Friday, January 20th 2012

AMD Vishera Packs Quad-Channel DDR3 IMC, G34 En Route Desktop?

AMD might be a little sore that its "Zambezi" FX processor family based on its much-hyped "Bulldozer" architecture didn't quite meet the performance expectations of a ground-up new CPU architecture, but it doesn't want to take chances and build hype around the architecture that succeeds it. From various sources, some faintly-reliable, we have been hearing that the next-generation of high-performance desktop processors based on "Piledriver" architecture, codenamed "Vishera", will pack five modules or 10 cores, and will be structured essentially like Zambezi, since Piledriver is basically a refinement of Bulldozer architecture. The latest leak comes from the Software Optimization Guide for AMD 15h family (read here), which was picked up by CPU World while most of us were busy with CES.

CPU World compiled most of the features of what it suspected to be AMD referring to its future processors based on the Piledriver architecture, that's "Vishera" (desktop high-performance), "Terramar" (high-density server), and "Sepang" (small-medium business server) parts. The three are not the first chips to be based on Piledriver, AMD has a new mainstream desktop and notebook APU in the works codenamed "Trinity", which is en route for a little later this year. Trinity basically has an identical CPUID instruction-set as Vishera, Terramar, and Sepang, confirming their common lineage compared to today's "Bulldozer" architecture. The most catchy detail is of Vishera featuring 4 DDR3 channels.

The plot thickens where "HyperTransport Assist feature" is listed as being supported on Vishera. HT Assist is a feature found on AMD's enterprise socket G34 processors, which facilitates better inter-die communication between the two dies of a typical socket G34 Opteron processor. The G34 (LGA1972) package is a multi-chip module of two quad-core, six-core, or four-module dies, which combined have four DDR3 memory channels, and a number of HyperTransport links to communicate with neighbouring sockets and the system's chipset. Could this be the first indication that AMD wants to take on Intel LGA2011 HEDT (high-end desktop) using Vishera chips in the G34 package? It will be a while before we find out.

Apart from using common silicon between client and enterprise platforms, AMD does have a history of colliding the two.Source: CPU World
Add your own comment

229 Comments on AMD Vishera Packs Quad-Channel DDR3 IMC, G34 En Route Desktop?

#1
Super XP
by: Steevo
Any reason you didnt use W1zzards test? Other than it shows BD failing in the same games?
http://www.techpowerup.com/mobile/reviews/AMD/HD_7970_CPU_Scaling/7.html
My point was different setups gain you different results. Both Anandtech & TPU are legit in there own right. There are several website reviews that show BD doing great in gaming. Though I do fully agree AMD has a lot of work in trying to crack open this performance riddle, and hopefully they come close with upcoming Piledriver.
Posted on Reply
#2
Steevo
by: Dent1
So you think the Civilization V reviewed with a the ATI 7970 with the unofficial drivers and no support from W1zzard is more valid than the Civilization V bechmarked from Anandtech with official and mature drivers??

or are you insinuating that Anantech's review is fake?
I'm just saying when a independant review with a card that is more than powerful enough to remove any GPU limitations is reviewed by an indeoendant site that has a hell of a Admin running most of its tests shows results that entirely disagree with one test by one site. Plus most of the BD users who also run Intel know and agree that it doesnt even meet performance of 3 year old Intel offerings in software availabe now, I would put my trust in all if their reviews and actual real life experiances. That was the exact reason I bought ny 1100T.
Posted on Reply
#3
Super XP
Well I fully agree real life is where it's at. Real life gaming is the upmost important for me anyway.
Like I said before, gaming benchmarks all depend on your setup. And I fully agree, Bulldozer was not what we all expected, but it still does fine for everyday gaming. And for me personally it was a performance boost to go with a new AMD FX-8120 platform vs. my late PII x4 940 platform. The price was right and that was important for me. There was no way I would go back to Phenom II and I was itching for a upgrade.

http://www.hardwareheaven.com/reviews/1285/pg10/amd-fx-8150-black-edition-8-core-processor-vs-core-i7-2600k-review-deus-ex-human-revolution.html
Posted on Reply
#4
Dent1
by: Steevo
I'm just saying when a independant review with a card that is more than powerful enough to remove any GPU limitations is reviewed by an independant site that has a hell of a Admin running most of its tests shows results that entirely disagree with one test by one site.
Had the W1zzards review been done a couple of months from now I would have agreed with you, because the official driver would have been ready. But the validity of any review is based on its scrutiny, and if there isn't drivers available whom is to say the GPU was removing the CPU limitations as effectively as it should/could have.
Posted on Reply
#6
WhiteLotus
by: WhiteLotus
I want to know what IOMMU is.

Anyone?
by: DigitalUK
its for VMWare etc to access graphics i believe?
No idea. I guess everyone here is to busy trying to bash each other to tell us.
Posted on Reply
#7
Suhidu
by: WhiteLotus
I want to know what IOMMU is.

Anyone?
IOMMU(now called AMD-Vi) is a feature of the Northbridge (Supported by 890FX/990X/990FX). What it can do is allow you to access actual PCI(e) devices directly from a virtual machine, instead of only via the host OS (in which case the hypervisor(e.g., Virtualbox) would have to implement a virtualized version of the actual device, connect that to the VM, and then redirect all I/O of the actual device to/from the virtual device).

Here are some links about it, but most of what I know about IOMMU/AMD-Vi came from Googling and getting back mostly links to Wikipedia, tech news sites, AMD docs(though AMD doesn't release as many docs as Intel does), and forum discussions. (Also, motherboard BIOS manuals are available.)

(I don't know if I read these particular documents back when I looked it up.)Intel's counterpart is called VT-d. Intel makes this feature unavailable on most consumer CPUs/Motherboards(depending on where the northbridge is located). To contrast with AMD, all AM3(+) Athlon II, Phenom II, and FX CPUs support AMD-Vi when paired with an 890FX/990X/990FX board that supports it in BIOS (You can download+read the mobo bios manuals, and search the web for actual user experience).
(Opterons/Xeons (as well as most server boards) also support these features, however, these are less relevant to this discussion.)
Posted on Reply
#8
Super XP
by: TheMailMan78
So does this mean AMD will change their sockets finally?
From what I've read, Socket AM3+ based Piledriver Cores will have Max 8-Core CPU's. But now with all this new info coming out of AMD, things have changed and now they are pushing 10-Core CPU's with Quad-Channel DDR3-1866 (Finally).

Why would AMD want to change it's socket? How would say Socket FM2 benefit over Socket AM3+? especially when AM3+ is serving it's purpose quite well.
Posted on Reply
#9
eidairaman1
by: Super XP
From what I've read, Socket AM3+ based Piledriver Cores will have Max 8-Core CPU's. But now with all this new info coming out of AMD, things have changed and now they are pushing 10-Core CPU's with Quad-Channel DDR3-1866 (Finally).

Why would AMD want to change it's socket? How would say Socket FM2 benefit over Socket AM3+? especially when AM3+ is serving it's purpose quite well.
Its went to become a Platform war, Technically AMD could cut manufacturing costs by going with a Single Socket and Provide a slew of CPUs.

Llano (Hudson 75 Chipset) Supports 1866 so that means that chipset is a lil stronger than AM3 for ram support.
Posted on Reply
#10
seronx
FM2 has an integrated Northbridge and PCI-E 3.0(1x16 that can be done to 4x4 which is basically the speed of 4 x 8 PCI-E 2.0) support supposedly
Posted on Reply
#11
spixel
by: Dent1
Had the W1zzards review been done a couple of months from now I would have agreed with you, because the official driver would have been ready. But the validity of any review is based on its scrutiny, and if there isn't drivers available whom is to say the GPU was removing the CPU limitations as effectively as it should/could have.
Posted on Reply
#13
spixel
by: Dent1
Whats your point. I can't read your mind you have to add some text to communicate your point.
You are doubting the results of wizzards review because of possible driver issues. This confirms wizzards review is correct and anandtechs is wrong.

Looking at the anandtech screenshot it looks like it might be an error as they have posted the results twice with the same resolution.
Posted on Reply
#14
Dent1
by: spixel
You are doubting the results of wizzards review because of possible driver issues. This confirms wizzards review is correct and anandtechs is wrong.
It's not that I doubt W1zzards review, I'm just saying beta drivers or unofficial drivers could have distorted 7970s true performance regardless of which CPU its running on.

For all we know in W1zzards review Sandybridge might have been held back slightly due to lack of a proper driver.


by: spixel


Looking at the anandtech screenshot it looks like it might be an error as they have posted the results twice with the same resolution
Maybe.

My feeling is with Anantech's review they used a Gulftown 6-core 990x with 12 threads. Maybe the software (Civilizations) wasnt coded to handle it, so the results were poor? But it is a strange result, personally I think its a glitch and the test should of been redone.
Posted on Reply
#15
spixel
by: Dent1
It's not that I doubt W1zzards review, I'm just saying beta drivers or unofficial drivers could have distorted 7970s true performance regardless of which CPU its running on.

For all we know in W1zzards review Sandybridge might have been held back slightly due to lack of a proper driver.




Maybe.
Your a very skeptical person... :)

In response to your edit... neither of that relates to the Bulldozers performance though, there is no way it should have reached 95 fps average so yes, its definitely an error in their benchmarks, hopefully not deliberately lol. Civilization is a very cpu demanding game but it seems having more cores does not help much, I saw in another review the i3 2120 having very similar fps to the i5 2500k however overclocking the cpu gives a large increase in fps.
Posted on Reply
#16
alexsubri
The data cache L1 should increase from 16 Kb to 32 Kb per core in the Piledriver. This would increase single thread performance. The prefetch has to be bigger otherwise the CPU is like a beast with very tiny claws. If there was a way to partner the cores in pairs so that one of them 'turns off' and boosts the other by allocating it's 64 Kb (not 32 Kb), this would result in 128 Kb (64/64 data instruction) and have new Turbo technology on top of that with reduced latency and quad Ram support with options to downgrade to dual channel via bios would make the Piledriver a winner. Let's hope AMD listens to consumer feedback because I am ready to jump ship.
Posted on Reply
#17
Super XP
by: Dent1
WOW.

I can't believe how badly the i7 990x got demolished, the FX 8150 performed twice as fast in Civalization V - is that game multi threaded or something?
I am bringing this up again, it depends on the hardware used. Both TPU and Anatech used different setups and updates, bioses, mobo's etc.

Bulldozer is a great gaming CPU, may not be the best but it still stands firmly with the rest. I have high hopes for this Quad-Channel Piledriver CPU.
Posted on Reply
#18
Patriot
by: xaira
how exactly do they plan to fit a g34 socket and 4 channels of memory on a single atx board?
I don't know...wait they have and it came out years ago...
:banghead: Seriously google is not that hard to use...

SUPERMICRO MBD-H8SGL-F-O ATX Server Motherboard So...


G34 actually has quite a bit of headroom...almost a shame they didn't go that route...
I have some ES chips that if they were only octos could probably be clocked much higher than the 3ghz they are now... (magnycours)...

by: Super XP
From what I've read, Socket AM3+ based Piledriver Cores will have Max 8-Core CPU's. But now with all this new info coming out of AMD, things have changed and now they are pushing 10-Core CPU's with Quad-Channel DDR3-1866 (Finally).
No....just no... lol
Server != Desktop
Server has 8 and 16 cores now and will have 10 and 20 cores then...
Server Already has quad channel... and I think desktop could use it...

There are 2 server sockets....
C32 which is a 2p almost copy of desktop with dual channel ram...
It has 4 and 8 cores now....
G34 is 2 of the c32 dies on one package hence 8 and 16...

From the looks of it there will be 2 new server sockets next round.
Which means AM3+ may or may not get 10 cores...

Its time for an LGA desktop socket with quad channel ram...

I honestly think they should combine C32s replacement with desktop for next gen.

by: Super XP

Why would AMD want to change it's socket? How would say Socket FM2 benefit over Socket AM3+? especially when AM3+ is serving it's purpose quite well.
Because its old and doesnt have enough bandwidth....
Fm2 is different vastly and is meant for APUs ... a cheaper socket.
Posted on Reply
#19
Super XP
Another example where the FX-8150 stands its ground in gaming. Once again for Price/Performance, they make good CPU's.

Posted on Reply
#20
seronx
G2012 -> Octo-channel vs G34 -> Quad-channel
C2012 -> Quad-channel vs C32 -> Dual-Channel
FM2 -> Dual-channel vs No Change

Vishera AM3+
Komodo/Trinity FM2
Sepang C2012
Terramar G2012
Posted on Reply
#21
Patriot
by: seronx
G2012 -> Octo-channel vs G34 -> Quad-channel
C2012 -> Quad-channel vs C32 -> Dual-Channel
FM2 -> Dual-channel vs No Change

Vishera AM3+
Komodo/Trinity FM2
Sepang C2012
Terramar G2012
Sounds like the relationship between the sockets will stay...
G34 is 2 dies of C32 2x dual channel = quad
G2012 being 2 dies of C2012

Hmmm... looks like Vishera gets stuck with dual channel
Posted on Reply
#22
Super XP
Haven't you guys read the 1st post?
Vishera gets Quad-Channel, which is one of the reasons for this thread. :D

Posted on Reply
#24
Patriot
by: Super XP
Haven't you guys read the 1st post?
Vishera gets Quad-Channel, which is one of the reasons for this thread. :D

http://www.techpowerup.com/img/12-01-20/186a.jpg
Yes and there are errors in it....and it clashes with the server roadmaps...


per first column IL has 4 channels not 2
per third column Terramar has 8 channels not 4

I don't think is accurate in the least...
Posted on Reply
#25
Super XP
AMD is making changes. I would take any roadmap with a grain of salt at this moment.
I can easily see Quad-Channel memory for Piledriver Desktop CPU's. Bulldozer was suppose to be Quad, but AMD made last minute changes.

In order for AMD to further better compete, Piledriver may be fully based on there server CPU's but with AM3+ and/or FM2.
Posted on Reply
Add your own comment