which processor

i want to build a computer and its main purpose would be blender but can’t decide which processor to get. i have been looking at these 2 amd 1055t or the amd 965 is more cores better or higher core clock speed? or does the # of cores not matter and is just the total speed of all cores combined that matters?

6 * 2.8 = 16.8 Ghz

4 * 3.4 = 13.6 Ghz

So clock and cores are important. The end result can be a good indicator.

In addition AMD cannot do hyperthreading, which means when you have
a long thread blocking one CORE you only have 3 left to do new render threads.
If you have a 6 core system you have 5 left to work.

I would go with the x6 because for pure 3d performance I think that will give
you more render speed.

so it does not matter if the cores are slower?

Slower core will mean slower performance for anything that is not multi threaded.

I’m looking for a new machine too!

the render engine is multithreaded which means it can use all your cores
the speed you can expect is related to the multiple of the number of cores and the clock speed as cekuhnen demonstrated
the 6 core cpu should be roughly 13% faster at rendering
baking is not multithreaded, it would be faster with the 4 core cpu (ignoring openmp)

The 1090T (6 @ 3.2Ghz) isn’t that much more $… why not pop for that?

because it is 100$ more and i just dont have enough for that right now

wait and save your money
how long can it take to save $100?
surely you have a bad habit you can cut out (if you don’t I’ll loan you one of mine;))
better to get what you want than spend years wishing you had it

question is if 0.6 Ghz is so much in speed.

Also even when some tools such as Texture Bake is not multi-threaded
those tasks are not those you spend most of the time on.

Rendering will require the highest amount of any CPU intense task.

Thus it makes more sense to go with a 6 core because rendering will be
faster then a 4 core. Here in a task which requires the most time you
will overall have the highest amount of time saved for the job.

If you go with a 4 core the single core jobs are faster but overall you
render jobs will take longer.

And in the end those render jobs are the big time suckers.

remember, it’s 0.6 Ghz X 6 = 3.6 Ghz
whole point of this conversation


I hasten to point out that the amount and type of memory you get will affect performance as much as a few gigahertz of CPU clock speed. Don’t forget about the motherboard it’s all running on too … some have faster data throughput, and other features, that can affect performance quite a bit too.
In computer building, the key is always matching components for the most efficiency while still using the components that will give you the most performance … a super-hot CPU is no good if the data-busses suck … Tom’s Hardware I’ve found can be VERY helpful in this regard.
good luck!! I too am saving up to build my next machine … i’m amazed at how much dual-CPU motherboards cost these days :stuck_out_tongue_winking_eye: … guess I don’t need to go that route any more tho. (I was hoping to build a dual 6-core machine with 6 - 8 Gigs of RAM and dual graphics cards and a 64bit OS … $$$$$$ a bit prohibitive for that yet =P )

We still have a computer about 5 years old and it still does blender excellently. But for rendering - speed should be your priority.


I was referring to single core tasks and for that it is only 0.6 Ghz difference.

Did you read what I wrote?

BTW Chippy AMD also has Turbo Boost, not as well implemented as Intel but this tool will overclock the single core being used when the other 5 cores are not used thus increasing your GhZ for that core and crunching through the job faster.

wow thanks for all the help. looks like i will go for a 6 core, and don’t worry there will be plenty of ram the mother board i am curently looking at is the ASUS M4A89GTD PROis this motherboard good/badd, thoughts?

no problem you are welcome

I am very sure the AMD six core will serve you well.

If you want to go via Intel

you can also use this guide:

The CPU and motherboard are price wise similar.

Tests showed that the AMD x6 is equal to the Intel i5 with 4 cores.

Actually overall (games/rendering/multuimedia) the X6-1055T is about 10% faster than the i5-750. However.

As I haven´t build an AMD system in quite a while now I can´t tell about AMD OC capabilities, but I can tell you:
The Ci5-750 references with 2.67GHz.
The last one I put hands on OC´d to 3.62GHz without any trouble and purrs like a kitten.

Not to forget the i5´s feature to power down 2 cores when he recognizes dual or single threaded applications and clocks up the remaining active cores.

My last AMD system was the Athlon 500Mhz :stuck_out_tongue_winking_eye: since then I’ve gone with Intel + nvidia. But I already bought my mobo (doing chassi mod, needed mobo to make things fit)

otherwise AMD 6core + ATi 5770 + ATi 5770 would have been a nice blender rig I guess. As Luxrenderer , BGE particles are going OpenCL, I think more and more aspects need to be going the ocl avenue.

Then the amount of CPU cores isn’t that much important taking in consideration the wast numbers of streamcores a graphic card has.

that said, my next linux rig will be i7 + nvidia 460 (found a manufactor, zotec where you can get a 460 for 180€ approx)

but dual graphics card is on my roadmap to harness ocl optimized programs.

That Zotac is the 768 MiB version. I would not get that one.
I´d get a 1 GiB version, they are around 220 Euro.
And then not the Zotac, it comes with 675/1350/1800 MhZ 1.00 VGPU
I´d grab a Gainward Golden Sample GLH or a Palit Sonic Platinum.
In difference to the zotac they got custom coolings, not the hairdryer Zotac uses, run with 0.987VGPU and are clocked at 800/1600/2000.
Also for 220 Euro.

Funky sidenote, the GLH in the Gainward is short for “goes like hell” :DD

I think 30-40 Euro is OK for 30% more memory and ~10% more performance.

The GTX460 will be nvidias new cashcow for the midend sector, I am sure of that. :wink:

yeah but I’m doing a chassi mod, and the zotac got nice reviews for being quiet and not getting to hot during use. afaik i know the 768Mb isn’t that big of diff from 1gb, but yeah. hum. ocl can actually benefit from loading big ass textures onto the gpu, so it doesn’t need to push between gpu cpu ram etc. just run gpu+vram when doing transformations.