Intel has only gained 22% performance from 2011 to 2017

[Fascinate]

https://www.overclock3d.net/reviews/...k_cpu_review/8

7600k at stock gets a 674 in cinebench R15. I just tested my 2500k which also runs at 4.2ghz and scored a 552. How pitiful is that lol, 22% gain in 6 years time.

It may be more like 23-24% because the stock 7600k only boosts 1 core to 4.2ghz but the fact intel could have easily released the 2500k with a 4.2ghz boost clock just kinda shows how little they have done in 6 years, they only have increased clocks since with hardly any performance gain.

Was just kind of curious of the actual gains since then because you see various reports, but cinebench is a great evaluator of raw performance.

[hulkgor]

They have no interest in 'rushing' technology further at a faster pace. It would only hinder their sales... it's more profitable to keep it at a slow pace.

[Chickat]

I think 15-20% per die shrink is acceptable. And 5ish on the tocks is okay. They need to do better though yes.

[Fascinate]

I think its more that we just dont need more desktop performance, mobile is the driving force these days and desktop performance is so light years ahead of whats needed for those types of apps that they are just focused on power savings.

Anyway i just found it interesting, i knew there havent been immense gains but i was honestly expecting more than 22% lol.

[schwarzkopf]

Performance in the last 5 years has been focused on multiprocess/multithread and floating point/maths/physics.

Check out those numbers... they might surprise you.

[Fascinate]

I think 15-20% per die shrink is acceptable. And 5ish on the tocks is okay. They need to do better though yes.

Well if that was right kaby should be 60% higher than sandy, as it sits we are seeing 22% lol.

- - - Updated - - -

Performance in the last 5 years has been focused on multiprocess/multithread and floating point/maths/physics.

Check out those numbers... they might surprise you.

Cinebench is a great test to compare apples to apples but sure maybe the other stuff they have gained a lot more in, but consumers dont really feel that.

[Chickat]

Performance in the last 5 years has been focused on multiprocess/multithread and floating point/maths/physics.

Check out those numbers... they might surprise you.

If only games benefited from that stuff as much as they should. Surely theres a way to make games benefit from at least multithreading. A lot of games still only really benefit form 2 cores let alone 4 or 8.

[schwarzkopf]

Cinebench is a great test to compare apples to apples but sure maybe the other stuff they have gained a lot more in, but consumers dont really feel that.

Well, depends on what kind of customers. If the customer is running 10 year old software, then probably not.

If they are running modern multi processing/multi tasking aware software - becomes quite noticeable.

Matter of fact - up to 11 times faster...

[Fascinate]

Well, depends on what kind of customers. If the customer is running 10 year old software, then probably not.

If they are running modern multi processing/multi tasking aware software - becomes quite noticeable.

Matter of fact - up to 11 times faster...

You may want to look a little closer at your "11x faster". That chart goes back to core 2 duo days, sandy was a massive jump and when you compare clock to clock from sandy to now intel has done basically nothing. Power consumption hasnt even came down on the performance desktop side, they have been in the ~90w TDP range the entire time.

[schwarzkopf]

If only games benefited from that stuff as much as they should. Surely theres a way to make games benefit from at least multithreading.

Multithreading is difficult to do well in live applications, because there is no room to do relevant stuff in secondary processing because the user can alter the requirements and does every moment.

The best you can do is parallel calculate some things, but that requires a rewrite of engines. So for example WoW will be stuck with its 4 thread max (or what ever it is) until it dies.

- - - Updated - - -

You may want to look a little closer at your "11x faster". That chart goes back to core 2 duo days

Was 10 years to 2015 (the zone covered by the red box in OP's link) ... but the main point of my argument was the detail of the article... that there are many factors at play.

[Chickat]

Hopefully unreal 5 and frostbite 2 etc the big engines coming in the next 5 or 6 years will finally use more of the cpu tech.

[Tehterokkar]

There hasn't been anything CPU intensive other than DICE's Battlefield series and some simulation games(eg. Cities Skylines, Ashesh of Singularity). So why would Intel try to go forward with CPU tech if nothing demands it?

[Nevcairiel]

Comparing an OC'ed CPU to a stock CPU seems rather dishonest, doesn't it. OC the 7600k to comfortable levels and compare those, or compare stock on both.
The 7600k is being named the new "overclocking king" afterall, just using stock boost clock isn't that.

[Fascinate]

Comparing an OC'ed CPU to a stock CPU seems rather dishonest, doesn't it. OC the 7600k to comfortable levels and compare those, or compare stock on both.
The 7600k is being named the new "overclocking king" afterall, just using stock boost clock isn't that.

No, that was explained above. Intel could have released the 2500k with the same clocks the 7600k has, just as they could have released the 7600k with the same clocks as the 7700k. Its just sleight of hand on intels part by manipulating clocks as they see fit to justify pricing increases. You need to normalize clock rates for this kind of comparison, and because all of these parts COULD have been released at the same speed out of the box the comparison is relevant.

[Speaker]

There hasn't been anything CPU intensive other than DICE's Battlefield series and some simulation games(eg. Cities Skylines, Ashesh of Singularity). So why would Intel try to go forward with CPU tech if nothing demands it?

I have a i5-3570k OC'd to 4.5 Ghz which runs at 100% load frequently in games like Overwatch, Diablo 3, WoW, and many others. Try running a few other programs in the background and the drop in FPS is quite noticeable. It's pretty disappointing. Can't wait for Zen.

[Ariktu]

The advantage for me is that my i5 3570k overclocked to 4.2 Ghz is still very nice in gaming so I can spend my money on other things like GPU upgrades.

[Fascinate]

I have a i5-3570k OC'd to 4.5 Ghz which runs at 100% load frequently in games like Overwatch, Diablo 3, WoW, and many others. Try running a few other programs in the background and the drop in FPS is quite noticeable. It's pretty disappointing. Can't wait for Zen.

What are you running in the background to be at 100% load in games like overwatch? I am playing OW right now with a twitch stream open and CPU usage does not go over 50%, hovering in the 40-45% range most times.

[Speaker]

No, that was explained above. Intel could have released the 2500k with the same clocks the 7600k has, just as they could have released the 7600k with the same clocks as the 7700k. Its just sleight of hand on intels part by manipulating clocks as they see fit to justify pricing increases. You need to normalize clock rates for this kind of comparison, and because all of these parts COULD have been released at the same speed out of the box the comparison is relevant.

I think he means that the newer chips overclock better than the older ones. So let's say the average 2500k would overclock to 4.5 Ghz, but the average 7600k overclocks to 4.9 Ghz, then on average the 7600k gains about an additional 10% increase in performance over the older chip. I doubt there is any concrete evidence on this however.

[Fascinate]

I think he means that the newer chips overclock better than the older ones. So let's say the average 2500k would overclock to 4.5 Ghz, but the average 7600k overclocks to 4.9 Ghz, then on average the 7600k gains about an additional 10% increase in performance over the older chip. I doubt there is any concrete evidence on this however.

Thats not true tho, most 2500k's can hit 4.8ghz, ive seen some hit 5.4ghz. That's part of why this comparison is relevant, they hit very similar max clocks.

[Speaker]

What are you running in the background to be at 100% load in games like overwatch? I am playing OW right now with a twitch stream open and CPU usage does not go over 50%, but hovers in the 40-45% range most times.

Nothing, but I do have a 144 hz monitor. So yeah, trying to push 144 fps is going to require more cpu usage. But in other games I barely get 60 fps and have a 100% usage.

- - - Updated - - -

Thats not true tho, most 2500k's can hit 4.8ghz, ive seen some hit 5.4ghz.

The numbers i made up were fake. I don't know what the average 2500k does, or the average 7600k. It was just supposed to be a theoretical explanation. No one knows because no has has run a large enough sample size to verify. And this is considering all things being equal, meaning comparing clockspeeds running the same thermals, using the same thermal compound, same heatsinks, same airflow. It's hard to compare these things.

I'm not trying to defend Intel in anyway. Even if the new chips did overclock slightly better, the performance gains since Sandy Bridge have been abysmal, and I hate Intel for milking the consumers in that way. I am just trying to say that comparing two chips at the same clockspeed isn't the whole picture since efficiency, thermal output, and overclockability also need to be taken into consideration.

Lastly, I am hopeful that AMD's Zen will shake things up a bit.