PDA

View Full Version : i7 4790k vs i7 5820K Which is good to have?



yoshiii
01-28-2015, 08:41 AM
Hello

I am probably go to get the i7 4790k system but I think I might be able to buy a i7 5820k system.

Which is better for Light Wave? Is it worth it to buy the i7 5820K?

Sensei
01-28-2015, 09:04 AM
According to
https://www.cpubenchmark.net/high_end_cpus.html
5820K is 15% faster.

If 5820K will render at 60 sec/frame, 4790k will do it in 69 sec/frame.

magiclight
01-28-2015, 09:13 AM
OpenGL will run at the same speed so no difference there, a little faster rendering (maybe), depends a little on what you use it for also, maybe better to go for more memory (or one that you can add more memory to later on).

Single threaded the 4790 beats the 5820.
you will need DDR4 memory for the 5820 (higher price)
looking at price you don't get that much for all the extra money.
I would go for the 4790.

I have a 4770 with 32 gig ram and it works very nice with Lightwave I would say.

Danner
01-28-2015, 10:03 AM
It's a close call, 5820K is a faster for rendering but a bit more expensive. The 4790 is in theory slightly better at modeling (better single thread performance). I'd go with the 5820k personally.

VirtualFM
01-28-2015, 12:14 PM
I got myself a 5820K and overclocked it to 4.125Ghz. The Asus Motherboard i got allowed to go up to 4.5Ghz without trouble, but I felt it was safer to play on lower speeds, specially because there will be a lot of rendering and it would get too hot for too long (even if I got an excelent cooler). It's perfectly stable like this even if it's rendering for a whole week.

The downside I found with this CPU was that it requires a motherboard that requires DDR4, which is twice as expensive as equivalent DDR3 and I don't think it's worthwhile. Other than that perfectly happy with it.

yoshiii
01-28-2015, 12:44 PM
I got myself a 5820K and overclocked it to 4.125Ghz. The Asus Motherboard i got allowed to go up to 4.5Ghz without trouble, but I felt it was safer to play on lower speeds, specially because there will be a lot of rendering and it would get too hot for too long (even if I got an excelent cooler). It's perfectly stable like this even if it's rendering for a whole week.

The downside I found with this CPU was that it requires a motherboard that requires DDR4, which is twice as expensive as equivalent DDR3 and I don't think it's worthwhile. Other than that perfectly happy with it.

Yeah, it seems these cpu makers like making 6 plus core cpus with lower clock speed. I wonder why. I want the power of the 6 cores(thats one reason why I was thinking about the FX8350 because it has 8 cores) but I dont like that the per core speed is lower than the i7 4790k. I like the speed of the i7 4790k, but dont like the fact that it is only 4 cores, I know it has hyperthreading butI want real cores.

So all of you who have the i7 5820K, are happy with your purchase?

rustythe1
01-28-2015, 01:16 PM
don't be fooled by the lower clock speed, I used to have a i7 980x and now have the i75960x, on paper there is not much between them apart from 2 extra cores, the other thing the 5960 has though is the L3 20 mb cache, and it can access any or all of the 4 memory channels, the upshot of this is that I had a 10,000 frame render that took 4 days on my old 980x machine, on my new 5960 it only took 24 hours! a small increase in one area can have an exponential effect in another, if you think how lightwave works, the processor starts at zero, upto 100% then drops off, so it creates a sort of wave, if you sharpen the up and down of that wave (ie by allowing the software to access faster and more parts at the same time, that's where the l3 cache comes in), the processor spends more time in the high part, that then speeds up frame times over the top of the speed up from the extra clock speed, if they were fairly short frame times you could see two or three times a speed up (like I did!) for not much extra,
e.g. if you had frames that were only taking 6 seconds, it could take 2 seconds to get process up to 100% and 2 seconds back down, if you could get the processor up in 0.1 seconds you have taken nearly 4 seconds off 6 and done it in just over 2.

netstile123
02-06-2015, 09:02 PM
I just put together a new system with the 5930K. I am doing a render with the marble benchmark right now. will post the results when done. Anyone no where ti find the page with everyones results?

well with stock speed @ 3.5 mhz. took 1 hr. 9 minutes . That is making now changes to the scene and just stock.

JamesCurtis
02-06-2015, 10:01 PM
My last system used a i7 920 and my newly built system is using the i7 5820 and LW renders with the newer chip is 2.4 times faster than my older system. Of course there is 7 year difference between the two systems, but hey, not a bad difference. Relative cost of the two machines is roughly the same!! Would have liked to go higher in i7 processor's but couldn't afford the cost.

VirtualFM
02-09-2015, 03:45 PM
I got an i7 5820K which I overclocked to 4.125Ghz and got around 2.5-3 times faster than my old Q9550 that was also 7-8 years old. I was expecting more... maybe because I was used to upgrade computers every 3 years, in the end 90's/early 2000's, and got increases of 3 to 5 times faster, so after almost 8 years I was expecting a real boost, which didn't happen. So either I already a really good system before (I had, otherwise it wouldn't last for so long) or the CPU progress rate is not going as fast as 10/15 years ago.

Oh, I got 1h 26m 41s (5201s). Even if I had other programs running I'm not overly impressed.

spherical
02-09-2015, 04:08 PM
In order to keep things simple and not get bogged down in differences in cache and memory handling, etc. and just examine CPU cycles,

Q9550 is 4 cores at 2.83 GHz = 11.32
i7-5820K is 6 cores at 3.3 Ghz = 19.80

Expecting 500% increase in performance using the above numbers is unrealistic.

Overclocking the 5820K = 24.75

24.75 / 11.32 = 2.186 which is about what you estimate the observed speed increase to be. I'd be satisfied; always wanting more, but getting what is possible.

VirtualFM
02-09-2015, 05:49 PM
Well, like I said, either I had a good system already or nothing much happened in terms of computer architecture lately. It's not only CPU cycles that count, there is RAM efficiency to take in consideration, motherboard architecture, CPU's internal architecture, etc And some years ago there were really big jumps happening.

I'm not even counting about changing from my Amiga 4000 (with a $1500 Warp Engine accelerator!) to a Pentium Pro 200 which gave me a 15 times (!!!) speed increase, I'm talking about going from a Pentium [email protected] (probably overclocked) to a Core 2 Duo [email protected] That gave a 5.15 times speed increase in a timespan of 3.5 years.

There were real changes at the core of the CPUs before, nowadays they just seem to add more cores. Why? I guess that for the common user computers nowadays are as good as they get, you can have a ton of apps loaded and running and still don't feel any slowdown, games are just awesome even with average graphics cards, so there is no need to invest anymore in computer speed unless you work in 3D or computer science. As someone who is an enthusiast of computer power and likes to squeeze everything I can out of it, the present situation is rather unfortunate and I expected better.