cycles speed changes with different motherboard/cpu

Hey, hope im putting this in the right area.
so, random stuff has slowly been dying on my motherboard, like graphics, sound, and sometimes usb. So, I bought a new motherboard and cpu. I have been using a GTX 780 GPU since the graphics were starting to conk out, and am using it on my new board. The original board had a pci 16x slot, and the new board has a pci 2.0 16x slot.
The computer started out as a HP P7-1003w, now the only HP left is the case, CD drive and the RAM (until my new sticks show up in the mail).
Base specs
700W corsair Power supply
8GB RAM (I think 1060MHZ clock)

Benchmark with the “venerable” BMW test

3.2GHZ AMD Athlon 2 x4 (quad core)
CPU render time 33 minutes 55 seconds

GTX 780 on pci 16x (HP motherboard) 5 minutes 11 seconds

The new mother board is a Gigabyte 78lmt-usb3
4.0 GHZ AMD FX-8350 (8 core 16mb cache)
CPU render time 15 minutes 7 seconds

GTX 780 on pci 2.0 16x slot
GPU render time 5 minutes 5 seconds

Swapping from a 3.2 GHZ quad core to a 4.0 GHZ 8 core dropped my render times 18.8 minutes…
Swapping from a pci 16x to a pci 2.0 16x dropped my times 6 seconds…

What does this all mean?
Raising CPU speed by 60% decreased my render time by 55.43%
Changing from pci 16x to pci 2.0 16x decreased render speed by 1.93%

One thing I don’t know is whether my cpu upgrade affected my GPU speed… The only area that I can think of where it may change things is before the actual render when its shuffling and loading materials and what… Mebe someone here is more informed about that then me and can inform.

When my new RAM busses show up I will see if they change my times and post my results.

> Swapping from a pci 16x to a pci 2.0 16x dropped my times 6 seconds…> Changing from pci 16x to pci 2.0 16x decreased render speed by 1.93%

I doubt it, i bet you the 6 seconds was more your cpu improving (which would effect the pre and post processing times for your render) rather then switching between pci1 vs 2

PCI 1 16x speeds operate at 4GB/s (780s have 3 gb so this would fill up in less than a second, if you were maxing out the ram)

PCI 2 16x speeds operate at 8GB/s (this would have the time taken… so if it took .75s before (the maximum ram transferred at max speed), it would take 0.375s… a total of 0.375s saved.)

Ram speeds shouldnt have much of an effect on GPU rendering either.

Ok, cool. I was wondering about that. Thank you for straightening it out.
I don’t expect any change with GPU render with a RAM change, but I am curious about the CPU render times.

Why did you decide to get an AMD FX chip rather than the newer Ryzen?

The FX chips are not true octo-core (there’s only four units that do floating point calculations, something critical for rendering). Even Ryzen 5 would beat them (6 true cores in those). The Ryzen 7 1800X is AMD’s top performer in rendering right now (8 true cores at an IPC that’s maybe 10 percent lower than in an Intel chip).

I went with the FX for a couple of different reasons. The price point was nice, and I primarily use my GPU for rendering, so my CPU just handles the load while I’m modeling and what. There was admittedly some ignorance involved to, I didn’t know that the extra 4 “cores” wouldn’t be quite up to snuff. I didn’t find anything about their performance in blender in my reading. I have been doing a bit more reading since your comment, now that I know what to look for! lol. Guess I got a little snookered by the “specs”! Oh well, live and learn and all that rot! Now I gots to go find how fast a Ryzen does the beamer benchmark!

Been tweaking with a little GPU overclocking via Afterburner. So here is my new GPU render times

Overclocked GTX 780
1228 MHZ core clock
3249MHZ memory clock
74c Temp during render (air cooled)

Render time
3 minutes 51 seconds

So a little bit of tweaking can get decent results

Upgraded RAM. I now have 4 sticks of Corsair Dominator Platinum 8GB rated at 2133MHZ. Total = 32GB RAM.
My motherboard Supports 1333MHZ RAM and the manufacturer states that its cool to clock to 1600MHZ. So that’s what I did.
I checked, and my old HP RAM was running at 1333MHZ.
Well, I satisfied my curiosity, and the verdict is that speeding up my RAM the little amount that I did didn’t do zip for my render times. Mebe if it was DDR4 running at outrageous numbers it may? I don’t know.

Are you sure the RAM you’re using isn’t DDR3 and not DDR4 (I’ve never heard of the latter having a version rated at just 1333 MHZ)?

Plus, if your machine is more than a couple of years old, chances are it doesn’t even support DDR4.

Yes, RAM is DDR3.
The BIOS said it was running at 1333MHZ, Then I just bumped it to 1600. That’s with the new Gigabyte motherboard.
I originally bought this computer in 2010…
My computer originally came with two of these

Then I put two sets of this in.

Well, Haven’t gotten around to updating my thread. But the FX processor is going to be a different computer to use as a farm. I ended up picking up a AMD Ryzen 2700X, I put it on a Gigabyte GA-350M-DS3H board. And paired it with Corsair Dominator platinum DDR4 3200MHZ 32 GB.
I had some issues when I first did it. I went into it knowing that my lovely windows 7 is technically not compatible with the Ryzen. Its true. There is more than USB issues. Had some BSOD issues, Like every time I rendered, or just randomly every five minutes. Had a fresh install of WIN 7. End result is I had to downgrade to WIN 10 and it fixed all of those problems. Enough drivel. Here is my numbers.

GTX 780 4GB
Ryzen 7 2700x
Corsair dominator platinum DDR4 32gb 3200MHZ
Win 10

BMW benchmark Blender 2.79b
GPU 4:58
CPU 4:33

Havent yet clocked my graphics card again yet. will post results when I do.

PS: Your right ACE, I shoulda gone the better route to begin with.