i7 6700K VS i7 4790K

[l_zheng101]

1 minute ago, Monarch said:

 

i7 and i5 have no difference in performance? Games don't use hyperthreading? Where did you get those silly ideas from? You people obviously have no idea what you're talking about, please do a thorough research first. Thank you.

 

 

I think you're the one who needs to do research.

[Vishal Gupta]

Just now, SteveGrabowski0 said:

Also, high quality DDR4 isn't much more expensive than high quality DDR3. I'd easily pay $10 for the upgrade.

 

PCPartPicker part list / Price breakdown by merchant

Memory: G.Skill Ripjaws V Series 16GB (2 x 8GB) DDR4-3000 Memory  ($79.99 @ Newegg) 
Total: $79.99
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2016-03-10 11:40 EST-0500

 

 

PCPartPicker part list / Price breakdown by merchant

Memory: G.Skill Sniper Series 16GB (2 x 8GB) DDR3-2400 Memory  ($69.99 @ Newegg) 
Total: $69.99
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2016-03-10 11:42 EST-0500

I guess comparing ddr4 to ddr3 on the basis of price is too complicated . Cuz with higher memory comes higher CAS latency which should be lower.haha

To be really accurate

that ddr3 is 4.53 nanoseconds while ddr4 is exactly 5 nanosecond timing.

No need to upgrade brother.There is a loss in performance between the ones you mentioned

( ACCORDING TO ME. I DONT KNOW IF IM RIGHT OR WRONG.)

[SteveGrabowski0]

48 minutes ago, Vishal Gupta said:

Which i5 would be the best? Lets leave price for a moment. And another question.

Whats hyper-threading really for?

 

 

 

Hyperthreading is designed to keep your cpu running at its full potential. Lets say you have a quadcore cpu like an i5. Then windows schedules one thread at a time onto each core. Say one of these threads has to load some data. Let's also say this data is not in cache, so it has to read it from your RAM. This is extremely slow relative to how fast the cpu can execute instructions, so the thread that's waiting for the data to be fetched from memory just sits there doing nothing until the data is fetched from RAM. Which means the core is doing nothing. 

 

Now if you had an i7 Windows could schedule two threads to each core simultaneously. So when the previous situation happens, instead of the core doing nothing while waiting to fetch data from memory it can switch to the other thread scheduled on the core. And then that thread executes until it stalls. So instead of having your core sitting there doing nothing while waiting for a fetch of data from memory, it's able to do useful work on another thread. The i7 can do this switching quickly since it has two sets of registers on each core: one for the first hardware thread, one for the second. The i5 only has one set of registers on each core, so it can't do hyperthreading.

 

TL;DR version: hyperthreading allows your cpu to keep doing useful work when an executing thread stalls, which happens all the time since there are many things like data stores/loads that will bottleneck your cpu's performance since they are so much slower. 

[Monarch]

13 minutes ago, l_zheng101 said:

 

I think you're the one who needs to do research.

LTT? Really? Those guys don't know anything. Their testing methodlogy is fundamentally flawed. To see a difference between an i5 and i7 you need a CPU-bound, multi-threaded scenario. You can find it in pretty much all AAA titles, but it depends on the level and what's going on in the game.

 

 

https://youtu.be/b6LUufXCPDM?t=3m14s

There's up to ~30 fps difference. 

 

You can also see huge difference here:

 

https://youtu.be/N0HlPN-69ck?t=1m12s

 

This is in Fallout 4:

Huge difference in both average and minimum fps.

 

@Ykno

[Vishal Gupta]

Just now, SteveGrabowski0 said:

Hyperthreading is designed to keep your cpu running at its full potential. Lets say you have a quadcore cpu like an i5. Then windows schedules one thread at a time onto each core. Say one of these threads has to retrieve some data, say the location of an enemy. Let's also say this data is not in cache, so it has to read it from your RAM. This is extremely slow relative to how fast the cpu can execute instructions, so the thread that's waiting for the data to be fetched from memory just sits there doing nothing until the data is fetched from RAM. Which means the core is doing nothing. 

 

Now if you had an i7 Windows could schedule two threads to each core simultaneously. So when the previous situation happens, instead of the core doing nothing while waiting to fetch data from memory it can switch to the other thread scheduled on the core. And then that thread executes until it stalls. So instead of having your core sitting there doing nothing while waiting for a fetch of data from memory, it's able to do useful work on another thread. The i7s can do this switching quickly since it has two sets of registers on each core: one for the first hardware thread, one for the second. The i5 only has one set of registers on each core, so it can't do hyperthreading.

 

TL;DR version: hyperthreading allows your cpu to keep doing useful work when an executing thread stalls, which happens all the time since there are many things like data stores/loads that will bottleneck your cpu's performance since they are so much slower. 

Thank you. Thats a wonderful explanation.

 

 

[SteveGrabowski0]

9 minutes ago, Vishal Gupta said:

I guess comparing ddr4 to ddr3 on the basis of price is too complicated . Cuz with higher memory comes higher CAS latency which should be lower.haha

To be really accurate

that ddr3 is 4.53 nanoseconds while ddr4 is exactly 5 nanosecond timing.

No need to upgrade brother.There is a loss in performance between the ones you mentioned

( ACCORDING TO ME. I DONT KNOW IF IM RIGHT OR WRONG.)

That DDR4 kit has minimally higher latency but much higher burst speed. The DDR4-3000 kit will likely mop the floor with the DDR3-2400 kit, though @MageTank is the person on this forum who seems to understand memory best, so maybe he can throw his two cents in here.

[l_zheng101]

2 minutes ago, Monarch said:

LTT? Really? Those guys don't know anything. Their testing methodlogy is fundamentally flawed. To see a difference between an i5 and i7 you need a CPU-bound, multi-threaded scenario. You can find it in pretty much all AAA titles, but it depends on the level and what's going on in the game.

 

 

https://youtu.be/b6LUufXCPDM?t=3m14s

There's up to ~30 fps difference. 

 

You can also see huge difference here:

 

https://youtu.be/N0HlPN-69ck?t=1m12s

 

This is in Fallout 4:

Huge difference in both average and minimum fps.

You say LTT doesn't know anything yet you are a member of this forum, get lost.

[Monarch]

Just now, l_zheng101 said:

You say LTT doesn't know anything yet you are a member of this forum, get lost.

I'm here for the community, not Linus and his videos.

[Vishal Gupta]

3 minutes ago, SteveGrabowski0 said:

That DDR4 kit has minimally higher latency but much higher burst speed. The DDR4-3000 kit will likely mop the floor with the DDR3-2400 kit, though @MageTank is the person on this forum who seems to understand memory best, so maybe he can throw his two cents in here.

Would you mind to elaborate and explain?

[Ykno]

17 minutes ago, Monarch said:

1 minute ago, Monarch said:

I'm here for the community, not Linus and his videos.

 

 

 

6700K and 4790k equal? i7 and i5 have no difference in performance? Games don't use hyperthreading? Where did you get those silly ideas from? You people obviously have no idea what you're talking about, please do a thorough research first. Thank you.

 

 

we on about real world preformance aswell not what a score sheet says real world looking at that game play you wont see a diffrence!

[Monarch]

1 minute ago, Ykno said:

we on about real world preformance aswell not what a score sheet says real world looking at that game play you wont see a diffrence!

I just gave you evidence that there is a huge difference in real world usage. If you can't see the 30 fps difference then you're either not paying attention or your vision is impaired.

[Godlygamer23]

2 minutes ago, Ykno said:

we on about real world preformance aswell not what a score sheet says real world looking at that game play you wont see a diffrence!

In Battlefield 3, the 4770K was superior to a 4670K at the same clocks. 

https://www.techpowerup.com/reviews/Intel/Core_i5_4670K_and_i7_4770K_Comparison/

[Ykno]

2 minutes ago, Godlygamer23 said:

In Battlefield 3, the 4770K was superior to a 4670K at the same clocks. 

https://www.techpowerup.com/reviews/Intel/Core_i5_4670K_and_i7_4770K_Comparison/

6fps diffrence is that worth the extra £100 when you can spend that extra £100 on a better gpu and get 20-30 fps increase?

which is what i been trying to say from the start

[Atmos]

39 minutes ago, SteveGrabowski0 said:

The 1231v3 is not a 4790, it's 200 MHz slower.

You realize that if I were to take my 4790k and down clock it to 3.8ghz, it doesn't stop being a 4790k... right?

Its the same architecture, and aside from extra features likes support for ecc memory, it is quite literally a 4790, with the integrated graphics disabled. And if you were just trying to be cheeky with me, then shame on you; because there's honestly people that believe that kinda stuff.

 

51 minutes ago, Kirky2k15 said:

Wish I'd known that when I was building my rig...

Yea when I was first building I didn't realize it either until a member here on the forums pointed it out to me. I would have gone for one, had I not fallen into a generous raise from my at the time employer. It's a damn good cpu, I recommend it way more than any 6600k or 4690k, because honestly 90% of people won't overclock those cpus far enough to make up for the lack of hyperthreading in heavily multithreaded applications. It's also just good to have a bit of slack when it comes to the cpu, rather than running it pegged all the time streaming, gaming, and doing w/e else at the same time.

[Godlygamer23]

1 minute ago, Ykno said:

6fps diffrence is that worth the extra £100 when you can spend that extra £100 on a better gpu and get 20-30 fps increase?

It depends on the game, and it's a subjective decision. Would I automatically purchase a 4790K? Not necessarily. 

[Ykno]

2 minutes ago, Godlygamer23 said:

It depends on the game, and it's a subjective decision. Would I automatically purchase a 4790K? Not necessarily. 

ye i get that but all im doing is giving him advise to say u might not need the i7 you might be better off saving extra cash getting better gpu and a i5 but end day it his decision and he should buy what he feels is right for him and his situation

[Godlygamer23]

Just now, Ykno said:

ye i get that but all im doing is giving him advise to say u might not need the i7 you might be better off saving extra cash getting better gpu and a i5 but end day it his decision and he should buy what he feels is right for him and his situation

I agree with that, and a 6600K is more than likely sufficient. 

[Ykno]

1 minute ago, Godlygamer23 said:

I agree with that, and a 6600K is more than likely sufficient. 

definetly but if his budget is tighter then haswell is a good option which is just as good obvs not the same preformance but a great price to preformance subsitute

[Monarch]

@Ykno

If you're CPU bottlenecked you're not going to get more fps with a better GPU. Building a PC is very simple. You get a good CPU to get good framerate in CPU-bound scenarios, you get a good GPU to get good framerate in GPU-bound scenarios. If you get a decent CPU and a great GPU you're going to have a bottleneck. Now, whether that bothers you or not and whether you think an i7 is worth its price or not is up to you, but let's stick to facts: i7 > i5.

[SteveGrabowski0]

2 minutes ago, Ykno said:

we on about real world preformance aswell not what a score sheet says real world looking at that game play you wont see a diffrence!

One flaw I see in this argument is that people don't buy their cpus to play just right now. But games tend to get harder to run over time as they have better hardware to run on. There is no question 2015 games hit the cpu a lot harder than 2014 games with games like GTA V and Fallout 4 that were very cpu bound and then Witcher 3 that had strongly cpu bound sections too (eg Novigrad). Seems like 2016 is shaping up similarly. Rise of Tomb Raider dies on an i3 when all 2015 games I know of could play pretty well on one.

[Vishal Gupta]

Just now, Ykno said:

definetly but if his budget is tighter then haswell is a good option which is just as good obvs not the same preformance but a great price to preformance subsitute

I don't have a tight budget.

 

[Atmos]

7 minutes ago, Monarch said:

If you're CPU bottlenecked you're not going to get more fps with a better GPU. Building a PC is very simple. You get a good CPU to get good framerate in CPU-bound scenarios, you get a good GPU to get good framerate in GPU-bound scenarios. If you get a decent CPU and a great GPU you're going to have a bottleneck. Now, whether that bothers you or not and whether you think an i7 is worth its price or not is up to you, but let's stick to facts: i7 > i5.

What does that have to do with what I said?

the xeon 1231v3 is a locked i7... for the price of an unlocked i5... it's a win win scenario rofl. You can afford an excellent cpu, with an excellent gpu.

[Monarch]

2 minutes ago, Atmos said:

What does that have to do with what I said?

the xeon 1231v3 is a locked i7... for the price of an unlocked i5... it's a win win scenario rofl

Didn't mean to quote you, but Ykno. 

[SteveGrabowski0]

33 minutes ago, Vishal Gupta said:

Would you mind to elaborate and explain?

You have a sequence of delays in going from cpu to memory. You have latency from trying to find out from where in the grid of memory to retrieve that data, as well as going to that specific location (think CAS latency). Then you have a transmission delay as the data is transmitted from that location of memory onto the bus (think memory speed). Then you have a propagation delay as the data travels on the bus from the memory to the cpu. This goes at the speed of light and is dependent on how far your DIMMs are from the CPU, which is why memory is always put right next to the cpu (and one of the reasons registers are so much faster than cache and cache is faster than memory, since registers and cache are on the physical processor).

 

Think of it this way. You want to download a movie either from some download site like depositfiles vs getting it off a torrent. Your latency is lower on depositfiles since you just click a link, while on torrent you have to wait for it to find seeders. But depositfiles will only let you download maybe 500 KB/s while once enough seeds are found maybe you're getting 2MB/s or more. I know I'd rather go with the torrent.

 

Of course this analogy is more extreme than the memory thing. The latencies seem very similar for the DDR3-2400 kit vs the DDR4-3000 kit I posted. For instance, the CAS latency is the time to find the correct column in memory, and at 11 cycle per strobe at 1200 MHz = 1200 * 10^6 cycle/second you're looking at a latency of 9.17 nanosecond per strobe with the DDR3-2400 kit. While with CAS 15 DDR4-3000 you're looking at 15 cycle per column strobe at 1500 MHz, giving a latency of 10 nanosecond per column strobe. So you're taking what looks like a 9% hit in latency. But you're getting a 25% bump in throughput thanks to the DDR4 kit doing 3000 MT/s vs the DDR3 kit doing 2400 MT/s.

[Vishal Gupta]

3 minutes ago, SteveGrabowski0 said:

You have a sequence of delays in going from cpu to memory. You have latency from trying to find out from where in the grid of memory to retrieve that data, as well as going to that specific location (think CAS latency). Then you have a transmission delay as the data is transmitted from that location of memory onto the bus (think memory speed). Then you have a propagation delay as the data travels on the bus from the memory to the cpu. This goes at the speed of light and is dependent on how far your DIMMs are from the CPU, which is why memory is always put right next to the cpu (and one of the reasons registers are so much faster than cache and cache is faster than memory).

 

Think of it this way. You want to download a movie either from some download site like depositfiles vs getting it off a torrent. Your latency is lower on depositfiles since you just click a link, while on torrent you have to wait for it to find seeders. But depositfiles will only let you download maybe 500 KB/s while once enough seeds are found maybe you're getting 2MB/s or more. I know I'd rather go with the torrent.

 

Of course this analogy is more extreme than the memory thing. The latencies seem very similar for the DDR3-2400 kit vs the DDR4-3000 kit I posted. For instance, the CAS latency is the time to find the correct column in memory, and at 11 cycle per strobe at 1200 MHz = 1200 * 10^6 cycle/second you're looking at a latency of 9.17 nanosecond per strobe with the DDR3-2400 kit. While with CAS 15 DDR4-3000 you're looking at 15 cycle per column strobe at 1500 MHz, giving a latency of 10 nanosecond per column strobe. So you're taking what looks like a 9% hit in latency. But you're getting a 25% bump in throughput thanks to the DDR4 kit doing 3000 MT/s vs the DDR3 kit doing 2400 MT/s.

What I did to judge a good ram was:-

Latency/Memory speed *1000 = x nanoseconds

So this means I am wrong?If I'm wrong ,then what was I comparing between the two by finding out the nanoseconds??