Time to build a new computer.
polycounter lvl 10
My current rig is not doing well as I could hope. In 3D terms UE4 can freeze up in lag sometimes and the FPS is not great at most times. In gaming terms Rocket League at lowest everything at 1680x1050 runs at 64-30 Fps. In general I feel the problem with my current system is the GPU which is the Nvidia 545, most of my system is not that bad (Intel I7 2600, 8GB RAM) however after a lot of thought I would just rather sell my system as one whole thing than upgrade a few bits and have half old half new.
My current thoughts on the new parts is this: http://uk.pcpartpicker.com/p/66Jvxr
CPU
I feel that 4790 is where I want to be in terms of performance without paying too much for too little performance increase.
Motherboard
I just chose something that had good ratings and was compatible. However I have no desire to overclock so maybe more of the expensive motherboards would not be needed for this build.
RAM
I decided that 16GB was as high and as low as I wanted to go, I feel the extra 8GB will help me as there are definitely times where I hit my current RAM cap.
SSD and hard drive
I decided on a 256GB SSD as this is what I feel is the sweet spot for me where I can have an SSD and not worry too much about filling it up while not paying too much. I currently have a 2TB hard drive and while I have only used 600GB of it, I have also been deleting unneeded stuff off it for a long time and I decided the price difference was not too much between 1TB and 2TB.#
GPU
For a while I was split between the 960 and the 970 but I feel that 970 will last me longer before I need to upgrade. I also felt that since I had no plan to overclock a cheaper Zotac 970 would do me just fine.
Case
The 200R looked like a good beginner case that I could get roughly the things I wanted without having to pay too much for them.
Power supply
I am not 100% sure on this and I do admit motherboards and power supplies are my least confident areas in this build.
Feel free to point out any changes I need to make or make suggestions about anything I could change.
My current thoughts on the new parts is this: http://uk.pcpartpicker.com/p/66Jvxr
CPU
I feel that 4790 is where I want to be in terms of performance without paying too much for too little performance increase.
Motherboard
I just chose something that had good ratings and was compatible. However I have no desire to overclock so maybe more of the expensive motherboards would not be needed for this build.
RAM
I decided that 16GB was as high and as low as I wanted to go, I feel the extra 8GB will help me as there are definitely times where I hit my current RAM cap.
SSD and hard drive
I decided on a 256GB SSD as this is what I feel is the sweet spot for me where I can have an SSD and not worry too much about filling it up while not paying too much. I currently have a 2TB hard drive and while I have only used 600GB of it, I have also been deleting unneeded stuff off it for a long time and I decided the price difference was not too much between 1TB and 2TB.#
GPU
For a while I was split between the 960 and the 970 but I feel that 970 will last me longer before I need to upgrade. I also felt that since I had no plan to overclock a cheaper Zotac 970 would do me just fine.
Case
The 200R looked like a good beginner case that I could get roughly the things I wanted without having to pay too much for them.
Power supply
I am not 100% sure on this and I do admit motherboards and power supplies are my least confident areas in this build.
Feel free to point out any changes I need to make or make suggestions about anything I could change.
Replies
It is sadly not a K version nor does my pre-built motherboard support overclocking. Ram slots is also a problem because I currently only have two. I also specified that I don't want to upgrade my current build or overclock. With the motherboard I chose for the new computer It also has 4 RAM slots so if I so decide that 16GB is not enough I can easily upgrade at a later point. Also is that power supply not way to much for the parts since the estimated wattage is around 400w?
It would also be worth investing in a decent cpu cooler as stock coolers are simply shit.
Anyway it's a nice solid build, go for it :poly121:
All of that at a low price.
A 4790k is maybe 15- 30% faster than a i7 2600, depending on the application.
If you really really want to buy an all new system, go for skylake at least. Haswell (4790k) is 2013 tech after all
The importance of "future proofing" a computer build is often wildly exaggerated. Most people will never swap a CPU or make major changes without switching to a different chipset (which means new cpu, mobo and ram, typically), so what is the point?
Also the OP clearly stated he has no intention to OC, so the continued advice for OCing or expensive after market coolers is just... silly.
My current thoughts on what I could choose for the PSU is this:
https://uk.pcpartpicker.com/part/evga-power-supply-120g10650xr
Someone I know has this power supply in their computer so I can at least confirm it does its job.
This makes the full build this:
http://uk.pcpartpicker.com/p/MW3Dgs
It's a nice build with just one nag from me, if your going to use the stock cooler then add a fan to the side of the 200R. If not your blowing heat into the centre of the case. My pref to solve this would be a cooler that blows out the back (toward the back fan) and this will also cool your ram if it's a push & pull cooler.
good luck
This is incorrect. The 4970 is 3.6ghz standard clock, and 4ghz boosted. The 4970K however is 4ghz standard and 4.4ghz boosted. You're losing 400mhz automatically by going with the non-k variant, and only saving yourself maybe $20.
@Repete
I decided that I will build the PC and then check all of the temps and see if I need to order another fan. I'm not totally ignoring your advice but I would rather spend as little money as I can after having to order different parts from different suppliers due to potential problems with one.
Have to disagree here.
Between Rendering
Maya 2016 multithreaded ( and it appears more for the future )
light baking
upcoming Direct X 12 leveraging all those cores.
I would say that the multi threaded revolution is well under way. If not for the average consumer...
Certainly for the Polycount set?
Moore's law has hit a brick wall. The largest gains from here on out may be from more cores and the parallel superiority of powerful GPUs'.
Adopt now and force the market's hand. ( build it and they will come vs if they adopt it then we will build it )
https://udn.epicgames.com/Three/Swarm.html parallel evaluation in Maya 2016
[ame]
I honestly feel that multi-threaded processing is going to lead straight into GPU compute, and that CPUs with lots of threads is just going to be a stepping stone. DX12 is moving a lot of CPU tasks to the GPU, and people are trying to avoid baking more and more.
If you want to do a ton of baking and rendering, build a cheap AMD render farm, it's much more cost effective than putting a $1000 8 core / 16 thread CPU into your main rig.
Like Zac says, the 5820K while having more cores, has slower per clock cores so even at max load of all 6 cores, the 5820K is only about 15% faster than the 4970K, when you consider the cost difference in cpu, DDR4 ram and mobo, you can easily pay up to 2x as much for 15% gain, this is a poor value no matter how you cut it.
While I agree with and understand perfectly that apps in the future will be developed to work more efficiently with multi-core CPUs, this isn't really the case today, and with the two CPUs we're talking about, the differences are small to begin with.
It's very important to keep in mind that the differences we're talking about here are only going to be seen when you're maxing out all cores. AO baking, lightmap baking, offline rendering, etc, these sort of tasks. If you're spending enough time doing these tasks that it dictates your production, you should skip these consumer level CPUs and go straight to high end workstation level Xeons, dual Xeons, build a renderfarm out of AMD cpus like Zac suggests, etc. If you're really spending a significant portion of production on baking/rendering, a high end system like this will quickly pay for itself.
I don't think most game artists spend nearly that much time baking though. I certainly don't. Even when I am baking something like AO that maxes my CPU, I throttle the app so I can continue being productive doing other tasks, so minor differences in speed here are meaningless to me.
Also, just as more applications will become more efficient with multiple cores in the future, as Zac mentions again, they'll make better use of the GPU. GPU baking of texture maps is already much, much faster than CPU baking (check out Knald), so the days of your average artist requiring a beefcake CPU for baking are likely numbered anyway.
Have to agree lots of exciting cores vs price happening with the xeon cores.
I find my experience with multi core machines are directly opposite tho.
where day to day vrt pulling ANY machine can handle without lag. Therefore I perceive none of the speed advantage of single compute performance. Unless I hit a brick wall in computationally oppressive operations I need to deal with. Then I suffer where I can just throw the job out to the renderfarm. In other words...
I only get any advantage out of todays cpu's by throwing cores at the problem. Otherwise there is no problem. ( most of the time, and I will piece meal things to fit that kind of pipline... breaking things up from sculpt to Maya etc... and even then I see CPU and GPU parallel compute as the tech on the horizon to solve the problem )
As the OP is building a machine for Unreal I assume Static Global Illumination and it's concerns.
As I understand DX12...
parallel compute distribution for GPU is made better by finally having better parallel compute distribution that leverages multi-core utilization of today's CPU?
http://www.anandtech.com/show/9543/the-windows-10-review/13 Edit:
yikes I assumed we were comparing to consumer flagship. Haswell-E 8 core ( thought I was in the twilight zone with performance comparisons. But I see everyone was comparing to 6 cores )
After looking at some benches I see how someone would be mightily tiffed with
5930 multi threaded performance compared to 4790k!
But then again this cycle iz a bit of a fluke compared to the regular tic toc...
( where the 4770k was the 4 core performance base and Intel took an unprecedented release cycle and did a one up with the 4790k like the 2600k jump to 2700k )
And even then going cheaper with the 4790k only gives u the multi threaded performance of the 5 year old i7 flagship... ( 990x )
( On a sidenote... I am skipping this round and saving fer summer myself.. pascal with HBM and NVlink just seem like to much of a "potential" game changing wrench to pull the trigger on another $3000 just yet considering the Pascal wafers were rumored to be taped out this june already )
But let's do it anyway. The 5960X is about 42% faster than the 4790K even with double the cores, yet again single core performance of the 4790K is better (about 25% better). So if you're dealing with apps that can't take advantage of more than 4 cores, the 4790 will give better performance.
42% under full core load is significant, but you will pay about 3x the cost if not a little more for that improvement, which again is simply a terrible value unless you spend a large portion of your work day doing offline rendering/baking, in which case you'll probably invest in something even more powerful than the 5960X.