KSP1 Computer Building/Buying Megathread

[Camacha]

21 minutes ago, Endersmens said:

And this is what I currently use for: Gaming, Recording, Blender rendering. I just wanted a big step up. I don't wanna be the very best, obviously, since I have a budget. I want the very best for my budget, and I want something that will be a lot better than my current setup. Something that doesn't drop the framerate to 15fps when I record and the game is already on a lower graphics setting. Does that make sense?

If at all possible budget wise, I would look at an i5. Those are absolutely great value for money. It would a doable sum more, depending on model choice. Meanwhile, depending on the exact choice you make, you would gain almost 100% performance in rendering and encoding, while games also tend to like an i5 a lot. It would make the system ready for pretty much any task you can throw at it now and in the foreseeable future. Due to Intel's socket strategy, upgrading to another processor will be a pain. With a modern i5 you will be set - processor wise - for the next years. Swapping out a GPU later will be much easier. Looking at what you want to do with your system, it would not be overkill, but allow you to make a leap in performance and work with a fair amount of comfort. If you are prepared to overclock, an i5 could be even faster, though it would require some additional investments.

It would probably be a good idea to run a render based benchmark on your system, like Cinebench. That way, you get a really decent comparison of modern hardware to your current system. Yes, I always say that real world benchmarks are superior and this is an artificial one, but since you do not have access to multiple systems and the workload is in this case fairly close to the real thing, it is a very decent alternative. It will give you a realistic idea of what to expect when rendering.

Quote

When I said you conflicted yourself, you said hardware twice. I bolded it in my quote. You said hardware, as compared to hardware.

That is why I corrected my post and clarified any confusion in the new post

[Alphasus]

4 hours ago, Endersmens said:

@Alphasus I was looking for i3s, but I kept forgetting that I was editing a parts list, so compatibility check was on the whole time. That's why I ended up with that processor. So the second intel build, how much better would that be compared to my previous intel build? I'm mostly wanting this computer to be a rig that I can game and record off of, and possibly stream as well. I will also be doing blender renders, and video editing. I'm not really worried about KSP potential, more of just modern computer intensive games that I can't play right now. 

 

Also I don't have a preference for miniITX, in fact I think I would rather go MicroATX instead, if that's a possibility with those other parts.

Off the top of my head, the sky lake build should be about 25% faster(10% per generation). MiniITX doesn't make it much more expensive. The extra cores on the AMD 880K make it better for streaming than an i3, but most games prefer strong single cores, where the i3 winds. The 1060 should be about as fast as my 970 for gaming, which is pretty quick. It also goes really fast for rendering and did the BMW benchmark(new version) in sub 2 minutes.

Edited July 14, 2016 by Alphasus

[Endersmens]

3 hours ago, Camacha said:

It would probably be a good idea to run a render based benchmark on your system, like Cinebench. That way, you get a really decent comparison of modern hardware to your current system. Yes, I always say that real world benchmarks are superior and this is an artificial one, but since you do not have access to multiple systems and the workload is in this case fairly close to the real thing, it is a very decent alternative. It will give you a realistic idea of what to expect when rendering.

I ran cinabench. Here are the results, I have no idea what they mean or how bad my current system is.

CPU: 287

OpelGL: 17.17fps

Edited July 14, 2016 by Endersmens

[Camacha]

2 hours ago, Endersmens said:

I ran cinabench. Here are the results, I have no idea what they mean or how bad my current system is.

CPU: 287

OpelGL: 17.17fps

The i3-3240 gets a score of 306. That is barely any faster than your quadcore i7 now when it comes to rendering and encoding workloads. Mind you, in single threaded applications it will be a fair bit quicker (so in KSP too), but in properly threaded workloads very little is gained. The i5-6500, however, scores 544 points, which means you double your rendering performance. As a reference: the i7-6700K scores almost double that with a whopping 1060 points, in part due to its high thread speeds and in part due to hyperthreading being ideal for rendering workloads. It is unfortunately proportionally expensive.

This all assumes you ran CineBench R15, which by the looks of it, you did. The number is exactly what you would expect from your current CPU.

[Endersmens]

38 minutes ago, Camacha said:

The i3-3240 gets a score of 306. That is barely any faster than your quadcore i7 now when it comes to rendering and encoding workloads. Mind you, in single threaded applications it will be a fair bit quicker (so in KSP too), but in properly threaded workloads very little is gained. The i5-6500, however, scores 544 points, which means you double your rendering performance. As a reference: the i7-6700K scores almost double that with a whopping 1060 points, in part due to its high thread speeds and in part due to hyperthreading being ideal for rendering workloads. It is unfortunately proportionally expensive.

This all assumes you ran CineBench R15, which by the looks of it, you did. The number is exactly what you would expect from your current CPU.

I've look up scores for the other processors in mind. The i3 6300 (the one in question for me) gets a multicore score of 406, which is only slightly less than the multicore score of a fx-6300, at 415 single core blows it out of the water. And the athlon core is even worse in the mid 300s. Out of the processor options laid before me, the i3 6300 is clearly the best. Not only for single core either, since it's multicore score is right near the fx 6300 and athlon 880k, despite it's core disadvantage. Unfortunately, I will probably not be able to fit an i5 in. I'm pushing price as it is. Goal was $500, I let it slip to $600. I'll see what the options are for i5s but I don't have high hopes.

 

Basically, it's keep the i3 6300, SSD, and gtx 960 4gb, or drop one of the latter for an i5.

Edited July 14, 2016 by Endersmens

[Camacha]

6 minutes ago, Endersmens said:

I've look up scores for the other processors in mind. The i3 6300 (the one in question for me) gets a multicore score of 406, which is only slightly less than the multicore score of a fx-6300, at 415 single core blows it out of the water. And the athlon core is even worse in the mid 300s. Out of the processor options laid before me, the i3 6300 is clearly the best. Not only for single core either, since it's multicore score is right near the fx 6300 and athlon 880k, despite it's core disadvantage. Unfortunately, I will probably not be able to fit an i5 in. I'm pushing price as it is. Goal was $500, I let it slip to $600. I'll see what the options are for i5s but I don't have high hopes.

Upgrading to an i5 would certainly push the computer from office and light gaming to serious gaming and production work. Maybe you can convince your parents that building a skill set with rendering and video editing is a valuable advantage in the job market

That is a fairly serious remark, by the way. I am not sure how old you are, but assuming you are in the first half of your teens, discovering and learning these kinds of skills is absolutely invaluable. You will be much more comfortable with these kinds of tasks and skills later on. I started fiddling with computers and programming when I was in my early teens too, and it has certainly paid dividends in my education and professional life. I litterally could not make my living if I did not start when and where I started then.

[Endersmens]

12 hours ago, Camacha said:

Upgrading to an i5 would certainly push the computer from office and light gaming to serious gaming and production work. Maybe you can convince your parents that building a skill set with rendering and video editing is a valuable advantage in the job market

That is a fairly serious remark, by the way. I am not sure how old you are, but assuming you are in the first half of your teens, discovering and learning these kinds of skills is absolutely invaluable. You will be much more comfortable with these kinds of tasks and skills later on. I started fiddling with computers and programming when I was in my early teens too, and it has certainly paid dividends in my education and professional life. I litterally could not make my living if I did not start when and where I started then.

Did you see what I added to the post you quoted? About what parts I can keep and what parts I would have to ditch?

[cantab]

Movement in GPU prices might make budget for the i5. You could also consider buying second-hand parts but of course there are risks involved. That said the i3-6300 is no slouch.

[Alphasus]

16 hours ago, Camacha said:

The i3-3240 gets a score of 306. That is barely any faster than your quadcore i7 now when it comes to rendering and encoding workloads. Mind you, in single threaded applications it will be a fair bit quicker (so in KSP too), but in properly threaded workloads very little is gained. The i5-6500, however, scores 544 points, which means you double your rendering performance. As a reference: the i7-6700K scores almost double that with a whopping 1060 points, in part due to its high thread speeds and in part due to hyperthreading being ideal for rendering workloads. It is unfortunately proportionally expensive.

This all assumes you ran CineBench R15, which by the looks of it, you did. The number is exactly what you would expect from your current CPU.

He renders in Blender. Blender uses video cards. His CPU doesn't matter for rendering. So the question is, does video editing need an i5? Also, the 6300 has a 403 in CineBench R15, which is a good deal faster than his laptop.

Edited July 14, 2016 by Alphasus

[cantab]

1 hour ago, Alphasus said:

He renders in Blender. Blender uses video cards.

Since that's the case, may as well compare GPU performance:

http://www.videocardbenchmark.net/compare.php?cmp[]=3114&cmp[]=108

A synthetic benchmark so won't be precise, but going from a 330M to a 960 (non-M) is about a fifteen times performance improvement. It's huge.

[Endersmens]

1 hour ago, cantab said:

Since that's the case, may as well compare GPU performance:

http://www.videocardbenchmark.net/compare.php?cmp[]=3114&cmp[]=108

A synthetic benchmark so won't be precise, but going from a 330M to a 960 (non-M) is about a fifteen times performance improvement. It's huge.

How would a 960m compare? Because a new option just opened up. For almost the price of this computer (including the monitor and OS) I could get a laptop with a skylake i7 and a 4gb gtx 960m. Same card I was gonna get, much beefier processor. 

@Alphasus, blender uses CPU or GPU. On my laptop it uses my CPU since it doesn't support my old graphics card. Although I've heard GPU is much better.

Here is that laptop: http://www.bestbuy.com/site/lenovo-y700-15-6-touch-screen-laptop-intel-core-i7-8gb-memory-1tb-hard-drive-black/4474900.p?id=1219751154856&skuId=4474900

Processor/Graphics:
Intel® Core™ i7-6700HQ processor
NVIDIA GeForce GTX 960M graphics
 

[Alphasus]

14 minutes ago, Endersmens said:

How would a 960m compare? Because a new option just opened up. For almost the price of this computer (including the monitor and OS) I could get a laptop with a skylake i7 and a 4gb gtx 960m. Same card I was gonna get, much beefier processor. 

@Alphasus, blender uses CPU or GPU. On my laptop it uses my CPU since it doesn't support my old graphics card. Although I've heard GPU is much better.

Here is that laptop: http://www.bestbuy.com/site/lenovo-y700-15-6-touch-screen-laptop-intel-core-i7-8gb-memory-1tb-hard-drive-black/4474900.p?id=1219751154856&skuId=4474900

Processor/Graphics:
Intel® Core™ i7-6700HQ processor
NVIDIA GeForce GTX 960M graphics
 

6700HQ single threaded is much slower than the i3. It WILL throttle in multi threaded tasks like editing, so performance won't be much better than the i3. A 960M is similar to a 950, which is 3/4 as fast as a 960. The 960M will thermal throttle as well while rendering, seriously slowing down your render times. Trust me, I know a friend with the same laptop that throttles even though he cleans it regularly with compressed air. Take the desktop.

 

I use Blender all the time by the way with a GTX 970. It is really fast, and a 960 is about 60-70% as fast. Expect speeds similar to a 970M or 770 in rendering.

[Endersmens]

Ok, I will look into i5s. What would be the best i5? My range for i5s is small, can only do up to about 200. That includes the 6500, but it almost includes the 2500k, which has lots of good remarks (although slightly older)

What would be the absolute best i5 for under or around $200?

[Endersmens]

Alright, I put together an i5 build with the 6500. Can't do the 2500k, already pushing budget majorly. Here is the build: http://pcpartpicker.com/list/V7MKBP

And it's only $20 more than the previous i3 skylake build: http://pcpartpicker.com/list/fw3dXH

Plus I was able to put DDR4-2400 RAM in.  If you guy's response to this is anything about better than the i3 build or that I couldn't do better with  my budget I'm buying this tomorrow.  

[Camacha]

10 hours ago, Alphasus said:

He renders in Blender. Blender uses video cards. His CPU doesn't matter for rendering. So the question is, does video editing need an i5? Also, the 6300 has a 403 in CineBench R15, which is a good deal faster than his laptop.

I would consider an improvement of just barely 33% not much of an improvement in computer land, to be honest. That just does not seem worth spending all that money on.

Regarding the CPU versus GPU rendering - it might be worth looking into that to see whether it is actually faster and whether the quality and compatibility is also what you would want and expect. GPU rendering is sometimes viewed as the end-all solution, while performance does not always match up with these expectations. Sometimes it does offer a very nice shortcut, though. It all depends on the circumstances.

6 hours ago, Endersmens said:

Alright, I put together an i5 build with the 6500. Can't do the 2500k, already pushing budget majorly. Here is the build: http://pcpartpicker.com/list/V7MKBP

And it's only $20 more than the previous i3 skylake build: http://pcpartpicker.com/list/fw3dXH

The 2500K is not really a chip I would buy any more. I have had one for a while and absolutely loved it, but it is, by now, old technology. If you have one, enjoy it, but if you are going to buy something new, the 2500K is not the best choice. The new chips are faster at stock speeds and much, much more frugal.

That 6500 looks like a good choice. It is the chip I was referring to earlier. It has more cache and a Max Turbo Frecuency that is close to the i3, so they should be about as fast in single threaded applications. As far as I can tell the i5-6600K is another 20 dollars more expensive and would allow for an overclock. The 6600K is the modern equivalent of the 2500K, though I understand you are pushing your budget already. I just want to make sure you have weighed your options before committing The i5-6500 certainly allows some serious gaming and serious production work, rendering and so on.

Also, drop that Kingston SSD and switch back the the BX200 you had before. The controller on the Kingston (and many other SSDs) is notoriously unreliable. The Crucials have a great track record.

Are you still sure you want a small motherboard, now that you have an ATX case? Everything should fit now (it never hurts to double check), but expanding anything in the future will be a bit harder.

Edited July 15, 2016 by Camacha

[Alphasus]

@Camacha

The 2500K is similarly fast to said i3. Really, the laptop isn't faster for anything except starting editing. I switched him back to the BX200 each time by the way. He doesn't have the budget for an EVO in my opinion. A 6500 should be fine, because 300 mhz don't matter.

http://pcpartpicker.com/list/r7h8Yr

Fixed with Camacha's suggestions, has an i5 in it. 

http://pcpartpicker.com/list/sWXNQV

Same thing, but a step up with a hyperthreaded Xeon. Now, this is important! If you have a MicroCenter nearby, buy the Xeon there. They have those chips for $200, which will give you 8 threads for editing. To buy that from a local microcenter saves you enough money that the price of the Skylake i5 build is comparable, and the Xeon build should be similar in single thread. But. hyperthreading will improve editing performance.

http://www.microcenter.com/product/442355/Xeon_E3_1231V3_35GHz_LGA_1150_Boxed_Processor

The Xeon box has full ATX, and the same case.

Edited July 15, 2016 by Alphasus

[Camacha]

16 minutes ago, Alphasus said:

@Camacha

[clip]

A single result can skew the view on matters. The clip maker/poster commented in the comments: "GPU vs CPU speeds really depend on the scene you are rendering." That is why doing research is so important. Without all the facts, making the right decision is a lot less likely. GPU rendering can be great, but it is no silver bullet.

16 minutes ago, Alphasus said:

He doesn't have the budget for an EVO in my opinion.

I see no reason the EVO would be preferred. Real life performance should be identical.

16 minutes ago, Alphasus said:

A 6500 should be fine, because 300 mhz don't matter.

It could be a lot more than 300 MHz, because the 6600K can be overclocked. That is the relevant difference, not the clockspeed in itself. If you can an overclocking upgrade for 20 dollar, I would go for it.

16 minutes ago, Alphasus said:

http://pcpartpicker.com/list/r7h8Yr

Fixed with Camacha's suggestions, has an i5 in it. 

http://pcpartpicker.com/list/sWXNQV

Same thing, but a step up with a hyperthreaded Xeon. Now, this is important! If you have a MicroCenter nearby, buy the Xeon there. They have those chips for $200, which will give you 8 threads for editing. To buy that from a local microcenter saves you enough money that the price of the Skylake i5 build is comparable, and the Xeon build should be similar in single thread. But. hyperthreading will improve editing performance.

http://www.microcenter.com/product/442355/Xeon_E3_1231V3_35GHz_LGA_1150_Boxed_Processor

The Xeon is a bit older, but could be an interesting choice. It pays to closely review relevant benchmarks of both chips, as they seem to have roughly similar performance and it depends on the task which will be faster. The Xeon will be less frugal and is slower clock for clock, but has the advantage of hyper-threading.

[Alphasus]

4 minutes ago, Camacha said:

A single result can skew the view on matters. The clip maker/poster commented in the comments: "GPU vs CPU speeds really depend on the scene you are rendering." That is why doing research is so important. Without all the facts, making the right decision is a lot less likely. GPU rendering can be great, but it is no silver bullet.

I see no reason the EVO would be preferred. Real life performance should be identical.

It could be a lot more than 300 MHz, because the 6600K can be overclocked. That is the relevant difference, not the clockspeed in itself. If you can an overclocking upgrade for 20 dollar, I would go for it.

The Xeon is a bit older, but could be an interesting choice. It pays to closely review relevant benchmarks of both chips, as they seem to have roughly similar performance and it depends on the task which will be faster. The Xeon will be less frugal and is slower clock for clock, but has the advantage of hyper-threading.

I'm saying that its faster in general. I exclusively render still images on GPUs, and the only time I see CPUs get close in speed to a GPU is in noisy, extremely complex images. Then, its a factor of 2 instead of 10.

Overclocking on any of the motherboard chipsets that were chosen doesn't work. That upgrade to a reasonable chipset is $50.

The Xeon's hyperthreading means that in <4 threaded loads, it is faster. In single core, its a tad slower, but not enough to matter as far as I've seen.

[Camacha]

1 minute ago, Alphasus said:

I'm saying that its faster in general. I exclusively render still images on GPUs, and the only time I see CPUs get close in speed to a GPU is in noisy, extremely complex images. Then, its a factor of 2 instead of 10.

That was the point: you cannot make broad claims like that. You need to look at the application and the hardware. Another major factor is flexibilty, as CPU renderers tend to be less finicky than GPU renderers from a programming and end user point of view. The only right answer is it depends.

1 minute ago, Alphasus said:

The Xeon's hyperthreading means that in <4 threaded loads, it is faster. In single core, its a tad slower, but not enough to matter as far as I've seen.

Hyper-threading does not work that way. It is rather more complicated. I wrote a more extensive explanation a little while back, but the workload really needs to fit a specific profile for hyperthreading to work out.

If you have found relevant results, you would be most welcome to share them.

[Alphasus]

5 minutes ago, Camacha said:

That was the point: you cannot make broad claims like that. You need to look at the application and the hardware. Another major factor is flexibilty, as CPU renderers tend to be less finicky than GPU renderers from a programming and end user point of view. The only right answer is it depends.

Hyper-threading does not work that way. It is rather more complicated. I wrote a more extensive explanation a little while back, but the workload really needs to fit a specific profile for hyperthreading to work out.

If you have found relevant results, you would be most welcome to share them.

I use the exact same application, and hardware that could be considered just a tier up. That video was of the same application, Blender, that Endersmen users.

http://www.cpu-monkey.com/en/compare_cpu-intel_core_i5_4690k-412-vs-intel_xeon_e3_1231_v3-437

Source. The only difference between the 1231 and 4690k is that the 1231 is 100 mhz slower. In cinebenches multithreaded benchmark, it pulls ahead. Blender uses every thread available as well, so I'm quite sure that it would fit said profile. Editing, not so sure.

http://www.cpu-monkey.com/en/compare_cpu-intel_core_i5_6500-524-vs-intel_xeon_e3_1231_v3-437

Faster in Cinebench's multi threaded load, but not all. 

[Camacha]

37 minutes ago, Alphasus said:

I use the exact same application, and hardware that could be considered just a tier up. That video was of the same application, Blender, that Endersmen users.

I might have chosen my words more carefully, though I figured the context made the intention clear. I was referring to the application of the program (how you use a program), not the application as a program (what program you use). As is indicated by the person that made the clip you used to show a difference, it depends on what you render, with what elements, lighting, settings, et cetera. A video card is very good at doing certain things efficiently, though it is also terrible at doing other things.

We are not the first to discuss this. The top answer here has a very decent explanation on how and why.

 

37 minutes ago, Alphasus said:

http://www.cpu-monkey.com/en/compare_cpu-intel_core_i5_4690k-412-vs-intel_xeon_e3_1231_v3-437

Source. The only difference between the 1231 and 4690k is that the 1231 is 100 mhz slower. In cinebenches multithreaded benchmark, it pulls ahead. Blender uses every thread available as well, so I'm quite sure that it would fit said profile. Editing, not so sure.

http://www.cpu-monkey.com/en/compare_cpu-intel_core_i5_6500-524-vs-intel_xeon_e3_1231_v3-437

Faster in Cinebench's multi threaded load, but not all. 

Yes, rendering is the type of task that lines up perfectly with what hyper-threading does. Internally, there are lots of small tasks/instructions that require constant switching. Hyper-threading thrives in those conditions. The results with other tasks vary a bit. The Xeon is an interesting chip, even if it is from an earlier generation. I view the fact that it comes with a full size motherboard also as an advantage. Hyper-threading will not help out with gaming as much, though, and its age also means that you need DDR3 memory, which is on the way out. You might want to pick according to the type of workload you prefer (GPU rendering, CPU rendering, gaming or something else).

Regardless, either choice should yield a very fast system that will last a while.

[Endersmens]

10 hours ago, Camacha said:

I

Also, drop that Kingston SSD and switch back the the BX200 you had before. The controller on the Kingston (and many other SSDs) is notoriously unreliable. The Crucials have a great track record.

Are you still sure you want a small motherboard, now that you have an ATX case? Everything should fit now (it never hurts to double check), but expanding anything in the future will be a bit harder.

I couldn't find the crucial. I have no preference on board, I went with cheapest with good reviews  I also looked at the 6600k, but the problem is other people are already trying to convince me to get a rx 480 instead of a gtx 960 and I definitely can't squeeze the 6600k and the 480  

9 hours ago, Alphasus said:

@Camacha

The 2500K is similarly fast to said i3. Really, the laptop isn't faster for anything except starting editing. I switched him back to the BX200 each time by the way. He doesn't have the budget for an EVO in my opinion. A 6500 should be fine, because 300 mhz don't matter.

http://pcpartpicker.com/list/r7h8Yr

Fixed with Camacha's suggestions, has an i5 in it. 

http://pcpartpicker.com/list/sWXNQV

Same thing, but a step up with a hyperthreaded Xeon. Now, this is important! If you have a MicroCenter nearby, buy the Xeon there. They have those chips for $200, which will give you 8 threads for editing. To buy that from a local microcenter saves you enough money that the price of the Skylake i5 build is comparable, and the Xeon build should be similar in single thread. But. hyperthreading will improve editing performance.

http://www.microcenter.com/product/442355/Xeon_E3_1231V3_35GHz_LGA_1150_Boxed_Processor

The Xeon box has full ATX, and the same case.

Don't know about the Xeon. I'm wanting this primarily for gaming and recording. Editing and blendering come second. 

[cantab]

11 hours ago, Camacha said:

As far as I can tell the i5-6600K is another 20 dollars more expensive and would allow for an overclock.

The gotcha there is the 6600K doesn't come with a cooler, so add another 20 dollars or more for an aftermarket cooler, and overclocking wants a Z170 motherboard which is a further extra cost. Of course the 6600K is a good choice for many people but the overall cost difference versus the locked i5's is bigger than it looks.

35 minutes ago, Endersmens said:

 I also looked at the 6600k, but the problem is other people are already trying to convince me to get a rx 480 instead of a gtx 960 and I definitely can't squeeze the 6600k and the 480  

Try and find some Blender benchmarks of the RX 480. Relative performance - and compatibility and other factors - might not be the same as for gaming. (Not that all games are the same either.)

[Alphasus]

2 hours ago, Endersmens said:

I couldn't find the crucial. I have no preference on board, I went with cheapest with good reviews  I also looked at the 6600k, but the problem is other people are already trying to convince me to get a rx 480 instead of a gtx 960 and I definitely can't squeeze the 6600k and the 480  

Don't know about the Xeon. I'm wanting this primarily for gaming and recording. Editing and blendering come second. 

Its a crucial BX200.Amazon has them commonly.

Blender doesn't like openCL, the way that AMD and intel cards render. In general, it is slower at doing so. The RX 480 isn't worth it for what you're doing anyways. It also can kill your motherboard sooo...

Blender much prefers CUDA, which Nvidia cards prefer.

https://blenderartists.org/forum/showthread.php?402080-RX480-or-GTX-1070

A 290x is similar in speed to an RX 480, as they are both close to the GTX 970 for games(not compute).

A 680 is similar to a 960. An RX 480 should be similar to a 960, but it has poor optimization for Blender because of OpenCL, runs hotter, won't have an impact on performance for what you play, doesn't work with GameWorks, and can kill your motherboard. And you don't have budget for it. But its faster for gaming, however your games will max with a 960.

https://steamcommunity.com/discussions/forum/11/620703493334221287/

Is there a microcenter near you that you or your parents could drive you to?

Edited July 15, 2016 by Alphasus

[Camacha]

@Endersmens

- We are discussing both CPU and GPU rendering. If you look at both and spend money to do both, you will have a flexible system that is also prepared for other renderers you might want to use later on. Flexibility, on the other hand, costs money.
- If gaming is your main concern, the focus of the discussion on rendering seems a bit trivial.

 

3 hours ago, cantab said:

The gotcha there is the 6600K doesn't come with a cooler, so add another 20 dollars or more for an aftermarket cooler, and overclocking wants a Z170 motherboard which is a further extra cost. Of course the 6600K is a good choice for many people but the overall cost difference versus the locked i5's is bigger than it looks.

I referred to that before, but that might have gotten buried between the configuration posts. It will take more of an investment, aye. If money is tight already, that might be a problem. An air cooler is upgraded easily later on, but a motherboard is not easy to replace. Considering we are already stretching the budget, dropping overclocking seems to be a logical choice.

 

2 hours ago, Alphasus said:

Blender doesn't like openCL, the way that AMD and intel cards render. In general, it is slower at doing so. The RX 480 isn't worth it for what you're doing anyways. It also can kill your motherboard sooo...

Blender much prefers CUDA, which Nvidia cards prefer.

https://blenderartists.org/forum/showthread.php?402080-RX480-or-GTX-1070

A 290x is similar in speed to an RX 480, as they are both close to the GTX 970 for games(not compute).

A 680 is similar to a 960. An RX 480 should be similar to a 960, but it has poor optimization for Blender because of OpenCL, runs hotter, won't have an impact on performance for what you play, doesn't work with GameWorks, and can kill your motherboard. And you don't have budget for it. But its faster for gaming, however your games will max with a 960.

https://steamcommunity.com/discussions/forum/11/620703493334221287/

Is there a microcenter near you that you or your parents could drive you to?

You seem to have a strong bias against AMD. Please let that not get in the way of proper advice. The point of this discussion is to get Endersmens the most bang for his buck. I will comment point by point.

- An RX480 does not kill your motherboard. The story about the GPU not adhering to ATX specifications is blown out of proportions, mostly by people that do not understand how things are properly measured. No examples of an RX480 actually causing any problems whatsoever exist, which means the actual problem is non-existent.
- We should note rendering is not the primary concern, but gaming is.
- When it comes to gaming, the RX480 appear to be much more value for money than Nvidia's competition.
- With a modern processor (which both the i5-6500 and the Xeon are), the GPU will pretty much always be the limiting factor. A faster card means more frames.
- There seems to be few direct comparisons between OpenCL and CUDA rendering. It mostly is people saying the same things without proving them. It does seem that OpenCL has been under development longer.

The RX480 is an very interesting choice at the moment. Nothing Nvidia has to offer for the same money comes close to it and it is not that fast that the performance is wasted. If you look at a random review, you will see it generally makes a big difference at 1080P in games where the framerate is not already over 60-80 fps. The difference between 30 and 50 fps is very noticeable and most games show similar performance improvements.

Mind you, I am not saying the GTX960 is not a viable option. I am just saying that the RX480, if you have the money to spend, seems to be a better option when it comes to gaming at 1080P. If the money is not there, sure, buy a slightly less nimble card that will still provide an good experience. If needs be, a GPU can be replaced with ease later on, so punching below your weight on the GPU is not a disaster.