trinity based X4 740 and Radeon 7770 IOMMv2 support ?

[William Gaatjes]

Am i searching at the wrong website or has AMD layed off the website maintenance crew ?

It is for sale in shops but it is nowhere to be found.


EDIT:

I changed the topic title to a more interesting one.
See post 8.

 

[Prey2big]

http://www.cpu-world.com/CPUs/Bulldozer/AMD-Athlon II X4 740.html

 

[William Gaatjes]

I am assuming that the X4 740 technically is the same as the A8 - A5500 but with the GPU section disabled.

I am just looking around how critical the memory specifications are and was looking for a memory validation test.
Should i have a look at the MSI site ?

I am planning for the near future a decent budget pc with this MB : FM2-A75MA-E35 from MSI.


Post Scriptum.
Thank you for the website.

 

[William Gaatjes]

The X4 740 is remarkable similar in specification when compared to the A8 A5500.

http://www.amd.com/us/products/desk...s/Pages/a-series-model-number-comparison.aspx

 

[inf64]

These are Trinity chips with disabled iGPU. They are very cheap,top unlocked model is around 65-68 euros. FM2 socket of course and one needs a dedicated card to run the system,naturally .

 

[Insert_Nickname]

The X4 740 is remarkable similar in specification when comapred to the A8 A5500.

http://www.amd.com/us/products/desk...s/Pages/a-series-model-number-comparison.aspx

If you're serious about getting a 740, I'd highly recommend getting the 750K instead. It should not be that much more expensive. Its fully unlocked and if you do a moderate overclock on it its easy to get it to 5800K performance...

http://www.anandtech.com/bench/Product/675?vs=677

I have even seen people get it up to 4.5GHz with proper cooling...

my 2c...

 

[William Gaatjes]

If you're serious about getting a 740, I'd highly recommend getting the 750K instead. It should not be that much more expensive. Its fully unlocked and if you do a moderate overclock on it its easy to get it to 5800K performance...

http://www.anandtech.com/bench/Product/675?vs=677

I have even seen people get it up to 4.5GHz with proper cooling...

my 2c...

To be honest, i am not going to overclock. I want the system to be as reliable as possible for a non ecc system. I want the system as energy efficient as possible within budget boundaries.
At the moment i have created a list for a complete system including OS for a little more then 500 euro's.

It will have a dedicated card by need, otherwise i would have opted for an integrated GPU. Saving money. I might still decide to go for an integrated GPU for now and then later add a dedicated card.

 

[William Gaatjes]

I was reading this article about piledriver (the core inside the trinity chips)
and i found this to be interesting.

http://www.anandtech.com/show/5831/amd-trinity-review-a10-4600m-a-new-hope/2

IOMMU v2 is also supported by Trinity, giving supported discrete GPUs (e.g. Tahiti) access to the CPU's virtual memory. In Llano, you used to take data from disk, copy it to memory, then copy it from the CPU's address space to pinned memory that's accessible by the GPU, then the GPU gets it and brings it into its frame buffer. By having access to the CPU's virtual address space now the data goes from disk, to memory, then directly to the GPU's memory—you skip that intermediate mem to mem copy. Eventually we'll get to the point where there's truly one unified address space, but steps like these are what will get us there.

I assume support is needed from this by the OS and directx and opengl ?
Only the Tahiti gpu is mentioned.
Will the 7700 series (Cape Verde XT) also benefit from this ?
Thus if i would assemble a system based an AMD X 740 and an radeon 7770, the system can benefit from this removed "mem to mem copying " feature in graphics intensive programs ?
Has anybody an idea how much speed gain this would give in real life applications?

http://en.wikipedia.org/wiki/Southern_Islands_(GPU_family)#Radeon_HD_7700

Southern Islands architecture had the code name "Graphics Core Next" (GCN).[3]
Support for x86 addressing with unified address space for CPU and GPU.
64-bit addressing[citation needed]
Support for PCI-E 3.0[4]
GPU sends interrupts to CPU on various events (such as page faults).
Usage of RISC SIMD instructions for GPGPU instead of VLIW MIMD (Which was only one option in previous AMD GPU-architectures).
Support for Partially Resident Textures,[5] which enable virtual memory support through DirectX and OpenGL extensions.
PowerTune support, which dynamically adjusts performance to stay within a specific TDP.
Usage of Liquid-chamber cooling technology over Vapor chamber.[6]

These changes could lead to better utilization of the GPU for compute along with traditional graphics.

 

[ShintaiDK]

To use AMD-Vi(IOMMU)/VT-D you need chipset, CPU and BIOS support and virtualization software that supports it.

The unified address space by the GPU got nothing to do with virtualization. Its essentially only about GPGPU compute with HSA. In terms of DX and OpenGL it makes no difference.

 

[William Gaatjes]

To use AMD-Vi(IOMMU)/VT-D you need chipset, CPU and BIOS support and virtualization software that supports it.

The unified address space by the GPU got nothing to do with virtualization. Its essentially only about GPGPU compute with HSA. In terms of DX and OpenGL it makes no difference.

Ah, thank you. Would such a feature not be handy for directx and openGL with respect to texture loading or is there already an existing optimized technique to get data into gpu memory directly from mass storage such as DMA with MMU ? I am just guessing here.

 

[ShintaiDK]

Ah, thank you. Would such a feature not be handy for directx and openGL with respect to texture loading or is there already an existing optimized technique to get data into gpu memory directly from mass storage such as DMA with MMU ? I am just guessing here.

There would be no benefit compared to now.

 

[William Gaatjes]

There would be no benefit compared to now.

I am the curious kind who likes to know about the secrets under the hood.
Thus i ask thee : Why ?

 

[ShintaiDK]

I am the curious kind who likes to know about the secrets under the hood.
Thus i ask thee : Why ?

Because none of the things you list are related to loading textures.

First one is about I/O virtualization and would, if anything cause a slowdown. The second part is about GPU sharing address space with the CPU for usage in specific compute.

 

[pablo87]

I would have bought this a few months ago instead of 2500k, a lot less money and can still OC, plus its a current platform FFS. Lots of buyers don't need the iGPU, plus it gives more headroom for the CPU.

Now just need to make its own mask instead of a disabled part, integrate the FCH in the next generation, add another "30.9mm^2" module and...make another payment to GF.