[FW - Chiphell] FM2 CPU not compatible with Socket FM1 board??

[imamage]

http://www.chiphell.com/thread-307338-1-1.html
From this post , poster stated some info about the FM2 APU platform


Image on the left is Socket FM1 CPU(Llano), the right pic is Socket FM2 CPU (Trinity)
FM1 CPU got 905 PIN ， FM2 got 904 PIN , they have different pin count

The next two pic shows
FM2 CPU with Socket FM1
It won't fit into Socket FM1 due to incomaptibility. The pin blocked the insertion...

The last series of pic shows the "Engineering spec"
which shows the clock speed , GPU clock , and Max TurboCore clock

Trinity "Top end model" = Quad Core , 3.8GHz , TurboCore up to 4.1GHz , GPU Clock at 709MHz

[Andi64]

He also said that despite the high frequency the performance is near Llano.

So I guess IPC will suck as Bulldozer, maybe a little better, but without the L3 cache it will be near BD. Just a guess.

The top model will be a 2M/4C CPU? I don't get how it could best Llano, that has 4 cores, and much better IPC, on CPU intensive tasks.

[DedEmbryonicCe1]

The slides indicate they changed the power delivery section to allow for Turbo Core on the GPU. It's not the missing pin that is the problem there, they move around the pins sliding one group (upper left) of blocked off pins further from the center, and the two groups on the right closer to the center. With the rumors about a FM2 socket before Llano even launched I expect they already foresaw this and couldn't get it finished in time for Llano. Nobody should be too surprised that FM1 had such a short life.

[sniper_sung]

AMD fans will be disappointed and hurt.

[Oliverda]

He also said that despite the high frequency the performance is near Llano.

So I guess IPC will suck as Bulldozer, maybe a little better, but without the L3 cache it will be near BD. Just a guess.

The top model will be a 2M/4C CPU? I don't get how it could best Llano, that has 4 cores, and much better IPC, on CPU intensive tasks.

One of the AMD Linux engineering systems for Trinity is running nicely even on Ubuntu 11.04 with the Linux 2.6.38 kernel. The CPU string is AMD Eng Sample 2M252057C4450_32/25/16_9900_609 and its graphics are the Trinity Devastator Mobile with 512MB of video memory and an AMD Pumori motherboard. The PCI ID on the Trinity Devastator appears to be 0x9900. This Trinity APU is quad-core and running at 2.50GHz. The current quad-core Llano offerings are clocked at 2.6GHz (A6-3650) and 2.9GHz (A8-3850), while this Trinity part is clocked slower, it's numbers are nice compared to my A8-3850 Linux system.

http://forums.overclockers.com.au/sh...9&postcount=51

[haylui]

AMD fans will be disappointed and hurt.

already no more feeling...lolz

[FlanK3r]

and Trinity is CPU without L3 (up to 2 modules). Here saing, LLano performance is simillary as Trinity performance at 3800 MHz= about 3.45 points in R11.5. FX 4100 at 3600 has "only" 3 points and + L3 cache

[Hornet331]

http://forums.overclockers.com.au/sh...9&postcount=51

I like how suddenly anything windows related is basically irrelevant for AMD and everything is on linux?

I don't like the hyping again... ppl just got burned hard and now they gona push the linux joker... its all nice and dandy if you use linux... but in the consumer market windows dominates, if your performance sucks there no linux benchmark will save you. Especial in the notebook market where trinity is supposed to make its biggest impact...

[Oliverda]

I like how suddenly anything windows related is basically irrelevant for AMD and everything is on linux?

I don't like the hyping again... ppl just got burned hard and now they gona push the linux joker... its all nice and dandy if you use linux... but in the consumer market windows dominates, if your performance sucks there no linux benchmark will save you. Especial in the notebook market where trinity is supposed to make its biggest impact...

Nobody says that anything windows related is irrelevant. Currently we don't have any windows based results. That's all.

[imamage]

Is that possible FM1 CPU will fit to Socket FM2 board by any chance?

[FlanK3r]

no....because Llano has think 905 pins, but socket for Trinity 904 think. But I dont know at 100%. Maybe Trinity can work with Fm1 boards after BIOS update

[Tao~]

I will be surprised if this is true. How is 3.8 Ghz 2-Module Bulldozer almost equivalent to 2.9 Ghz 4 Core Llano .

EDIT: OK. Not Bulldozer, Piledriver. Hmm.. if this is really true, I am hopeful the next FX series might be competitive.

[Sparky]

FM2 CPU not fitting FM1 socket. Gee, I wonder why. After all, the socket name is different

[Olivon]

no....because Llano has think 905 pins, but socket for Trinity 904 think. But I dont know at 100%. Maybe Trinity can work with Fm1 boards after BIOS update

Physically, a FM2 processor can't fit in FM1 socket

http://www.xtremesystems.org/forums/...=1#post4988818

[Kuroimaho]

Ooh, the irony, yesterday in the haswell thread everybody was surprised that a new arch will needs a new socket, today it's AMD's turn.
I hope the bashing continues.

[3NZ0]

Ooh, the irony, yesterday in the haswell thread everybody was surprised that a new arch will needs a new socket, today it's AMD's turn.
I hope the bashing continues.

As AMD are the 'underdog' we are supposed to feel sorry for them and gloss over when they change sockets while attacking intel for the same thing.

[PatRaceTin]

I see intel socket 1050 for year 2013

[tifosi]

I would like to politely remind fans of Int-El... to sod off

EDIT: Forgot that Int-El fans rather choose not to read... :P it says there in bold, 2.5 Ghz Trinity does well compared to Llano at 2.6 Ghz... don't know what to say...

Also, I think in the next chipset USB3 is native... and may be more SATA3 ports, like 1090?

[iddqd]

As AMD are the 'underdog' we are supposed to feel sorry for them and gloss over when they change sockets while attacking intel for the same thing.

AMD were the original socket-changers.

Does nobody remember people getting all butthurt about the whole s754 -> s939 -> AM2 thing? Some people have pretty short memories.

Anyway, it's all because of the integrated memory controllers. Every time you make changes to that, you'll need a new socket; mostly to prevent people from sticking an incompatible CPU into a board where it won't work (eg: s1156 <-!-> s1155). There's other reasons, obviously, but that one makes you change them more frequently.

[FlanK3r]

Olivon: right, I forgot watch it at photos

[breakfromyou]

Admitting i'm an AMD fanboy, i'm extremely disappointed in AMD. If they somehow manage to screw Llano up by pairing the next gen up with a worse CPU core, i'm done for good. No company making as many mistakes as they have should still have business.

Maybe i'm just tired of seeing llano come out ahead in some benchmarks when put against a much higher clocked Bulldozer. Llano does NOT need a bulldozer core. AMD needs a new core that doesn't suck so badly.

It's almost like Prescott, except that they're not finished ruining their entire product line yet. At least Intel had the Pentium M on the mobile side of things. What's AMD going to do? Release a 1 GHz quad core based off this new bull architecture for laptops? Bring IPC up a lot, at least to the point where an 8 core can easily exceed your LAST GENERATION 6 core chips, or bring power consumption way down. What the hell is happening?! This is the time for extremely low power consumption and high performance. ARM version of Windows 8 as well as x86. If Intel is about to pull off a very power friendly x86 chip that might be tablet-worthy, and AMD is planning on ruining the one decent thing they have going for them (aside from the outdated Phenom II chips), nobody is going to be happy. Competition is absolutely necessary, and AMD isn't providing.

[imamage]

Admitting i'm an AMD fanboy, i'm extremely disappointed in AMD. If they somehow manage to screw Llano up by pairing the next gen up with a worse CPU core, i'm done for good. No company making as many mistakes as they have should still have business.

Maybe i'm just tired of seeing llano come out ahead in some benchmarks when put against a much higher clocked Bulldozer. Llano does NOT need a bulldozer core. AMD needs a new core that doesn't suck so badly.
It's almost like Prescott, except that they're not finished ruining their entire product line yet. At least Intel had the Pentium M on the mobile side of things. What's AMD going to do? Release a 1 GHz quad core based off this new bull architecture for laptops? Bring IPC up a lot, at least to the point where an 8 core can easily exceed your LAST GENERATION 6 core chips, or bring power consumption way down. What the hell is happening?! This is the time for extremely low power consumption and high performance. ARM version of Windows 8 as well as x86. If Intel is about to pull off a very power friendly x86 chip that might be tablet-worthy, and AMD is planning on ruining the one decent thing they have going for them (aside from the outdated Phenom II chips), nobody is going to be happy. Competition is absolutely necessary, and AMD isn't providing.

Honestly I think maybe AMD just make Quad Core Bobcat , clock it up 3GHz and it won't be as bad as Bulldozer core
Bobcat works quite nice BTW

[imamage]

By the way I am AMD fanboy but seeing AMD getting the wrong path isn't going to help them to "save" market share

For Llano (Socket FM1) , I do own an Zotac A75-ITX setup pair with A4-3400, and I am quite happy with its performance
One of my friend (qcmadness) recently grabbed an A8-3800 from Taobao *(Online shopping in China , similar with ebay)
It cost almost the same as A8-3850 but the system setup use almost the same idle + full load wattage ,
If AMD tweak the default VCore lower it will be quite a winner for HTPC / All-in-One for OEM

The sad part is GF have production yield issue with 32nm and they can't produce it in higher volume
I don't mind if they switch the Socket if AMD manage to provide higher performance parts (CPU + GPU) in FM2 , but with "Faildozer" , I don't see the point (Same goes for AM3+ )

Intel fanboy complaint about LGA775 when they changed the VRM version requirement for Core 2 , LGA1156 -> LGA1155 for Sandy Bridge
But Both changes do serve a purpose, the processor performance increase , not decreasing !

[EniGmA1987]

He also said that despite the high frequency the performance is near Llano.

Isnt that because Trinity is based off the Bulldozer core? I thought I had read that. Havent kept up too much on this though. So that would be why performance is around the same with the large clock advantage

[Manicdan]

i think that PD will be fine for fusion since people who get those chips typically dont run all 4 cores fully loaded 24/7, so if single threaded perf is better with Trinity vs Llano then it can probably do well with gaming. 4.1Ghz sounds promising in that regard, however it needs a better task scheduler to see a solid 5% gain on top of that.

now what i want to see the die shot, and get a nice view how how much extra space they threw in there compared to BD, where the cores (minus L3 and extra stuff) took up only half the space. 4 Llano cores might be larger than the 2 Trinity modules, which shows the architecture is not all that horrible.