MSI AM3+ 990FX Spotted

Quote:

---

Originally Posted by Manicdan

i noticed the AM3+ socket has the same number of pins missing on each corner, how the hell will we know how to line up the cpu right before pushing it in the socket?!?!?!?!?

---

How's that different from AM3?:confused:

If I were ASUS, MSI would have some explaining to do...

http://img813.imageshack.us/img813/1...usheatsink.jpg

http://img441.imageshack.us/img441/3636/msiheatsink.jpg

Quote:

---

Originally Posted by BeepBeep2

If I were ASUS, MSI would have some explaining to do...

http://img813.imageshack.us/img813/1...usheatsink.jpg

http://img441.imageshack.us/img441/3636/msiheatsink.jpg

---

if you look closely you'll see why AM3+ chips won't fit in AM3.
extra pins come back in a different spot.

Four pin CPU 12V, weird.
edit: nevermind, didn't notice the small cap

Quote:

---

Originally Posted by AlleyViper

Four pin CPU 12V, weird.
edit: nevermind, didn't notice the small cap

---

It is 8 pin but there is a 4 pin dummy plug in it.

@demonkevy i think he was posting the pics because how similar the boards look, really indentical just diff brand and diff colours..

Quote:

---

Originally Posted by Johnny87au

@demonkevy i think he was posting the pics because how similar the boards look, really indentical just diff brand and diff colours..

---

The only thing really the same are the PCIe slot placements. Even the slot count is different. Even the PCI slot count is different. Fan header placement is different, RAM power circuitry placement, and finally, the NB chip on the ASUS is closer to the I/O port by >1" vs the MSI board.

Quote:

---

Originally Posted by jeremyshaw

The only thing really the same are the PCIe slot placements. Even the slot count is different. Even the PCI slot count is different. Fan header placement is different, RAM power circuitry placement, and finally, the NB chip on the ASUS is closer to the I/O port by >1" vs the MSI board.

---

Talking about the heatsink design as well.

MSI never had a heatsink design like this and wouldn't have had a slashed line type design unless ASUS had it first...

the mosfet sink is basically the ASUS one flipped upside down then mirrored with the (I assume) aluminum military class sticker on it

Well, to be honest, I've seen that design on my 785GM-E65 MSI mobo, of course, it doesn't have the slanted lines, but still the unified NB/MOSFET HS. And my Rampage II GENE doesn't have them slanted lines o.0

Yea it looks like BD2 with pcie and/or gpu will need new socket(ddr4 maybe), it doesnt mean that there wont be more am3+ cpus tho.At least one more revision seems likely as it was always the case with amd.
And on the intel side things look even worse.

Those sockets look mighty weird, i see a lot of bent cpus when troglodytes force push cpus into them ,there is still keying in the slot, but not edge wise.

Quote:

---

Originally Posted by RaV666

Yea it looks like BD2 with pcie and/or gpu will need new socket(ddr4 maybe), it doesnt mean that there wont be more am3+ cpus tho.At least one more revision seems likely as it was always the case with amd.
And on the intel side things look even worse.

Those sockets look mighty weird, i see a lot of bent cpus when troglodytes force push cpus into them ,there is still keying in the slot, but not edge wise.

---

Or AMD might be looking to go the Intel route, with two different sockets splitting high end and low end. It would be kinda stupid, since C32 is nothing more than 1207 with DDR3, and G32 was kinda maxed out from Mangy Cours MCM, but you never know, lol....

hm, we will see, but Im thinking, Komodo will be still AM3+ (4-up 10 cores maybe)

Very very interesting but i dont like MSI mobos

http://www.msi.com/product/mb/890FXA-GD65.html - first motherboard which has AM3+ CPU Support (look at mb image). But I'm feeling it is mistake here :)

Quote:

---

Originally Posted by Rsk01

http://www.msi.com/product/mb/890FXA-GD65.html - first motherboard which has AM3+ CPU Support (look at mb image). But I'm feeling it is mistake here :)

---

You are right ;)

Quick look at the socket shows it's AM3 only :p:

any word on SLI capability with this mobo? thats what i really want. also, will it support PCIE rev. 3.0? im looking to build my first monster rig in a few years at the end of the year and i want to make it a little bit future proof. i was going to go the intel route but with my rigs cost already ballooning a bit (blame that on water cooling :D ) i wouldnt mind reducing the cost by going the amd route. i use my desktop for gaming mostly so i think in that respect bulldozer not performing as well as ivy bridge in a few different benchmarks and apps will be of little difference to me

Quote:

---

Originally Posted by FlanK3r

hm, we will see, but Im thinking, Komodo will be still AM3+ (4-up 10 cores maybe)

---

It won't be 10 cores only plac amd will have more than 8 cores is server right now.

Quote:

---

Originally Posted by Mats

1 - This slide shows that Komodo got an integrated GPU (not that I want it tho).
It's not an APU, but given what it's made for I'm not surprised that AMD calls it a CPU.
If you integrate a GPU that doesn't match the CPU performance wise (ie adding the Llano graphics to a 8C BD), you simply wouldn't call it an APU, IMO. I'd like to hear some official words about that, tho.

2 - The server counterparts, Sepang and Terramar, will get new sockets and have integrated PCIe (it doesn't mention any Komodo GPU).
Even the relatively new, lagacy-free C32 and G34 isn't enough even though they're guaranteed to work with first gen Bulldozer.

3 - JF have told us that PCIe can't be integrated without a new socket.

Desktop and server development has been close to each other since the AMP/XP, and I don't think AMD is prepared to separate them more =(higher cost) just to keep AM3+ for desktop.
So even if Komodo doesn't have a GPU, it most likely will have PCIe on die since that's what the server CPU's have.

Nothing is set in stone, and I hope I'm wrong, but this is why I think AM3+ won't work with Komodo.

Sn0wm@n: Have you seen any signs indicating that Komodo will work with AM3+?

---

and did they show am3+ compatibility or a new socket need for komodo??? not yet ... so its still in the air as far as we know ....


so again my first post about the not complaining about the am3+ sockets dead end ...


btw compairing server roadmap with client roadmap doesnt mean anything ... server roadmap is one thing .. and client roadmap is another ..

Komodo is desktop....it will be refresh Zambezi chip (4-8 cores, maybe more)

Hmm...
Why are all the brand's of mobo makers using those lame round power buttons ?

Does anyone know for sure if am3+ has quad chan support?
If it doesn't then the only thing I'm really interested in is the avx support on the cpu it's self (that would come in real handy).
And I'm not seeing any diffrence in the cpu socket either...

So why a new socket?
What in the world do we need another socket for when it's the exact same stuff re-hashed?

You gotta forgive me, I haven't been keeping up on the new cpu rumors..

Oh and one more thing.
We are in the age where we are paying almost $300 per board (yeah well $200...).
But why are the same priced intel board's getting 2x the pwm, 2x the cpu power conenctors and extra pcie power when the amd's get left out and look like mid range ?
Paying that much you'de think we'de have 4 cpu sockets.

dual channel support for desktop ... its been known for a while actually



and why would you need quad channel memory support???

And why would we need faster cpus ? ;).
Same reason... :)
Mem is way to slow on current amd's to put much of a workload through them.

Thanks for the heads up though.
I'm kinda glad I didn't wait it out for this gen.
Not that it isn't any better, it just not that much better except for the avx support I think.

Quote:

---

Originally Posted by NEOAethyr

And why would we need faster cpus ? ;).
Same reason... :)
Mem is way to slow on current amd's to put much of a workload through them.

Thanks for the heads up though.
I'm kinda glad I didn't wait it out for this gen.
Not that it isn't any better, it just not that much better except for the avx support I think.

---

ok .. wich program do you use that requires lots of memory bandwith ....

PCSX2 "might" be one, not 100% sure.
I haven't checked the diffrences in mem speeds with that yet.
But it's a super power hungry app that prefers more then 4ghz quads.

Avisynth I'm pretty sure could use more mem bandwith those.
Certain kinds of filters for both audio and video use one heck of alot of bandwith.
More so then what the stuff in my sig could provide.

I'm on a temp system for now until my mem gets back from rma though.
I haven't checked the 64bit ver of avisynth mt either.
With 64bit generic registers and 64bit mem access it might beable to pull off alot more.
But in 32bit, alot of stuff is just way to much for it in avisynth in terms of realtime filtering.

Other then those 2 apps.
I haven't seen anything come close to using all of my systems power.
So really all is good, it's kind of a golden age in terms of performance.
Just that pcsx2 could use a tiny bit more power, cpu mainly but mem would probably help alot.
Avisynth however I think is the main one that needs more then what we currently have.

The only filter for avisynth I can think off off hand, without re-testing is the sound eq.

The only reason I know is because I was making a new format for video's.
I would pack them, store ratio, in rar's, rename the extention.
Setup the registry for that new extention (scripting using js and batch), to play back in media player classic.
The new format I made for video's I'de download mostly off youtube, multi part videos.
So I could have them all put into a single file.
To piece them together I'de use an avs file, avisynth.
Which worked ok'ish, some of them had miss matches formats for video resolutions and audio types, etc.
So I needed to stretch the video's to a new resolution.
I worked on diffrent filter setups and etc to try to ensure I would get more quality then what was being put in and so it was valid for all my videos.
Well, I ran into a big prob with performance on some of the more fancy filters (higher res, noise, eq's, etc).
So I got the 1st build of the mt avisynth, to buggy.
I got the latest x86 mt build from doom9, worked ok.
Still to slow though.
The very latest are 64bit only.

Lol..
That is actually the only reason I'm looking at converting to the 64bit os.
Because everything I tried in the "past" with win2k3 x64 datacenter really didn't make any diffrence in performance.
But perhaps with all that raw data, the 64bit os might work well for that.

Seeing that intel boards with tripple chan, and now these sandy bridge intel's with dual chan blowing amd out of the water in terms of mem bandwith.
I think it could be very helpful for these extremely large workloads.

It's not like I want to render the original final fantasy movie in real time lol, I just wanna use avisynth to it's full potential.
:)

so your not sure if your programs really need more processing power or more bandwith yet you deffinatly think there is a big need for bandwith ... . why not try to max your memory controler now instead of wanting more that you wont even max out ..