Will upgrading from the i5-4690k to the i7-4790k really make a huge improvement?

[Nick Arno]

I'll be using this rig for recording 60fps 1080p gameplay. I don't care so much about how long rendering takes. I want to make sure that I'll see a noticeable performance increase in recording. And I know streaming is CPU intensive, so will this upgrade help with that too? I really do need the truth on this. Will spending the extra money actually help me specifically with recording and streaming?

 

[Sig2525]

interms of gaming not much

but in software rendering video rendering, streaming the i7 will be a huge upgrade - these processors were made for those kind of workloads

 

[Nick Arno]

interms of gaming not much

but in software rendering video rendering, streaming the i7 will be a huge upgrade - these processors were made for those kind of workloads

Perfect! and recording with a H.264 I'll also see a huge improvement?

 

[synphul]

You may want to check some benchmarks. That's a huge price hike of $100 over the 4690k for 2mb cache and hyperthreading. The 4790k comes clocked faster than the 4690k out of the box but also has less overclocking headroom so the two pretty much peak out within 100-200mhz of each other. The i7's with a big advantage over the 4690k are the 6 and 8 core versions. If it helps (it may) it will be very little boost in performance. Real world hyperthreading performance never has been more than about 10%, maybe 15% on average. Those are the types of increases someone would get going from a 3rd gen to 4th gen cpu and it's not recommended for the lack of value. I don't see this being any different, just my opinion on it.

 

[synphul]

This might make it a little more clear.

A direct comparison between the two you're asking about
http://www.anandtech.com/show/8227/devils-canyon-review-intel-core-i7-4790k-and-i5-4690k/3

Just doing quick math on the handbrake test, at stock speeds the 4790k has 12% more performance. Although I have a feeling it's more to do with the stock base clock being 500mhz higher out of the box.

When overclocked to the same speeds, the hyperthreading and extra 2mb of on die cache netted a little under 3% performance increase.

 

[Nick Arno]

You may want to check some benchmarks. That's a huge price hike of $100 over the 4690k for 2mb cache and hyperthreading. The 4790k comes clocked faster than the 4690k out of the box but also has less overclocking headroom so the two pretty much peak out within 100-200mhz of each other. The i7's with a big advantage over the 4690k are the 6 and 8 core versions. If it helps (it may) it will be very little boost in performance. Real world hyperthreading performance never has been more than about 10%, maybe 15% on average. Those are the types of increases someone would get going from a 3rd gen to 4th gen cpu and it's not recommended for the lack of value. I don't see this being any different, just my opinion on it.

But for recording with a H.264 Codec and streaming.. Would it make more of a difference? Because thats what I really need it for. Also It would be helpful with video rendering, though, thats not what I'm too worried about. I'm worried about the streaming and recording improvements.

 

[synphul]

I don't see it making a huge difference. It might, but in reality the i7 in the quad core has no magic behind it. It's a quad core same as the i5. They come clocked a bit different out of the box but if comparing both k models (since your i5 is a k version) they overclock just about the same. Hyperthreading is a thread optimizer, it's not a core doubler. 4 cores at xyz speed of the same family/generation can only process so much. I'm not saying hyperthreading can't help 'some' but to what extent is a questionable matter. Even though it can schedule 2 threads per core, each core can still only process a single thread at a time. It may alternate it's attention from one thread to the other, but it's not the same as a dedicated core processing a dedicated thread.

Recently I've noticed people talking about hyperthreading like it's the best thing since sliced bread and it's nothing new. It's been around for years and it wasn't a huge peformance boost back in the day either. I don't know if the current crowd doesn't remember it or what but I see a lot of talk of hyperthreading greatness like it's adding additional cores and the benchmarks continue to reflect a different story.

I'm trying to look at it from a realistic point of view, you just dumped $220-240 into that i5 and it's only 6mo old since it was released in what, august of 2014? Now you're looking to get rid of it (or maybe reuse it in another system) and in less than half a year are ready to dump another $320-340 into a processor with around 10% performance increase. For $600 in cpu's you should have a decent 6core with ht to show for all that, that would make a significant difference. It's ultimately up to you but it's a lot like someone with an fx 8320 hoping to see vast improvements moving to the fx 8370 or 9xxx. People may argue the comparison isn't the same and it's more like going from a 6xxx to a 9xxx in amd, but bottom line they're still capable of similar clocks at 4 cores. A hyperthreaded i7 quad doesn't perform like a 6 core or 8 core i7, it performs like a quad core since that's what it is. It's plenty generous to say that any advantage ht gives the i7 is around 10% which is roughly stepping up a generation in the same cpu line. Is that slight performance margin going to be worth an additional $340 to you?

 

[TofuLion]

i agree. the performance increase is not worth 150% cost. especially if you already have the i5. it's simply not logical, you should have got the i7 the first time. even then, it wasn't worth the extra cost. i've done my own "real world" evaluation by disabling hyperthreading and running a few tests (can't disable the L3 or i would). all-in-all, they are very similar, and the price difference is better invested in an after market cooler that can handle a hefty overclock. the core speed is where the difference is made between these two CPUs

 

[synphul]

The trouble is there are no accurate benchmarks showing streaming performance specifically. It has many variables including your internet's upload speed, whether you offload the work to the gpu for acceleration or not, whether using obs or xsplit etc etc. Many people say lots of things but it doesn't make it so unfortunately. I prefer to find facts if at all possible.

Earlier today was a perfect example, I was discussing photoshop and gpu acceleration with a friend of mine and landed on a very nice professional looking website - riddled with erroneous information. For starters, referring to i7 quad cores as 8 core cpu's (when the links provided to pcpartpicker were clearly 4 core hyperthreaded i7's) down to multiple suggestions for nvidia video cards for the purposes of gpu acceleration using it's cuda core power. Slight problem, adobe's site explicitly says that it's mge acceleration engine does NOT use cuda technology and instead uses opencl and opengl. There was no boost to be had in that application by sticking solely with nvidia cards and not a single amd/ati gpu was recommended. Misinformation runs rampant.

 

[synphul]

This isn't a streaming benchmark, rather just a general multitasking benchmark where you would assume that hyperthreading would show it's power. Notice where the 4th gen i5 and i7 falls in the lineup both at stock and when overclocked. When overclocked there's very little difference in an environment where multiple threads should be taken advantage of. It makes the existing cores a tad more efficient but doesn't make them process things any faster.

http://www.bit-tech.net/hardware/2014/07/03/intel-core-i5-4690k-review/4