Sunday, November 4th 2012

No New FX Processor From AMD in 2013

AMD's FX "Vishera" socket AM3+ processors are in for a long haul. According to a DonanimHaber report based on a leaked company roadmap slide, the company plans no new processor architecture to succeed it in 2013. The company recently launched its FX "Vishera" line of eight-, six-, and four-core chips just an year following FX "Zambezi," leading analysts to believe the company would launch a new micro-architecture each year to keep up with Intel's "tick-tock" product development strategy.

The roadmap slide, pictured below, shows AMD FX "Vishera" continuing through 2013 as the flagship desktop platform, followed by "Richland" third-generation desktop APU, which combines "Piledriver" CPU components with "Radeon 2.0 cores" (we're guessing those are Graphics CoreNext stream processors), which maintains socket FM2 platform; and low-power "Kabini" APU, which carries the mantle from "Brazos."


Source: DonanimHaber
Add your own comment

58 Comments on No New FX Processor From AMD in 2013

#1
Covert_Death
I for one am fine with this.... i just ordered my 8350 to replace my Pii x4 955 BE and can't wait to get it. it should last me until SR is released (hopefully with ddr4 and PCIe3.0)
Posted on Reply
#2
cdawall
where the hell are my stars
by: Steevo
You are missing the point.

A CPU core is just that, a X86 procesing unit, there are only so few ways to make it work.

The latency is a huge issue every time their inferior banch prediction fails. 5% worse hit rate added to a 10 cycle penalty due to the latency if it is even in L3 and you have your 15-20% slower per clock than Intel.

Bend of the knee my friend, AMD is there and it isn't a good place to be.
You are still assuming it is in the L3. Most major processes end up in the ram anyway at most the instruction set is there. There is next to no performance change per core on AMD between L3 and no L3 (FX series vs APU's). Heck even on Intel there is only a mild change the cacheless pentium and chips and i3 chips per clock.
Posted on Reply
#4
repman244
by: cdawall
I just want to point out it doesn't say there will be no new FX processors just that they will be Vishera based. We will likely see new revisions of Vishera with improved IMC's and improved power consumption/clockspeed.
I think the same, and maybe we see Vishera on 28nm as well. Maybe it's a good thing, they will use Vishera to improve the 28nm process before doing Steamroller.
Posted on Reply
#5
xenocide
I knew this was going to happen. I even called it back in the Piledriver discussion thread--and then was hastily berated by the very pro-AMD users. Oh well, I'd rather AMD took their time and released a killer product than rushed another Bulldozer out the door. Piledriver will be fine by now unless Haswell comes out and is substantially better than people expect.
Posted on Reply
#6
Dent1
by: xenocide
I'd rather AMD took their time and released a killer product than rushed another Bulldozer out the door.
Huh, Bulldozer wasn't rushed! It was delayed, and delayed and delayed again.
Posted on Reply
#7
Live OR Die
there CPUs suck mite as well give up and put all there brain power into there GPUs.
Posted on Reply
#8
Aquinus
Resident Wat-man
by: eidairaman1
L3 cache was made to compensate for the latency of the DDR3 DRAM for data access, so it means more common tasks are stored in the cache.

http://en.wikipedia.org/wiki/CPU_cache
Hahaha, Wikipedia. Cache in general is for this. L3 is shared by all the cores and the bandwidth on the L3 cache for AMD processors is similar to that of its memory speed, just with lower latency, which is fine but if you compare the multi-core efficiency of an AMD processor with and without that L3 cache, you will notice that L3 helps multi-core efficiency more than anything else. Intel does the same exact thing. Not to say it doesn't help other tasks, but there is a reason why L2 is write-back and L3 is write-through on AMD processors IIRC.

by: Live OR Die
there CPUs suck mite as well give up and put all there brain power into there GPUs.
GPUs don't handle linear applications well, they're built to run the same instructions on a huge set of data which the majority of applications in the world do not do or need... yet and I don't think that will change in the near future.
Posted on Reply
#9
ChristTheGreat
I guess we don't need right now faster CPU.. Maybe more optimization would be nice...
Posted on Reply
#10
Melvis
by: Live OR Die
there CPUs suck mite as well give up and put all there brain power into there GPUs.
So every Intel CPU from the 3570K/2600K and down suck also is that right? :rolleyes:
Posted on Reply
#11
sergionography
by: Steevo
You are missing the point.

A CPU core is just that, a X86 procesing unit, there are only so few ways to make it work.

The latency is a huge issue every time their inferior banch prediction fails. 5% worse hit rate added to a 10 cycle penalty due to the latency if it is even in L3 and you have your 15-20% slower per clock than Intel.

Bend of the knee my friend, AMD is there and it isn't a good place to be.
the massive size of the L3 cache masks the latency, not to mention its the L2 cache that is more important not to mention the load queues/L1 especially in amd designs which amd is improving. as for the L2 its larger size kinda masks for the latency bit so at this point that's the most affordable way for amd to work around the problem, and with power gating the cache in steamroller they solve the extra power consumption issue.anyway aslong as they improve the branch predicted the less the latency is going to matter, at this point i think its more important for amd to add some complexity to the modules than just work on the cache, since now u have a module with shared resources that has less resources than a single core sandy bridge. Dedicated decoders in sgeamrollers kinda solve that problem, adding more ipc and better multithread scaling, then That whole its not really 8 core argument might finally be put to rest

I believe the only advantage for amd to improve their cache would be to allow them to use less cache and have more performance rather than wasting all this die area, but as far as performance its gonna be single digit gains in very specific kinds of code so its not worth it for amd right now
Posted on Reply
#12
3870x2
by: Melvis
So every Intel CPU from the 3570K/2600K and down suck also is that right? :rolleyes:
what?
Posted on Reply
#13
Fourstaff
So they are not planning to release higher clocked versions?
Posted on Reply
#14
cdawall
where the hell are my stars
by: Fourstaff
So they are not planning to release higher clocked versions?
They are not planning on releasing steamroller the road map doesn't show individual chips.
Posted on Reply
#15
seronx
Vishera 2.0(Orochi Rev E or Rev C(x)) is what is coming out next year.

Then with Viperfish Rev B coming out early 2014. <-- With AMD returning to the early April-May launch dates.

Vishera 2.0:
FX-8370
FX-8340
FX-6400
FX-4400
New NB 8.0 GT/s
New SB (Hudson or Bolton D4)

Viperfish:
Socket GC36 only(1974+ pins)
Quad-channel DDR4
10 Steamroller Cores
80 PCI-E 3.x Links or 5 HT 4.0 Links(HTX4 GPUs?)
Internal NB and I/O Hub
External SB with Bolton D5?


----
LGA 1974(Reformatted) vs LGA 2011
FX vs i7

PGA 904 vs LGA 1150
A10 vs i7
----
^-2014 worth the wait?
Posted on Reply
#16
Dent1
by: Live OR Die
there CPUs suck mite as well give up and put all there brain power into there GPUs.
There CPU sucks? Look at the Piledriver reviews, AMD actually beats Intel in more tests overall. But apparently a CPU doesnt suck if it excels just in single threaded gaming.

by: seronx
Vishera 2.0(Orochi Rev E or Rev C(x)) is what is coming out next year.

Then with Viperfish Rev B coming out early 2014. <-- With AMD returning to the early April-May launch dates.

Vishera 2.0:
FX-8370
FX-8340
FX-6400
FX-4400
New NB 8.0 GT/s
New SB (Hudson or Bolton D4)
Where did you read that? I Googled it and found nothing!
Posted on Reply
#17
seronx
by: Dent1
Where did you read that? I Googled it and found nothing!
The names I am guessing but Vishera 2.0 will replace Vishera and will give a timeframe for the next CPU line up. The new NB and SB has been in the roadmap since 2007.
Posted on Reply
#18
cadaveca
My name is Dave
Don't quote me on this, but isn't steamroller really just piledriver on 28nm?


which would still be piledriver...


Just sayin'.
Posted on Reply
#19
xenocide
by: cadaveca
Don't quote me on this, but isn't steamroller really just piledriver on 28nm?


which would still be piledriver...


Just sayin'.
From what I remember it was a fine-tuned variation of PD on 28nm--and PD was only a refined version of BD. I know the implication is that we could still technically see new CPU's, but it's more likely revisions will come out next year (higher clocks, very minor tweaks) and they will lineup a Steamroller launch for Spring 2014 to finally be releasing around the same time Intel is, or just before them.
Posted on Reply
#20
cadaveca
My name is Dave
Yeah, I can see the easy potential for 5 GHz clocks, perhaps, if they can get the process ironed out nice, and really, that's all I expect from AMD, myself, but you never know what might happen.
Posted on Reply
#21
eidairaman1
by: cadaveca
Yeah, I can see the easy potential for 5 GHz clocks, perhaps, if they can get the process ironed out nice, and really, that's all I expect from AMD, myself, but you never know what might happen.
for now a 5GHz Chip is plausible.
Posted on Reply
#22
HumanSmoke
by: cadaveca
Yeah, I can see the easy potential for 5 GHz clocks, perhaps, if they can get the process ironed out nice, and really, that's all I expect from AMD, myself, but you never know what might happen.
Ye gods I hope not. I'd personally hope AMD could be able to get closer to Intel's overall spec- keep clocks relatively static while bringing power consumption and die size down. Every enthusiast would dearly love insane clockspeed, but the enthusiast market isn't what will keep AMD afloat. I'd sincerely doubt 5GHz@28nm is going to be any more frugal on power usage than 4GHz@ 32nm...which may mean squat to a lot here, but given that the same four module parts are going 2-up on the same package for server duty, and OEM's have a love for cheapening out on PSU's and cooling- it doesn't sound like a recipe for success.
Posted on Reply
#23
TRWOV
by: 3870x2
what?
He was quoting a guy that said the 8350 sucks. Since the 8350 trades blows with the 3570 so it's assumed that the 3570 also sucks. :laugh:



by: Live OR Die
there CPUs suck mite as well give up and put all there brain power into there GPUs.
by: Melvis
So every Intel CPU from the 3570K/2600K and down suck also is that right? :rolleyes:
Posted on Reply
#24
NeoXF
What about Piledriver 2.0 (or was it Vishera 2.0?)?
Posted on Reply
#25
nt300
by: btarunr
followed by "Richland" third-generation desktop APU, which combines "Piledriver" CPU components with "Radeon 2.0 cores"
Shouldnt this be with Steamroller Cores :confused:
What is AMD doing.
by: cadaveca
Don't quote me on this, but isn't steamroller really just piledriver on 28nm?

which would still be piledriver...

Just sayin'.
Steamroller is suppose to be the major design change and should look nothing like Bulldozer. Most of the shared coomponents in Bulldozer are suppose to be removed for Steamroller and should offer approx about 45% performance improvement.
Posted on Reply
Add your own comment