Intel Core i5-12600K

[Tsukiyomi91]

i5-12600K dethroned the 5600X in most benchmarks while maintaining a small gap with the 5800X, 5900X and 5950X. AMD is feeling the pinch now for being complacent.

 

[mechtech]

Yes, next week https://www.techpowerup.com/forums/threads/intel-core-i9-12900k.288509/post-4642317

awesome. Thanks

Will you also add the amd 5600g and 5700g to all the charts??

 

[dicktracy]

Basically no reason to get a 5600x or 5800x when both are overpriced right out of the gate. Intel dominates every single price point and you don’t need to participate in their mandatory AGESA beta-testing program .

 

[ncrs]

i5-12600K dethroned the 5600X in most benchmarks while maintaining a small gap with the 5800X, 5900X and 5950X. AMD is feeling the pinch now for being complacent.

You are comparing a completely new platform with an old platform. I don't see how you can say AMD is being complacent here. They can always slash prices, and we're still going to get the increased cache variants.
Intel winning here is the least they could do, to be honest I was expecting more (especially in the power department).

 

[W1zzard]

Will you also add the amd 5600g and 5700g to all the charts??

AMD has borrowed my samples "for 3-4 weeks", that was on October 6th But yeah, eventually I'll add them

 

[RandallFlagg]

i5-12600K dethroned the 5600X in most benchmarks while maintaining a small gap with the 5800X, 5900X and 5950X. AMD is feeling the pinch now for being complacent.

Actually it dethroned the 5800X in most benchmarks, and beats every Zen 3 in all 3 resolutions tested in games. From other review sites it runs quite well on DDR4 too, even DDR4-3200.

AMD is going to need to make the 5600X $200 and the 5800X $300 for them to make any kind of sense at all for a new build or significant upgrade. They would only make sense at current prices for someone who is just doing a CPU swap on a compatible motherboard.

 

[bobsled]

i5-12600K dethroned the 5600X in most benchmarks while maintaining a small gap with the 5800X, 5900X and 5950X. AMD is feeling the pinch now for being complacent.

I don't think AMD is being complacent, if anything, they're doing extremely well considering they're fighting for fab space and have a much lower budget than Intel.

Alder Lake was under design/testing long before the new CEO took the helm too - he'll no doubt take the credit regardless.

 

[HenrySomeone]

"Beats 5600x conclusively". Well, sure, in pure numbers with no other consideration; 14% overall performance increase in CPU tests with a multi-core increase of 50% power consumption (126w vs 189w). I'm sure each individual test varies, but it's not that impressive in relation to the power increase.

Nice reviews on all the new CPU's regardless!

It also beats the 5800x, at the same power efficiency. All that is left is to wait for more affordable motherboards to arrive (ddr4 ones for now) and it will be the undisputable champion of the mid-range (and better in single thread and gaming than any Ryzen, hehe)

 

[Dristun]

HWUnboxed said that DDR5 vs DDR4 difference is insignificant even against that ludicrous G.Skill kit so it's possible to count that out of equation in value comparisons. That leaves the motherboards and 100$ saving against 5800X handily covers the difference imo - wouldn't you rather be faster at the same price anyway? Can't wait to see how AMD responds with pricing and where Zen3D lands.

 

[HD64G]

So much so for the revolution that AL would bring...

To sum up, when you build a new AL system with a 12600K on DDR5 with a decent board you will spend more than building a system on AM4 with a 5800X that will use less energy, gain on average ~3% at cpu tasks and lose on average ~3% in gaming @1440P when using a high-end GPU. Where is the benefit to customers with the arrival of AL? I hope AMD is feeling threatened (I wouldn't) to lower the Zen3 CPU's prices by $50 for 5600X, 5800X and 5900X and by $100 for 5950X.

Anyone remembers the Intel test that showed them winning by much bigger % in gaming vs 5950X using the broken version of win11 for Ryzen CPUs?

Let's wait for Zen3D now to see if it will take back the gaming crown.

 

[RandallFlagg]

HWUnboxed said that DDR5 vs DDR4 difference is insignificant even against that ludicrous G.Skill kit so it's possible to count that out of equation in value comparisons. That leaves the motherboards and 100$ saving against 5800X handily covers the difference imo - wouldn't you rather be faster at the same price anyway? Can't wait to see how AMD responds with pricing and where Zen3D lands.

It's really not even $100. The problem is there are no midrange Z670 type socket 1700 boards. If you compare the lower end Z690 prices to X570 you're probably talking $50, and for that you get PCIe 5.0 with more unshared PCIe 4.0 lanes. It'll be a few months before we see those lower end boards.

This is a limited release for higher end enthusiast parts is what it boils down to. That is a negative, but only if you are not looking for higher end enthusiast parts.

 

[bug]

So much so for the revolution that AL would bring...

To sum up, when you build a new AL system with a 12600K on DDR5 with a decent board you will spend more than building a system on AM4 with a 5800X that will use less energy, gain on average ~3% at cpu tasks and lose on average ~3% in gaming @1440P when using a high-end GPU. Where is the benefit to customers with the arrival of AL? I hope AMD is feeling threatened (I wouldn't) to lower the Zen3 CPU's prices by $50 for 5600X, 5800X and 5900X and by $100 for 5950X.

Let's wait for Zen3D now to see if it will take back the gaming crown.

12600k isn't a high-end chip. Why would you put it in a DDR5 system?
The article is out already: you can pair this with a $250 DDR4 mobo even today. Though the ideal time to build a mainstream system is when the mainstream chipsets are launched, too.

 

[londiste]

@W1zzard, Single Thread Energy Usage based on SuperPi is an interesting point in this review - this is a single test result that seems to be very different from all the other single-thread test results when it comes to performance.

 

[thegnome]

Not worth a Z690 board and 12600K compared to the 5600x with a much cheaper motherboard. Will obviously be better with a B-series board but especially if AMD drops price it's barely worth it for gaming. For people doing more than just gaming I suppose it's a much better deal compared to the 5600x.

 

[HenrySomeone]

So much so for the revolution that AL would bring...

To sum up, when you build a new AL system with a 12600K on DDR5 with a decent board you will spend more than building a system on AM4 with a 5800X that will use less energy, gain on average ~3% at cpu tasks and lose on average ~3% in gaming @1440P when using a high-end GPU. Where is the benefit to customers with the arrival of AL? I hope AMD is feeling threatened (I wouldn't) to lower the Zen3 CPU's prices by $50 for 5600X, 5800X and 5900X and by $100 for 5950X.

Let's wait for Zen3D now to see if it will take back the gaming crown.

This is a straight out LIE!

 

[kruk]

It also beats the 5800x, at the same power efficiency.

But how? If you would read the review:

- It has worse single-thread efficiency than the 5800x
- It has the same multi-thread efficiency than the 5800x
- It's barely slower in CPU tests than the 5800x
- It's barely faster in gaming tests than the 5800x

It might have slightly better energy efficiency in fully threaded games and in fully threaded apps, but in everything else this is not true.

 

[bug]

@W1zzard, Single Thread Energy Efficiency based on SuperPi is an interesting point in this review - this is a single test result that seems to be very different from all the other single-thread test results when it comes to performance.

Probably one of the cases where the workload went to the E core

 

[HD64G]

12600k isn't a high-end chip. Why would you put it in a DDR5 system?
The article is out already: you can pair this with a $250 DDR4 mobo even today. Though the ideal time to build a mainstream system is when the mainstream chipsets are launched, too.

I said that it will end up being more expensive than AM4 platform. And even with a DDR4 board it will be more expensive. The truth of my post about the price difference doesn't change either way.

 

[R0H1T]

i5-12600K dethroned the 5600X in most benchmarks while maintaining a small gap with the 5800X, 5900X and 5950X. AMD is feeling the pinch now for being complacent.

AMD's biggest issue has been capacity, at TSMC, for at least the last 2 if not 3 years, including GPU's as well. Complacency has little to do with it, also no they're not feeling any pinch right now!

 

[Chrispy_]

Having read your reviews here and watched a couple of youtube reviews, it's clear that Intel have the lead in IPC when using DDR5, but can only achieve the wins at hideous power limits.

This 12600K is clearly where Intel shines this generation. Competitive power consumption and better performance across the board than the the 5600X for a similar CPU-only price.

What I'm waiting to see is the performance of the 12600K (and budget offerings, presumably 12500F etc) using DDR4 on B660 boards. As good as the 12600K is in this review, it costs as much as a 5900X, B550, and good quality DDR4-3600 kit just because the Z690 and DDR5 carry such a high premium that they utterly decimate the value proposition.

 

[Richards]

Slaps the 5600x & 5800x around and cheaper... i.m sure it makes 5600x & 5800x owners insecure about cores and performance (he he)

 

[W1zzard]

@W1zzard, Single Thread Energy Usage based on SuperPi is an interesting point in this review - this is a single test result that seems to be very different from all the other single-thread test results when it comes to performance.

Probably one of the cases where the workload went to the E core

I just checked for you, it gets scheduled on P cores, but bounces between two of them. I rather suspect that the new architecture isn't a best fit for the old x86 floating-point instruction mix of Super Pi

Maybe for the 2022 bench I can replace it with another 1T workload. Any suggestions? I rather not Cinebench but something else

 

[ncrs]

I just checked for you, it gets scheduled on P cores, but bounces between two of them. I rather suspect that the new architecture isn't a best fit for the old x86 floating-point instruction mix of Super Pi

Maybe for the 2022 bench I can replace it with another 1T workload. Any suggestions? I rather not Cinebench but something else

You are probably already aware of it, but the open-source Phoronix Test Suite also runs on Windows and has some ST-focused tests like simdjson and Ngspice. It would be nice to see some more Linux benchmarks on TPU as well

 

[W1zzard]

You are probably already aware of it, but the open-source Phoronix Test Suite also runs on Windows and has some ST-focused tests like simdjson and Ngspice. It would be nice to see some more Linux benchmarks on TPU as well

No plans for Linux and no plans for exotic tests that nobody has heard of

 

[ncrs]

No plans for Linux and no plans for exotic tests that nobody has heard of

Understandable, none of them are "mainstream", but Linux might, maybe, finally, perhaps, be having a shot at that with the Steam Deck.