Tom's Hardware i7-7700K OC Preview

[IntelUser2000]

The 'new' 14nm+ process appears to be extremely disappointing. Power consumption is through the roof.

I suspect this 14nm+ process 'improvement' is a marketing gimmick more than anything else, simply due to them not being able to continue with the tick tock model Any changes from 14nm to 14nm+ would probably not have been mentioned in previous years (such as 22nm improvements, from 4770k to 4790k etc).

We've been spoiled with breathtaking advancements in computers, for so long. Now, such times are coming to an end. I can see that such drastic changes are hard to accept.

Regarding process, part of the reason they parade smaller advancements with greater fanfare is because it likely would have took much effort to reach that point. Remember when with GPUs we got "free" clocks along with big increases in shader count? And it might use less power? With Pascal Nvidia was saying they needed dedicated design to optimize for frequency.

We still get nice gains, but that's only on lower power CPUs. The U chips definitely got its +12% boost. Process is just one part of a complicated formula. On its heyday, semiconductors got that much easy, but that's times past. There was never a semiconductor brick wall, just a sludge that's getting increasingly tougher to wade through as progress happens.

 

[guskline]

I picked up a 6700K for one of my systems recently, and I have been spending some time overclocking it.

I simply can't get beyond 4.7GHz ASUS RealBench stable, even when I force-feed the thing 1.45V. It's not a temperature/core problem, since the core temperatures don't go higher than, say, 80-85C.

I had this same issue with the other 6700Ks that I had in the past...4.7GHz with lots of volts was the limit, and I could forget about 4.8GHz or beyond.

If 7700K can do 4.8GHz+ regularly, especially at 1.3v, that'd be really nice.

Agreed. I had a custom water cooler set up for my 6700k but found it rock solid at 4.6Ghz and too hot and not stable at AsusRealbench running at 4.7Ghz while using a Corsair h110iGT a pretty high end AIO.

Looks like the 7700k can run solid at 4.8. Incremental increase but not bad for a "polished and refined" skylake core. PS, my 6700k runs like a champ. Waiting to see what if anything, Zen brings to the table.

 

[coercitiv]

I'm not sure why anyone was expecting KL to be revolutionary.

The real thought twister is that if Kaby Lake were a new arch bringing a flat 10% performance increase over Skylake, with the same max frequency and same power usage as Skylake, it would have received a far less hostile welcoming committee. Because everybody knows improving max frequency is far easier than improving IPC. Right? Right?!?

Look at Broadwell, the son Intel wishes never had, but was far too late in gestation to abort: it will still have a better welcome by comparison with the Kaby Lake pedigree breed. Of course six months later everybody plus dog will race their forum signatures to 5Ghz+, but let's not get ahead of ourselves and patiently wait for the Zen showdown.

 

[ddogg]

I picked up a 6700K for one of my systems recently, and I have been spending some time overclocking it.

I simply can't get beyond 4.7GHz ASUS RealBench stable, even when I force-feed the thing 1.45V. It's not a temperature/core problem, since the core temperatures don't go higher than, say, 80-85C.

I had this same issue with the other 6700Ks that I had in the past...4.7GHz with lots of volts was the limit, and I could forget about 4.8GHz or beyond.

If 7700K can do 4.8GHz+ regularly, especially at 1.3v, that'd be really nice.

Same here. I have a monoblock on mine and 4.7 seems to the sweet spot @~1.3V while keeping temps around 65C max. Bumping it to 4.8 requires me to push it close to 1.4V and the temps really start to skyrocket.

 

[RussianSensation]

I'm not sure why anyone was expecting KL to be revolutionary. It was never going to be anything but an optimization.

This was Anandtech's August piece on KL:

That's been my beef with some posters on this forum. They kept hyping up KBL, but it was confirmed a long time ago that IPC and architecture are identical. Many other sites incorrectly reported on KBL since the 12% improvement was related to transistors, not CPU performance/benchmark increases.

Let's say in a best case scenario a golden sample 7700K overclocks a to 5.1Ghz, a great 6700K hits 4.8Ghz. That's only a 6.25% increase. 5Ghz 7700K vs. 4.7Ghz 6700K is once again only a 6.38% increase. It's been a rule of thumb that less than a 10% increase in CPU speed is not noticeable by the end user without benchmark scores.

Moving onto the Z270 motherboard, again no major improvements. 24 PCIe lanes do little to change much compared to 20. Optane support is practically useless since Optane was delayed to late 2018:

"Intel has revealed that there will be a considerable delay in the arrival of 3D XPoint memory, and the blazingly fast modules apparently won’t pitch up until late 2018 at the earliest."
http://www.techradar.com/news/intels-superfast-memory-tech-gets-hit-by-a-major-delay

Even if Optane shows up earlier, knowing Intel's rip-off SSD/PCIe SSD prices, it will cost a lot of $ and have gimped capacities.

Next year 6-10 core Skylake doesn't excite since the planned launch is 2H 2017, by which point Skylake architecture is 2 years old. Historically, the workstation platform first led the mainstream architecture by a full year (920 vs. 860 Lynnfield), then it started to lag behind with Sandy (2600K vs. 3930K). Following Ivy, the delay started to increase more and more Skylake-X will be more or less 2 year "late."

I keep telling you guys, there will be nothing exciting out of Intel until 6-core Coffee Lake drops into laptops and Ice Lake on the desktop. Everything before that will be slightly improved iterations of the same Skylake architecture. 6700K users will have enjoyed their processor for close to 3-3.5 years before Ice Lake shows up.

Considering how Z170 has all the key features already, it will require a new chipset with DMI 4.0, PCIe 4.0, M.2 with PCIe 3.0 x8 minimum, next gen sound card chipsets, built-in WiFi to make upgrading worthwhile.

Unless game start moving the way I'd BF1 in droves, and people upgrade to 144-240Hz monitors, it doesn't look like 6700K is going to be a slow CPU even in 2018. I can easily see myself and others even skipping Ice Lake just like many did a 2 generational architecture jump moving from Sandy/Ivy to Skylake. Skylake to Ice Lake is just a 1 generation jump which won't be as good as going Sandy -> Skip Haswell -> Skylake. Ice Lake looks better suited for Haswell 4770/4790K users.

I keep telling you guys if a user bought a brand new i7 Intel architecture close to launch date, then this Intel CPU now lasts 4-5 years. With Intel moving from 2-2.5 year Tick-Tock to a 3-year Process-Architecture-Optimization cycle, this is looking to change to 6 years for anyone who will want a similar jump from Skylake as moving from Sandy was to Skylake.

Possibly next generation PS5/XB2 consoles, if powered by a 6C/12T or 8C/16T Zen could finally start a new generation that obsoletes a 4C/8T 6700K. I don't expect those consoles to launch until 2019-2020, at which point it should still take developers 1-2 years before they start coding games regularly for 8-16 threads. That's why I tentatively have 2018 Ice Lake and 2021 "Ice Lake 2" as my potential upgrade points. This all assumes Intel even hits those launch dates as I smell Intel delays, after already delaying Broadwell, Broadwell-E, Skylake-X. If Cannon Lake shows up Q1 2018, it's possible we won't even see Ice Lake until Q1 2019.

For budget users who cannot step up to an i7, I'd strongly consider getting an Asrock Z170 and an i5 before Intel completely closes the BLCK overclocking. In 2017 a $170-180 i3 7350K will look laughable compared to a $170-180 i5 6400 @ 4-4.5Ghz.

For those who have a MicroCenter near you, $259 6700K is a smoking deal. eBay $275 6700K is also a go.

7700K will cost $329-349 and offer almost nothing except an irrelevant clock speed increase and 4K Netflix streaming support.

I don't mind these trends though since it leaves $$$ for monitor and/or GPU upgrades. It also means people buying an i7 have now about 5-6 years before a meaningful upgrade shows up. Since there are a lot of Sandy/Ivy users still around, clearly even a 25-40% boost in CPU speed isn't enough for many. It might hard for even a 2021 Ice Lake 2 to achieve single core boost of 25-40% over a 6700K.

 

[RussianSensation]

Same here. I have a monoblock on mine and 4.7 seems to the sweet spot @~1.3V while keeping temps around 65C max. Bumping it to 4.8 requires me to push it close to 1.4V and the temps really start to skyrocket.

The blower myth keeps persisting in the GPU world, and the CPU temp myth keeps haunting the CPU world.

Modern Intel CPUs work up to 90-100*C without thermal throttle. Haswell went all the way to 100C:
https://www.google.ca/amp/www.techspot.com/amp/article/927-temperature-impact-cpu-performance/

I have been running my Ivy i7 3630QM 3.4 Ghz in a laptop at 92-93C since February 2013. By the time an i7 6700K dies from heat if ran at 80C, it will be unsuitable for next gen GPUs.

There is no longevity difference running Skylake at 65C or 80C.

I see many confusing Intel's TCase of 64C with max operating CPU die temperature. They aren't the same and the die can go much higher than 65C.

TCASE
Case Temperature is the maximum temperature allowed at the processor Integrated Heat Spreader (IHS).

 

[Arachnotronic]

That's been my beef with some posters on this forum. They kept hyping up KBL, but it was confirmed a long time ago that IPC and architecture are identical. Many other sites incorrectly reported on KBL since the 12% improvement was related to transistors, not CPU performance/benchmark increases.

This boggles my mind, too. Intel flat out said that it's the same Skylake micro-architecture, yet people were looking for IPC increases.

Let's say in a best case scenario a golden sample 7700K overclocks a to 5.1Ghz, a great 6700K hits 4.8Ghz. That's only a 6.25% increase. 5Ghz 7700K vs. 4.7Ghz 6700K is once again only a 6.38% increase. It's been a rule of thumb that less than a 10% increase in CPU speed is not noticeable by the end user without benchmark scores.

6.25% performance improvement, to be fair, is better than what you'd get with a typical "tick." Ivy Bridge perf/clock was up, what, 5%? And 3770K shipped at the same clocks as the Core i7 2700K.

Moving onto the Z270 motherboard, again no major improvements. 24 PCIe lanes do little to change much compared to 20. Optane support is practically useless since Optane was delayed to late 2018:

"Intel has revealed that there will be a considerable delay in the arrival of 3D XPoint memory, and the blazingly fast modules apparently won’t pitch up until late 2018 at the earliest."
http://www.techradar.com/news/intels-superfast-memory-tech-gets-hit-by-a-major-delay

Be careful here. Optane SSDs coming in 2017, it's just the 3DXP DIMMs for Purley that won't show up until late 2018. Here is BK on Optane SSDs:

Just a correction, the team here just caught me on. I think I had my years off. So, on the 3D XPoint it will be qualified at the end of this quarter. And we’re shipping thousands of samples to customers, we’re shipping samples already, we’ll ship thousands through this quarter. And it ramps in 2017, I think I said by mistake 2018, I’m sorry I’m just too many years that were talking through here. So, ramp in ‘17, revenue growth in ‘17, samples, thousands in the fourth quarter and qualified at the end of the quarter. Sorry for that confusion.

Even if Optane shows up earlier, knowing Intel's rip-off SSD/PCIe SSD prices, it will cost a lot of $ and have gimped capacities.

Yeah, it ain't gonna be cheap.

I keep telling you guys, there will be nothing exciting out of Intel until 6-core Coffee Lake drops into laptops and Ice Lake on the desktop. Everything before that will be slightly improved iterations of the same Skylake architecture. 6700K users will have enjoyed their processor for close to 3-3.5 years before Ice Lake shows up.

Some people seem to really want more cores on mainstream. If you were rocking a SNB/IVB/HSW quad (especially i5), then a hex-core mainstream Coffee Lake using KBL architecture could be a nice step up.

Considering how Z170 has all the key features already, it will require a new chipset with DMI 4.0, PCIe 4.0, M.2 with PCIe 3.0 x8 minimum, next given sound card chipsets to make upgrading worthwhile.

Z270 is a super modest upgrade, we'll see what Z370 (Coffee Lake chipset) brings. But I agree, I think Ice will deliver the "big bang" on the platform side.

I don't mind these trends though since it leaves $$$ for monitor and/or GPU upgrades. It also means people buying an i7 have now about 5-6 years before a meaningful upgrade shows up. Since there are a lot of Sandy/Ivy users still around, clearly even a 25-40% boost in CPU speed isn't enough for many. It might hard for even a 2021 Ice Lake 2 to achieve single core boost of 25-40% over a 6700K.

Intel says that the upgrade cycles in enthusiast/gaming are around 2-3 years, much shorter than typical PC upgrade cycles, FWIW.

 

[RussianSensation]

^ The vast majority of gamers on this forum don't upgrade the CPU platform every 2-3 years. Do you have data to back up those Intel claims?

If what you said were true, nearly everyone with Sandy/Ivy/Haswell would own a 6700K or X99. We are seeing the complete opposite -- most aren't gaming on those platforms. Steam users are even worse as their hardware lags far behind that of users on tech forums.

 

[arandomguy]

Some of us skipped Sky Lake and rolled the dice on Kaby Lake. I'm one of those and I'm rather regretting that decision. The problem now is any ugprade is also in a weird gap before Coffee Lake and Ice Lake which should be much bigger.

6.25% performance improvement, to be fair, is better than what you'd get with a typical "tick." Ivy Bridge perf/clock was up, what, 5%? And 3770K shipped at the same clocks as the Core i7 2700K.

Ivybridge had lower power consumption. The platform upgrade was also larger with PCIe 3.0 and native USB 3.0. There was a large GPU upgrade.

Ivybridge also was poorly received due the the downgrade from solder to TIM resulting in temperature issues.

It also released in a complete vacuum due to Bulldozer. Granted post Zen launch we could be looking at Kaby Lake much more favorably due to poor results as well.

 

[LTC8K6]

https://forums.anandtech.com/thread...-insights-benchies-overclocks-inside.2493250/

KL looking pretty good there.

 

[RussianSensation]

This boggles my mind, too. Intel flat out said that it's the same Skylake micro-architecture, yet people were looking for IPC increases.

Ya, I don't understand it myself. Although I'll wait for a real world experience from SpeedShift 2.0 to see if that makes a difference too.

6.25% performance improvement, to be fair, is better than what you'd get with a typical "tick." Ivy Bridge perf/clock was up, what, 5%? And 3770K shipped at the same clocks as the Core i7 2700K.

True.

On the desktop, I personally prefer upgrading on new achitecture cycles, not refresh cycles. KBL is a refresh to me. In the past, it would be equivalent to a bump from an i7 750 -> 760 or i7 860 to 870/875K/880. If it weren't for the GPU improvements, the architectural improvements Intel calls a "new generation" now were merely a new SKU in the past.

i7 6700K -> i7 6750K KBL
i7 6600K -> i5 6650K KBL

Be careful here. Optane SSDs coming in 2017, it's just the 3DXP DIMMs for Purley that won't show up until late 2018. Here is BK on Optane SSDs: Yeah, it ain't gonna be cheap.

Thanks for the correction. As almost all cutting edge storage tech, it's often better to wait 1-2 revisions/generations. PCIe M.2 SSDs improved a lot and it wasn't until 950 Pro series and 960 Evo that these became good value for the high-end.

Some people seem to really want more cores on mainstream.

Which is weird to me since relatively affordable i7 5820K was available August 2014, and it overclocked to 4.4Ghz.

If you were rocking a SNB/IVB/HSW quad (especially i5), then a hex-core mainstream Coffee Lake using KBL architecture could be a nice step up.

Maybe, but those are "no man's land upgrades". I already explained why that is earlier. A Sandy/Ivy/Haswell user that skipped Skylake would still end up with the same underlying architecture as Skylake should they get Kaby Lake, Skylake-X, Coffee Lake or Cannon Lake. If someone skipped i7 6700K and is using an i7 2600K-3770K-4770K-4790K, might as well skip everything and go for Ice Lake or Ice Lake-X. That's why I sold Sandy a while back since I knew this stagnation was coming. If someone wants more cores for workstation/productivity, then 8C/16T Zen is likely to be better than a 6-core Coffee Lake or a 6-core Skylake/KBL-X.

Z270 is a super modest upgrade, we'll see what Z370 (Coffee Lake chipset) brings. But I agree, I think Ice will deliver the "big bang" on the platform side.

Yup. At least Asus finally updated Z270-A to include dual M.2 ports, higher DDR4 speeds and better sound. The old, but highly popular Z170-A was an overpriced board, gimped on features and DDR4 overclocking headroom.

Some of us skipped Sky Lake and rolled the dice on Kaby Lake. I'm one of those and I'm rather regretting that decision. The problem now is any ugprade is also in a weird gap before Coffee Lake and Ice Lake which should be much bigger.

That's exactly how I feel. Same reason I am not at all excited for a 6-10 core Skylake-X or 4-core Kaby Lake-X processors. It's going to be ~ 2 years since Skylake architecture came out when Intel will give us a 6-10 core Skylake-X, and yet most of those cores won't be utilized in games. But wait for it, mark my words, if Zen 6/12T or 8/16T perform well, all of a sudden, 4C/8T CPUs will be claimed to be worthless and outdated.

IMO, the best time to upgrade in the Process-Architecture-Optimization cycle is on the architecture node. You get the latest and greatest for at least 3 years until the next architecture node. Anyone who skipped Skylake in 2015-2016 is going to be disappointed by every single Intel processor released on the desktop until Ice Lake (and will that even live up to the hype?), unless they actually need MOAR cores.

 

[RussianSensation]

https://forums.anandtech.com/thread...-insights-benchies-overclocks-inside.2493250/

KL looking pretty good there.

True, but he says "NOTE: This hasn't been stress tested for stability, I just booted straight to the desktop and ran the CR15 MT bench."

Spoiler: Here is another user with 5Ghz overclock @ 1.376V on an MSI Carbon board

http://wccftech.com/intel-core-i7-7700k-kaby-lake-benchmarks-leak/

It would be great if most 7700K's could hit 5Ghz reliably as opposed to a few golden samples achieving it. If true, then the +200mhz Toms got is going to translate to a 300-400mhz increase, a much better outcome.

 

[Arachnotronic]

On the desktop, I personally prefer upgrading on new achitecture cycles, not refresh cycles. KBL is a refresh to me. In the past, it would be equivalent to a bump from an i7 750 -> 760 or i7 860 to 870/875K/880. If it weren't for the GPU improvements, the architectural improvements Intel calls a "new generation" now were merely a new SKU in the past.

i7 6700K -> i7 6750K KBL
i7 6600K -> i5 6650K KBL

You aren't going to get an argument from me here in terms of delivered performance. But I just think it's becoming so hard to wring out more frequency, especially at the >4GHz stock speeds that Intel is shipping processors at, that Intel need to do a lot more to get those boosts without blowing out the power budget.

Thanks for the correction. As almost all cutting edge storage tech, it's often better to wait 1-2 revisions/generations. PCIe M.2 SSDs improved a lot and it wasn't until 950 Pro series and 960 Evo that these became good value for the high-end.

Yep, bleeding edge tech is always really expensive and can have teething issues.

Which is weird to me since relatively affordable i7 5820K was available August 2014, and it overclocked to 4.4Ghz.

X99 is more expensive a platform, though, which might be what has people avoiding that. You can get some really nice and cheap Z170 boards.

IMO, the best time to upgrade in the Process-Architecture-Optimization cycle is on the architecture node. You get the latest and greatest for at least 3 years until the next architecture node. Anyone who skipped Skylake in 2015-2016 is going to be disappointed by every single Intel processor released on the desktop until Ice Lake (and will that even live up to the hype?), unless they actually need MOAR cores.

It really depends on the software landscape. If consumer software starts using more cores, then "moar coars" becomes a viable strategy for future generation mainstream Intel CPUs. If it doesn't, then most of them sit there useless.

 

[daxzy]

This boggles my mind, too. Intel flat out said that it's the same Skylake micro-architecture, yet people were looking for IPC increases.

It's probably optimized for mobile. Reading some KabyLake-U laptop reviews, they seem to offer better battery life in normal usage (at higher base clocks). Of course their updated HVEC means that their video watching battery time is way higher.

For Desktop/Server, its a non-issue.

 

[Ranulf]

"Which is weird to me since relatively affordable i7 5820K was available August 2014, and it overclocked to 4.4Ghz."

Uh, ok RS. I don't remember seeing that many great examples of 4.4 for a 5820k. X99 was worth it if you needed 6 cores and were willing to take the time to OC but otherwise, a waste of cash.

 

[Lepton87]

Uh, ok RS. I don't remember seeing that many great examples of 4.4 for a 5820k. X99 was worth it if you needed 6 cores and were willing to take the time to OC but otherwise, a waste of cash.

Agreed, 4.4GHz is not a typical overclock for the 5820K.

 

[lolfail9001]

Agreed, 4.4GHz is not a typical overclock for the 5820K.

On early batches or low voltages, maybe. Last batch before BDW-E was averaging like 4.7Ghz at about 1.3V i believe.

 

[2is]

the milking won't stop until AMD provides something actually competitive. Intel has no motivation to innovate currently.

AMD really has little to do with this. The problem Intel is facing is economics and necessity. The R&D costs are a lot higher with these smaller processes while generational benefits are less and less. This results in people buying your product less often then they would have otherwise. This is compounded by the fact that, overall, there is less demand today for x86 processors then there was 5-10 years ago.

I don't think a competitive AMD would change this a whole lot. You still need the general public lining up to buy your product and I just don't think the demand is there anymore. Just look at the members of this and similar forums. How many of us are still on Sandy Bridge? I'm personally still on Ivy Bridge, which is 4.5 years old now and I demand more of my CPU then your average person does. When I first started building my own computers I was upgrading the CPU every year or two. From the looks of it, this one will last me at least 7

 

[cytg111]

AMD really has little to do with this. The problem Intel is facing is economics and necessity. The R&D costs are a lot higher with these smaller processes while generational benefits are less and less. This results in people buying your product less often then they would have otherwise. This is compounded by the fact that, overall, there is less demand today for x86 processors then there was 5-10 years ago.

I don't think a competitive AMD would change this a whole lot. You still need the general public lining up to buy your product and I just don't think the demand is there anymore. Just look at the members of this and similar forums. How many of us are still on Sandy Bridge? I'm personally still on Ivy Bridge, which is 4.5 years old now and I demand more of my CPU then your average person does. When I first started building my own computers I was upgrading the CPU every year or two. From the looks of it, this one will last me at least 7

There is one thing though, this struggle to get this one arch anywhere, you got Core ranging from 2C/4t tablets in ~3 watts to desktop class 4C/8T ~95watts to HEDT ~140watts. Maybe they're stretching it a bit thin and maybe the design decisions made both on the arch and the process node hurts potential peak performance somewhat. One core to rule them all may be too thin of a gamble come competition.

 

[beginner99]

Look at Broadwell, the son Intel wishes never had, but was far too late in gestation to abort: it will still have a better welcome by comparison with the Kaby Lake pedigree breed. Of course six months later everybody plus dog will race their forum signatures to 5Ghz+, but let's not get ahead of ourselves and patiently wait for the Zen showdown.

Well the 5775c actually had something going for it, namely the L4 cache /eDram which had some major benefits in some games. Kaby Lake has nothing going for it over Skylake. I was in the market for a new chip but all we know is pretty meh and I rather go Skylake-X and hexacore.

 

[coercitiv]

Well the 5775c actually had something going for it, namely the L4 cache /eDram which had some major benefits in some games. Kaby Lake has nothing going for it over Skylake. I was in the market for a new chip but all we know is pretty meh and I rather go Skylake-X and hexacore.

How many sites/reviewers/forumists did you see recommend 5775c as a worthy upgrade over Devil Canyon? The L4 advantage turns into thin dust once high performance memory is being used, at least as far as gaming is concerned, so it's way more effective in mobile, where memory usually comes in high latency & low frequency flavors, but in desktops the combination of high pricing and lower scaling simply didn't add up. The lower max clocks and poor BIOS support didn't help either.

The L4 really did open some eyes, and I strongly criticized Intel for keeping this away from the desktop platform as it showed enormous potential in some apps / configurations, but it turns out Intel wanted consumers to pay a premium for eDRAM, and both consumers and OEMs laughed in their face and said pass. To this day OEMs still bundle small&crappy dGPUs in cheap laptops instead of Intel chips with L4 cache, and that is a sad story indeed.

But I digress, let's get back to the topic, the disappointment that is Kaby Lake. Are people already turning to the dark side of 5Ghz overclocking?

 

[LTC8K6]

KL desktop with edram and higher clocks than BW would probably be interesting in games.

 

[TheELF]

But I digress, let's get back to the topic, the disappointment that is Kaby Lake. Are people already turning to the dark side of 5Ghz overclocking?

Might be disappointing for the high end but for the low end KL is awesome sauce, prices for official OC dropping by ~$100 (even if it's just an i3) i3 even getting oc,prices for HT dropping almost by half if the ~$70 pentium will actually end up having it,getting the hardware acceleration to decode multiple 4k videos at once or even encoding hevc while decoding other streams and so on and all on low tier CPUs.
Suddenly you get a whole lot more for your money.

 

[krumme]

The only people that should be interested in increasing IPC is shareholders be it IBM or Intel, and even AMD.
The only purpose it have is monopolizing the market by keeping the software from using more threads and making it more expensive for entry by new competitors and competing products. Classic top management strategy.
Its a shame. And what we have now is just pathetic improvement of performance per year. Like 10% improvement what we had 10 or 25 years ago. And ofcource at high prices. Pretty obvious looking at the margins and record profit.
An Arm a73 cpu core is 0.65mm2 on 10nm. It just shows how idiotic this monopoly race for IPC have been. Next step 5 wide x86 cores used with AVX 512 bit. Way to go for the shareholders.
Unfortunately we had the crappy bd core, and Moar cores is still parroted. But what a bunch of selfinflicted pain it is.
Looking in hindsight a oc sandybridge perf is about excactly the same performance than a oc KL. Same stuff with a lot of PR lipstick.
Go 10 or 20 years back and people expected more from 6 months progress. Not 6 years.
X86 is now a ugly cashcow with pathetic performance for the customers instead of beeing a place of progress and innovation.

 

[TheELF]

The only purpose it have is monopolizing the market by keeping the software from using more threads and making it more expensive for entry by new competitors and competing products.

We have quad cores for ~10 years now,what exactly is keeping the software from using at least these 4 threads?