jststojc:

Im sorry but the title is highly misleading! As you might know Power=Voltage times Current. What is listed here are Power and Current, but no Voltage.

The current listed is for the 12 volt psu output at the motherboard header, not the CPU core voltage, the PSU voltage is the same between 10nm and 14nm and all other CPUes made the last 10 years. If it was Vcore the amps would be much higher like: 100A * 1Volt = 100Watts. I am glad I ordered a 5600G instead of something like a i5 12400 for my mediacenter/server, if the Intel 35W TDP uses 20Amps * 12V = 240W peak. My passive cooled AM1 5350 25W TDP runs fine on a 65W laptop power brick, a 35W TDP Intel would blow the PSU immediately according to this.

cucaulay malkin:

so are big cores when they downclock,big whoop how many little low power cores does it take to match a regular big core ? can you provide your estimates please ? cause you seem very confident. show me the data then.

You're assuming the big cores will have the opportunity to downclock, or that a downclocked big core is still less power hungry than a little core under full load. Remember: Intel's TDP is measured by base clocks. ARM is where big.LITTLE started. The differences between ARM's big and little cores has got to be much more subtle compared to what Intel is doing, and yet the differences were still justified, even in non-mobile devices. The big.LITTLE approach is still being used in ARM, so clearly, it has value. It has so much value that Intel is doing it and AMD seems to be seriously considering it. So what really begs the question is where your confidence is coming from. Intel's market is primarily focused on Windows. The Windows task scheduler is notoriously bad, and yet, these companies (let's not forget Qualcomm, which has big.LITTLE WOS machines too) are investing in the technology anyway. That to me instills a lot of confidence that this approach is a necessary way forward for peak efficiency.

when I see that small cores can outperform big ones in power eficiency I'll believe they are useful. or are they only good in comparison to very inefficient big ones ?

Efficiency depends on the task, which is then determined by the task scheduler. If Windows decides to route Cinebench threads to the little cores, their performance-per-watt is bound to favor the big cores. Think of it like vehicles: A Prius is more fuel efficient than a V8 truck, but the Prius will most likely use up more energy and operate slower towing a boat than the truck. Meanwhile when there is no load, there is nothing you can do to make the truck more efficient than the Prius. Small cores are only efficient when you're not using them for heavy workloads. Big cores are inefficient when you try to make them do menial tasks.

Undying:

Its not true. I had Sandy-E 3930k before i went to amd again.

Exact same move I did. 3930k to 1500x. The 1500x was faster then the 3930k in gaming.

asturur:

at 60hz you are never at max load with any cpu of the last 4 years.

My 8 years old 4770K was glad to retire, it had lots of micro-stuttering due to 4 cores.

TheDeeGee:

My 8 years old 4770K was glad to retire, it had lots of micro-stuttering due to 4 cores.

When I was fixing a friends fiance computer, he has a 6700k. Even at 4.6 oc, it was pushed 85-90% load in all of the newer games I was testing. It had alot of micro stutter in very cpu intensive games. Need for Speed Heat was really bad for it too. It felt alot less smother then the r5 1600 @ 3.85 ghz which was around 65-75% load in her rig I had sitting next to it. Both were running the same GPU for testing. GTX 1080 at the time. 1080p 75hz panel. I didnt have FPS capped due to letting it push everything to the max. Capped at 75, the 6700k did alot better for smoothness but still, wasnt as good frametime wise as the 6 core.

I'm going for low powered 11th gen next month, probably i5 this time, maybe 11600k £230 can't complain got the mobo all updated for Rocket lake, call it a day and hope it lasts a while. The 6700k was quite impressive at the time, in TW3 for ex.

schmidtbag:

It has so much value that Intel is doing it and AMD seems to be seriously considering it.

no wonder they're doing it selling you a lesser product for more money,you think they'd miss the opportunity ? small cores are a waste of die space for enthusiast desktops.period. i'd rather not have to pay for that but seems like i'll have no choice.

Agonist:

When I was fixing a friends fiance computer, he has a 6700k. Even at 4.6 oc, it was pushed 85-90% load in all of the newer games I was testing. It had alot of micro stutter in very cpu intensive games. Need for Speed Heat was really bad for it too. It felt alot less smother then the r5 1600 @ 3.85 ghz which was around 65-75% load in her rig I had sitting next to it. Both were running the same GPU for testing. GTX 1080 at the time. 1080p 75hz panel. I didnt have FPS capped due to letting it push everything to the max. Capped at 75, the 6700k did alot better for smoothness but still, wasnt as good frametime wise as the 6 core.

same experience with 5775c,though only in some games. division2 ran great,wildlands ran great,bf1 too.no stutter,or close to none. odyssey was the worst,there was quite a lot of it.a little less in wd2,but it was still there at times.and it was pretty much pushed to the limit with 4.3G oc,2G oc on l4 cache and tuned ram. 1600 has other problems,not cpu load but purely fps numbers.especially the early revisions running under 4G and very limited in ram tolerance.

cucaulay malkin:

no wonder they're doing it selling you a lesser product for more money,you think they'd miss the opportunity ? small cores are a waste of die space for enthusiast desktops.period. i'd rather not have to pay for that but seems like i'll have no choice.

By that logic, why have a GPU? Why not just have a 500 core CPU that does everything? The whole point of different processors is to maximize performance and efficiency. If you want good performance, you don't want your big cores wasting clock cycles on background tasks. When they do waste clock cycles on such tasks and they're still clocked high, you're generating heat, which limits how long boost clocks can be sustained or how stable your OC is. For 99% of people, the small cores will offer more performance for less money, because they will allow CPUs to boost for longer and spend less time managing cache between menial tasks. These small cores will cost significantly less to produce. When it comes to desktop users, the point of having many cores is for multitasking, not for parallelization (that's what server CPUs, workstation CPUs, and GPUs are for). You don't need only big cores, because not all tasks warrant them. So for most people, 4 small cores with 6 big multi-threaded cores is going to be a cheaper, cooler, and faster option than 8 big multi-threaded cores, even though both will yield 16 total threads. So if you think it will be a waste of die space for enthusiasts, then you're buying the wrong CPU. I still can't tell if workstation or server CPUs will have little cores but even if they do, it'll be a small percentage. Those are what you'll want if you actually have a good reason to care about many big cores.

Just as those new power limit rules have come through for people in the state of California... guess those peeps won't be buying Alder Lake

schmidtbag:

By that logic, why have a GPU? Why not just have a 500 core CPU that does everything? The whole point of different processors is to maximize performance and efficiency. If you want good performance, you don't want your big cores wasting clock cycles on background tasks. When they do waste clock cycles on such tasks and they're still clocked high, you're generating heat, which limits how long boost clocks can be sustained or how stable your OC is. For 99% of people, the small cores will offer more performance for less money, because they will allow CPUs to boost for longer and spend less time managing cache between menial tasks. These small cores will cost significantly less to produce. When it comes to desktop users, the point of having many cores is for multitasking, not for parallelization (that's what server CPUs, workstation CPUs, and GPUs are for). You don't need only big cores, because not all tasks warrant them. So for most people, 4 small cores with 6 big multi-threaded cores is going to be a cheaper, cooler, and faster option than 8 big multi-threaded cores, even though both will yield 16 total threads. So if you think it will be a waste of die space for enthusiasts, then you're buying the wrong CPU. I still can't tell if workstation or server CPUs will have little cores but even if they do, it'll be a small percentage. Those are what you'll want if you actually have a good reason to care about many big cores.

yeah I don't need 8 mobile cores for background tasks. I don't need any mobile cores for background tasks. 12 big ones would deal with background tasks very well. and how do you know 4 small cores are faster than 2 big ones with HT ? imo they're prolly not even close tbh. the only reason they're pushing it is because 12/24 would need a nuclear reactor to power if 8/16+8 small ones are gonna need much more power than already power hungry rkl-s imagine this thing is already gonna pull 250-300w,maybe more.now replace the mobile cores with another 4 big ones. your points are valid for mobile devices,not for enthusiast desktops.

cucaulay malkin:

yeah I don't need 8 mobile cores for background tasks. I don't need any mobile cores for background tasks.

It isn't going to be just background tasks, because for most background tasks, a single core would be enough. Think of them more as supplements than replacements. You hardly hear anyone complain about Intel including an iGPU. Unlike the little cores, that's actually totally useless in a lot of desktop PCs. The little cores will most likely take up less package space and will actually improve the overall performance-per-watt. How is that a bad thing?

12 big ones would deal with background tasks very well.

You're missing the point here... The little cores can do the same tasks at a tiny fraction of the power consumption, while leaving the big cores to not be "distracted" with petty tasks. On Intel, cores boost for a finite amount of time, so you are effectively losing performance for every clock cycle that isn't being used on your task. On AMD, cores boost based on environmental conditions. If you have big cores focusing on menial tasks you're adding more heat than necessary, thereby losing performance. The fact of the matter is, x86 cores are getting bigger, hotter, and more expensive to manufacture. The vast majority of software out there doesn't need most of the transistors in a core to function. However, applications are becoming more multi-threaded and cross-platform (where many of the advanced instructions can't be used for the sake of compatibility). As long as you aren't doing major number crunching, rendering, or compiling, having only big cores isn't really a good option anymore.

and how do you know 4 small cores are faster than 2 big ones with HT ? imo they're prolly not even close tbh.

I didn't say they are faster; as you said, they're "prolly" not. But tasks that don't require 100% of a core's attention and don't require advanced instructions will [mildly] improve the performance of big cores. Two of the reasons I already mentioned: the big cores aren't wasting cycles on menial tasks, and, it keeps the whole CPU cooler, allowing for longer-lasting boosts. The other reason is because of the cache. The bigger the cache, the slower it gets. But as software and instructions get more complex, you need more cache to fit them in. Every time the CPU has to get new instructions for the cache, that is wasting time and potentially even bottlenecking the foreground app. By devoting certain tasks to the little cores, cache performance ought to improve. And since the little cores will probably have little caches, they might even perform some basic tasks faster than the big cores.

your points are valid for mobile devices,not for enthusiast desktops.

No it isn't. Like I said, unless you do a lot of number crunching, rendering, or compiling, you will not have a need for only big cores in the future. Of those 3 categories, I'm sure there will be options for only big-core CPUs. Go ahead and take a look at every program and game you run. Excluding the 3 categories I just mentioned, how many of them demand more 8 threads to run smoothly? Hypothetically, let's say you have a game that demands more. Well, not all threads of a game require big cores. You'll find that a lot of games that use every available thread do not proportionally load all of them, and that's because some of them are calculating totally different things. Or in another situation, maybe you have a game that demands 8 threads, but you also like to stream your games. Little cores can handle the streaming no problem, while lowering your overall power consumption and cost compared to a CPU with more big cores.

This is ridiculous!!! The cpu eats 40% of the energy of my water boiler 🙁

So no 16 big cores on this socket. Even the next something lake couldn't have 16 big cores under this conditions.

Agonist:

Exact same move I did. 3930k to 1500x. The 1500x was faster then the 3930k in gaming.

not likely if it's slower than 2600k https://www.purepc.pl/test-procesorow-amd-ryzen-5-1500x-vs-intel-core-i5-7500?page=0,31 in every game in the review

cucaulay malkin:

not likely if it's slower than 2600k https://www.purepc.pl/test-procesorow-amd-ryzen-5-1500x-vs-intel-core-i5-7500?page=0,31 in every game in the review

The benchmark shows i5 7600 is one of the fastest cpus. It must be very enjoyable playing some battlefield on a quad core without ht. :P

Undying:

The benchmark shows i5 7600 is one of the fastest cpus. It must be very enjoyable playing some battlefield on a quad core without ht. 😛

it includes avg. fps only so yeah it could be possible,despite the actual experiece hub shows the same on bf5,7600k doing 150fps average,they're fake news too,right ? [youtube=5VnRUFz-m0M] that said, an oc'd 3930k w. qc ram would've crushed it.maybe vs. stock sandy running 3.7G on 1333 ram it's possible for 1500x to be faster,but I don't think in any other scenario. 3930k to 2700x is a sidegrade.3930k to 1500x is a major downgrade.

cucaulay malkin:

it includes avg. fps only so yeah it could be possible,despite the actual experiece hub shows the same on bf5,7600k doing 150fps average,they're fake news too,right ? [youtube=5VnRUFz-m0M] that said, an oc'd 3930k w. qc ram would've crushed it.maybe vs. stock sandy running 3.7G on 1333 ram it's possible for 1500x to be faster,but I don't think in any other scenario. 3930k to 2700x is a sidegrade.3930k to 1500x is a major downgrade.

1% lows are bad on quad cores and it shows. 7600k running 5ghz isnt smooth as r5 1600.

Undying:

1% lows are bad on quad cores and it shows. 7600k running 5ghz isnt smooth as r5 1600.

yup but 4/4 is not 6/12

cucaulay malkin:

not likely if it's slower than 2600k https://www.purepc.pl/test-procesorow-amd-ryzen-5-1500x-vs-intel-core-i5-7500?page=0,31 in every game in the review

That review uses an overclocked 2600k. Base clock is 3.4ghz with turbo 3.5ghz, not 4.5ghz.

cucaulay malkin:

yup but 4/4 is not 6/12

7700k and 2700x is the same story. Despite average fps i bet ryzen feels smoother in many recent games. 8700k is faster tho.