Thoughts on Xeon vs i7?

[Notiusweb]

Anyone have any opinions on Xeon's higher cache abilities vs i7's overclock-abilities when it comes to Octane? (Interactions with GPU, scene compilation, plugin workings within other applications, or anything else...).
Thanks and regards!

[rappet]

Here is more to read about it: viewtopic.php?f=9&t=56874&p=294123#p294123

[smicha]

Notiusweb,

I'll ask a guy with new xeons and 7 gpus to join the thread and he may give some test results for xeons (2x 2640v4).

[KonstantinosD]

I am that guy that now owns this amazing system build by smicha, and I will only focus on the Xeons now.(2 x 2640 v4)
Before that let me say that I also own a system with an i7 4930K OC'ed a little to 4.3Ghz.
Let me say this, If you can go with Xeons, do it. They are amazing. In my case at the beginning I thought I would sacrifice single core performance but I was wrong. It runs like butter through anything and it is so much faster than my i7 it's ridiculous. It runs much cooler also, both on water, the Xeons don't reach more that 38C under full load. The i7 goes much higher than that. So to give you some numbers, the i7 OC'ed at 4.3GHz runs Cinebench 15 at 1074 and 144 for single core. Now the Xeon runs the CB15 at 2691 points Multithreaded and Single at 138! Even in single thread its very fast and at 3.4 GHz it's almost as fast as the OCed i7 at 4.3GHz! That shows that clock for clock the Xeon runs faster, much more efficiently (at 95W) and much cooler.
Everything else also runs supersmoothly .

Hope it helps.

[rappet]

Yes, that helps,.. Thanx a lot for sharing your experience

[rappet]

The first image shows the cpu threads while loading an Octane scene.
The other one shows the CPU core behaviour while generating heavy floorplans in ArchiCAD (my 3d model software)
Both have one core with high workload, but You can see the difference of the other cores/threads have a also pretty much workload in Loading Octane compared to a very small workload in ArchiCAD task.

So for the Octane scene I can imagine the much more cores/threads turn out to be a big gain in total performance,
because a lot more threads are helping out, and therefor the one hard working core has less impact on the total performance,
but for this ArchiCAD task the i7 would be much faster... at least for this task set as an example, becuase it looks like this task is really one core depending, which does not say anything about the overall work performance in ArchiCAD.

Favouring cpu probably All depends on the workflow, and when I thinking about the workflow with ArchiCAD using OctaneRender workflow,
my experience is that waiting for big scenes to load into Octane takes most time waiting and counting sheeps or get me a cup of coffee
Probably give in a little in cpu performance for some tasks (like loading the heavy floorplan in my example) Might be worth in gaining a lot more by loading the scenes a lot faster.

Just my thougths to get 'our' thoughts straightended out

KonstantinosD, if it not too much to ask, it would be great if you could test the difference between the dual xeon and the i7 in loading a heavy and complex Octane scene

Cheers,

image.jpg

image.jpeg

[smicha]

Here are some screenshots - thanks to Konstantions - of how well Xeons 2640v4 at 2.4Ghz perform in playing 8K content vs my old i7 4 core 4.5Gzh.

My cpu at 8K indicates over 80% usage and is not able to play the video -very choppy, voice is disappearing.. completely not usable (4K is fine at 40% of CPU usage). On xeons 8K is super smooth at ..17% usage

Feel free to post your impressions on your machine on 8K video.

https://youtu.be/QPdWJeybMo8

[Tutor]

What one is producing and how he/she produces it matters much in determining whether to go with i7 or Xeon.

It's been my experience that the fastest i7s will compile a scene faster than the fastest Xeon with all other system components being equal in ability. But until rendering is realtime, that advantage isn't that significant when rendering multi-frame projects. As to scene interactivity and tweaking while rendering is occurring, I prefer and rely on EVGA's GT X40s (mine have been and are so far the ones where 6=X) and very rarely do those GPUs participate in rendering. However, because I mainly produce animated training videos with thousands and thousands of frames, I've been steady moving all of my GPU assisted 3d rendering jobs to systems with two or four Xeons. Generally, systems with more than one CPU can support more GPUs. The more GPUs at work, the faster the animation can be finished. Thus, my hybrid 3d rendering projects (simultaneous CPU and GPU rendering w/TheaRender) and my solely GPU based rendering projects (w/Octane, Furryball and Redshift 3d) fare better on systems with lots of CPU cores and more than one CPU. My production of videos isn't, however, just all about creating animation. There's video rendering and audio processing of voice and music creations (both processes with lots of plugins) at work. These tasks usually benefit more from Xeons. Also, frequently some of that video and audio work has to be done while 3d rendering is occurring on the same system; so more CPUs running in a system are an advantage here also. Xeon's higher cache and core counts are also indispensable for what I do as frame dimensions and resolution increase. In sum, the nature of what one produces with their software tools and their workflows affect whether fewer, faster clocked cores (i7) vs. a lot more somewhat slower clocked cores (Xeon) fills the bill.

P.S. While new Xeons typically cost a lot more than new i7s, I been completely satisfied with every purchase of used Xeons that I've made through Ebay, not having spent more than $500 (USD) for any one of more than thirty of them. However, the number of i7 systems that I own do outnumber the number of Xeon systems that I own. That's a relic of my initiation because since I began building Xeons systems in 2013, I haven't once considered adding another i7 system to my render farm.

[Notiusweb]

Wow, the discussion here is mind blowing! More!

Hot Head.mp4