FX line is dead, just that Vishera is the last core developed for AM3+. In fact AMD's retort just confirmed it: AM3+ wont get Steamroller, it won't even get a 28nm refresh.

Now what Bt should do is change the title: didn't know this many people were dialexic, lots are reading "End of the Line" as "End of Life".

 

[TheoneandonlyMrK]

many on here are , thats ahh feckit


"AMD will continue to supply AM3+ and AMD FX processors for the foreseeable future, as per AMD's official roadmap update at APU'13"



no end point or date was given for FX or AM3+



just because a chart ends does not mean anything the next chart will dictate whats next or how long am3+ is here........................

no metion of vischera's reign or any future plans at all were mentioned in fact the only vischera comment was that they had just brought out the fx9xxx's

how would it be wise to talk up what's next while you are still clearing stock of what's here??

 

[xenocide]

"AMD will continue to supply AM3+ and AMD FX processors for the foreseeable future, as per AMD's official roadmap update at APU'13"

There's a huge difference between supply and support. Sure, they will continue to manufacture and sell AM3+ parts, but will they develop a new CPU to put in there? No. The big concern is AMD either consolidating the FM2+ socket or developing a whole new socket for FX-series CPU's. People have grown very accustomed to AMD continually supporting upgrade pathes, removing them might piss some folks off.

 

[NeoXF]

There's a huge difference between supply and support. Sure, they will continue to manufacture and sell AM3+ parts, but will they develop a new CPU to put in there? No. The big concern is AMD either consolidating the FM2+ socket or developing a whole new socket for FX-series CPU's. People have grown very accustomed to AMD continually supporting upgrade pathes, removing them might piss some folks off.

Well, FM2+ is here to stay untill mid-2015 at least, when I predict AMD will have to update it's APU platform for the DDR4 version of Carizzo.

It's also the best time when I think AMD should make a real succesor to 990FX and AM3+, since DDR4 will be there for awhile. Sure, once can speculate SATA-Express, PCI-Express 4.0, maybe support for more than 2 64bit buses and so on.
Edit: So they might as well merge the two.

 

[micropage7]

I suppose 6 core APUs are a strong posibility once AMD moves to 28/22nm.

6 cores with better performance per watt ratio

 

[itsakjt]

I want a Phenom III X8 with all new advanced instruction sets, high clock speeds and an awesome IMC.

 

[TheoneandonlyMrK]

I want a Phenom III X8 with all new advanced instruction sets, high clock speeds and an awesome IMC.

hows about you have a think because many went from a phenom II to an FX as i did and i can resolutely say your suggestion is daft ,Fx easily bests any phenom and in scenarios that DO use all supplied cores ie folding@home crunching some games(bf3-4 for eg)etc the FX chips keep up with intels similar easily, i suppose Amd could just call it a phenom 3 to please you though that would'nt make sense


and micropage etc Amd would wisely use that extra space for MOOOAR shaders not two more x86 cores maybe multi core will truley be multi (type) core in this future though too.

 

[enkidu_WM]

You know one thing I have to say is that AMD has been given a ish end of the stick. It has really been the true innovator all along, and each processor has been capable of greater performance gains over their Intel counterparts, it is just an issue in which code is processed. Intel cannot truly attain independent processing with its cores relying heavily on 4 cores with one process and virtualizing it all. AMD is able to perform in a manner befitting individual cores, however programs send information in a format that benefits Intel. 4 people can do one thing faster but 4 people can do four things individually thereby getting more done.