Intel Core i7-12700K Review
tomshardware.com
tomshardware.com
Sacrificing ~5% "game performance" (such a nebulous number given how varied the CPU/GPU load can be between games) for being able to sit comfortably in the same room seems like a no brainer.
This still feels like a halo product? Except does the i5 or such compete well enough with Ryzen for that to work?
I'm curious to see how AMD responds, I wonder if they're chiplet method lends itself well to launching their own hybrid E/P core architecture alongside "3D V-cache".
It seems insane but if true then consider that the combination of CPU and GPU is going to be 840W, and that is before the rest of the system is taken into account.
At some point we are going to need to start worrying about installing dedicated venting for computers to the outside of the house.
At some point you need to start integrating desktop computers into your HVAC designs.
At some point you need to start having dedicated electric circuits installed like for your oven.
My current computer (5950X with 3080) was already hot enough when gaming that I found it physically uncomfortable to have it under my desk and had to move it out from under there. My legs were burning up.
By running it at 1.1V it reaches 90°C hotspot, below the 110°C threshold so it runs at maximum frequency constantly.
Many GPUs (and CPUs) have their stock voltage quite high because it improves yield for the manufacturer. Unless a GPU is really bottom of the barrel regarding silicon lottery, it can usually be undervolted without issues. This is however less true for flagship models which are already pushing the silicon as far as it can.
The higher voltages don't improve yields since manufacturers usually bin parts by testing them at higher voltages, decreasing them until the part passes the whole test suite for a grade.
The primary pressure on stock voltage is headline numbers like core count and clock speed that consumers have been trained to compare by decades of marketing. This is especially obvious with laptops and desktops geared at gamers, which will often come with an even higher voltage than the recommended stock Intel profile at a detriment to performance and efficiency - doing it just to be able to list higher numbers on the specs page at the expense of the usability of the machine.
Point in case, I have a newly built Ryzen 5900X tower that reuses a GPU from an old build. I chose a high quality 80+ Platinum 750W PSU thinking that would give more than enough headroom for upgrading the GPU at some point, but if the next generation of GPUs are as power hungry as rumored I’ll be limited to current generation cards unless I buy a new PSU.
I hope efficiency comes to be a focal point again soon, because if the trend continues you’ll need a 1kW PSU just to comfortably accommodate current enthusiast parts and have any hope of upgrading in the future, which is ridiculous.
Even so, 1KW+ personal computers are a bit strange, back in the day the very largest workstations (think IRIX fridges) were in that domain.
What would be nicer would be if manufacturers aimed for minimal power consumption with the same performance as a previous generation to give people the option.
It took us almost 10 years to move the standard from 1024x768 to 1920x1080 (~2.5x increase in pixels/sec), yet in the past 5 years we've gone from 1920x1080@60Hz to 4k@120Hz (~8x increase).
I'm not sure if it's because graphics aren't naturally improving the same way they were in the 00s and early 2010s, or streaming culture showing things off with the highest resolutions and framerates possible, but it's absolutely a modern phenomenon and I'm not sure how long it can last.
I think the current increase in interest in high refresh rate gaming came from the shift away from TN panels. Once IPS became reasonably available, companies with TN products couldn't push for them as premium products from an image quality perspective, but high refresh rate was something they could still market with TN that there was a window where IPS and VA could not do that. So they did, gamers got exposed to high refresh rates and then moved those expectations to other products.
Also tech like gsync/freesync meant that if your monitor had a higher refresh rate than your hardware could produce, you no longer had to deal with tearing.
I think monitor tech is leading GPU tech at the moment, with 1440p240 and 4k120 monitors. There's certainly _some_ games where those resolution/framerate combinations are achievable, mostly esports titles and older/smaller titles, but for the most part, no (also for me at least, past like 90hz you're into dimishing returns territory). But now people have these monitors capable of X resolution and Y refresh rate, people want to have their cake and eat it, wanting games that have modern effects for higher fidelity and yet also higher fps to take advantage of their monitors.
Yeah, I think you're actually right there. People talk about high-end monitor technology much more than they used to, but I'm not sure how many ordinary people actually have them.
>Once IPS became reasonably available, companies with TN products couldn't push for them as premium products from an image quality perspective.
That's really interesting. I recently changed monitors from 4k60 to 1440p/144Hz under the assumption that the increased frequency would make more of a difference than the additional resolution. Immediately noticed an inferior quality picture (less pixels aside) - perhaps that's why!
I think 4k@120 monitors are entering more people's price range and many will choose one on their next upgrade cycle.
Many avid gamers likely want the best possible rig (in terms of computing power) they can comfortably afford. Hardware vendors try to meet this demand.
Powerful setups are useless without games that benefit from power. This creates a demand for super highres graphics, even though those might not add much to the actual gameplay experience.
https://store.steampowered.com/hwsurvey/Steam-Hardware-Softw...
This is kind of a move of desperation. GPUs are massively parallel so it's straightforward to make performance scale linearly with power consumption. They could always have done this.
Nvidia is used to not having strong competition. Now AMD and Intel are both gunning for them and the market share is theirs to lose. Things like this are an attempt to stay on top. But the competition could just do the same thing, so what good is it?
The RTX 3080 (Samsung 8N) is slower than the RX 6900 (TSMC N7) and has a 20W higher TDP. The RTX 3080 Ti and RTX 3090 are faster but have a 50W higher TDP. Their design isn't magic, they're just compensating for the worse process technology by using more power.
You will need a small DC welding machine for a power supply.
>10 Gbps USB/Thunderbolt are also a pain longer than a desk's length away but not the end of the world. Things like the power button are the easiest, just pop them into some single pin jumper extenders and you're good to go for any reasonable distance.
Any pair of wires will do...
Seeing this makes me worried that maybe it wasn't quite enough.
I have dedicated circuits for both my gaming pc (120v) and the amp (230v). It didn’t cost all that much while we were doing electric work anyway.
The fans are, via extension cables included with extra fans I bought, plugged directly into motherboard, zhe cable pulled trough opening for second gpu connectors.
I think you have that reversed
At some point Intel and Nvidia shall start using their brain. VLSI design is hard, but this is no excuse for making crappy products.
Noctua NH-D15 chromax.Black, Dual-Tower CPU Cooler (140mm, Black) https://www.amazon.com/dp/B07Y87YHRH/ref=cm_sw_r_apan_glt_fa...
Would be able to dissipate 180W probably without even throttling up the fans that much.
That's very likely. I have an NH-D14 with only the center 140 mm fan installed on an i7-3930k overclocked to 4.3 GHz, and it barely ramps up the fan. The loudest fan in my computer is the PSU (an old 600 or 650 W Seasonic). It runs in a "silence-oriented" Define R3 case with closed door.
Intel announces 135 W TDP for that CPU. Under load, OCCT says 160 W.
In my case, the CPU has 40 PCIe lanes, and all PCIe 16x slots on the motherboard are electrically 16x.
---
[0] For example, the i5-8500 only has 16 lanes total: https://ark.intel.com/content/www/fr/fr/ark/products/129939/...
That's not what you buy an i7 for though, is it?
An i7 is typically for "I have work to do when not gaming" or content creators starting out, where as an i5 is more the category for gaming and light productivity / "the family pc".
Getting an i7 exclusively for gaming seems more for bragging rights on "fastest CPU".
A recent example is Battlefield 2042. When I first got the game I had an AMD 3700X, which is no slouch of a CPU. But it could only drive the game at about 70fps, no matter the resolution. After an upgrade to a 5900X, I can run the game at 110-120fps. The strongest i5s would likely struggle to hit a stable 60fps on this game.
Perhaps to specifications that would satisfy someone else. My requirements are ~120fps and 4k. For those requirements, no existing i5 would cut muster.
> But the CPU "limiting" you when you're comfortably above 60fps
I never used the word "limiting" in my comment. Not sure what you're quoting from. That being said, the 3700X was objectively limiting my framerates. It's not a value judgment, it's just an objective fact that such CPUs are inadequate to satisfy my preferences, and those of many other PC gamers. If you have different preferences, that's fine, but it's not really relevant when I'm talking about my own.
I didn't say most gamers are CPU limited, though. I said "Depends on the type of game . . . there are many cases today where a strong CPU is required for gaming." I didn't say most cases.
Conversely, it would be fine to observe that many console gamers have historically been satisfied with 30fps, and therefore PC gamers ought to only "require" a strong -2 gen i5 processor. While you're at it, you could also say that console gamers game at 1080p, so PC players should be satisfied with that as well. And you'd be right, under a certain configuration of preferences, and a certain interpretation of the word "require."
or from a different perspective, if a new CPU can give you a 70% framerate increase, you either spent too much money on your GPU, or you spent too little on the original CPU. bottlenecks that severe are a sign that you have misallocated your budget.
There seems to be for example optical TB3 cables, maybe something like this could be part of the solution.
https://www.macrumors.com/2020/03/26/optical-thunderbolt-3-c...
I'm sorry but it reads more like an ad. And just because it often needs to be said, I have no brand loyalty here, and mostly think brand loyalty in this space is kind of dumb.
[0]: An example:
```Given its more amenable $409 price tag, it is quite shocking to see the Core i7-12700K deliver such a stunning blow to the $549 Ryzen 9 5900X in threaded work, highlighting the advantages of the x86 hybrid architecture.```
I'd love to save $140 and get better performance, but where am I getting a motherboard that doesn't eat up most of that savings? Maybe it exists, but it's not mentioned. That's fine, it's a CPU review, but then leave the dollar figures out of it if they aren't complete.
At the moment, if you're living a US city with a Microcenter, you can apparently get Asus Z690M-Plus Prime DDR4 for $170 if you buy it together with a CPU.
(Aside from gaming, there's definitely plenty of wins from going to higher end CPUs, but I'm curious about the gaming case specifically.)
Though at the low end, factorio is also a really well optimised game and can run a 100spm factory at 60ups pretty comfortably on a mid range cpu. It's only the large multiplayer games or megabases where it starts to hit performance limits.
Cities Skylines is probably the best example of a game where the average player has big gains in performance in regular play they could get from a cpu upgrade.
There are, but megabase (>1k spm) territory is where players start making design decisions around ups impact like avoiding large logistics zones or avoiding heat pipes.
There are a few games that do strongly benefit from a better CPU, but those are the exception, and they still have significant dismissing returns
Fwiw my simple reaction time is 220ms on a basic 60hz display vs 170ms on a top end 360hz, so being able to push those frames has a significant quality of life delta, even though I’m unable to hit 360hz in most games due to cpu bottlenecks on my system.
There’s many factors to consider in TCO like performance / watt as well combined with motherboard ecosystem which has been a historical AMD weak point. So it’s tough to compare CPU pricing on an apples to apples basis even if the CPUs were otherwise exactly the same performance and pricing
Is this referring to pre AM4 socket? Or the faff around Ryzen 5000 not supporting the earlier AM4 motherboards?
- The Intel processors with DDR4 are doing better than with DDR5. This is unlike some of the other benchmarks we have seen. What's happening here?
- In Blender, the 12900k impressively got really close in performance to the 5950x ($200 more expensive) and beat the 5900x (similar price). That's... somewhat unexpected.
- I wonder whether their cooling could keep up all the way through. They mention a custom water cooling loop. What is the performance with a 'typical' cooling setup? How much performance is lost if these CPUs have to throttle to something you can expect a typical boxed cooler to handle?
- Will you actually be able to buy these high-end CPUs? Or are they pure marketing that won't be profitable for Intel and will only be available in small quantities?
Here https://www.tomshardware.com/reviews/hot-spot,365-4.html they "tested" throttling in a somewhat famous burning Athlon video (later revealed they used board with disabled/non functioning thermal cutout) during Intel Pentium 4 payola timeframe. Some quotes:
"AMD did not bless the Thunderbird core with ANY thermal protection whatsoever."
In reality AMD socket certification required thermal cutout, same as Intel for Pentium 3. AMD processors do include thermal diode just like Intel ones.
"Intel's older processor is also equipped with a thermal diode and a thermal monitoring unit"
is a lie. Pentium 3 thermal throttling is performed by the Bios just like in Athlon case. Serendipitously tomshardware picked broken Siemens motherboard for AMD system as recommended by Intel, imagine that.
Tomshardware eventually published a non-retraction retraction after AMD pointed out Siemens lie/defect (not adhering to socket certification requirements) and showed proper AMD setup safely shutting down. Of course you cant read that one because its buried down, excluded from wayback machine and deadlinked http://www.tomshardware.com/column/01q4/011029/index.html
(And a larger gaming market/audience over the past four or so years- gaming on mobile / live game streaming)
More $ / TAM = more R&D and competition in the space. (And more power/capabilities available for gaming may just be a nice byproduct )
As much as I dislike Apple's overall business practices, they are 100% correct about ARM being the future of high performance, high efficiency computing.
And we need high efficiency in a world that is becoming starved of energy as more and more people get connected every day.
Intel, which hit a brick wall about 10 years ago when the competition disappeared due to AMD's Bulldozer, is finally responding to both of those chips from last year and while it has been able to pull some decent performance improvements for once it definitely dipped into the power bucket for a lot of that gain. I wouldn't count that as proof x86-64 is unable to compete in high performance high efficiency computing anymore - I'd hold my judgement until Zen 4 and "M2" (or whatever it will be called) release around this time next year as it seems each will be a really big update for each chip. Who knows, maybe Intel will have something new by then too now that they have actual competition again.
Size (mm^2) TDP (w)
Apple APU
M1 119 18
M1 Pro 251 ??
M1 Max 425 90 (peak -- not TDP)
AMD CPU/APU
4800U 156
5800U 180
5800X 206
5950X 286
Epyc Rome 1008
Intel CPU/APU
Ice Lake 123
Tiger Lake 146
Comet Lake 206
Rocket Lake 276
Alder Lake S 209
Alder Lake M 215
AMD GPU
Vega 20 331 300
Navi 14 158 150 (80 mobile)
Navi 10 251 235 (120 mobile)
Navi 23 237 160 (100 mobile)
Navi 22 335 230 (145 mobile)
Navi 21 520 300 (no mobile)
Nvidia GPU
TU106 445 175
TU104 545 250
TU102 754 280
GA106 276 170 (115 mobile)
GA104 392 290 (150 mobile)
GA102 628 350 (no mobile)
M1 Pro has worse multi-threaded integer performance than a 5950, but better floating point performance. It does this in a 251mm^2 die vs a 286mm^2 die AND while most of the die is used up by the GPU.
M1 Max has the equivalent of GA104 tacked on to its 8 CPU cores (plus 2 little cores). AMD's equivalent would be a 5800H plus a Navi 22 mobile. Together, those would use more than double the power while being 100mm^2 larger.
- GPU on the M1 Pro/Max is really impressive but irrelevant
- M1 Pro/Max came out a year later than the CPUs they are being compared to here
- The M1 takes 133% the transistor count on a generation newer TSMC node to get as far past the 5800 as it does
So from a CPU architecture perspective I really can't write x86 as having hit a brick wall to the point it will never be able to keep up with ARM, that's just not what the M1 has shown.
From an overall perspective though yes the M1, particularly Pro/Max with their larger GPUs (can't wait for mine to arrive!), are hands down the best overall systems of their classes right now I just don't think that has as much to do with the CPU being ARM instead of x86 as GP made it out to be.
It makes their performance look good relative to M1 for now. But what happens when Apple push out their next chip design? Is Intel going to up the watt usage by a huge amount again?
Nah , this is a dead end. Alder Lake is a blind alley.
A15 performance cores had significant gains and really improved integer performance all while reducing power consumption slightly.
A15 efficiency cores are even more amazing. They got something like 30-40% faster. They are now roughly as fast per clock as the big cores in x86 systems while still using only a fraction of the power of the big cores. This roughly translates into 8 full-size x86 cores, but with 4 of those being even faster.
Then there's the massive cache (32mb on mobile and probably 48/64 on M2) to further push down power consumption. They'll also be bumping the GPU by 30% (more if it doesn't throttle).
AMD really needs to either pick back up their ARM designs or start heavy investment into RISC-V designs.
And I'll believe that they're off of their node formally known as 10nm when I see it. It's been close to ten years straight that Intel has been slipping their public deadlines for bringing a new node up.
So, yeah, for a company that used to be a full node ahead of everyone, there's some serious stagnation going on.
[1] https://www.extremetech.com/computing/311275-intel-doesnt-wa...
Because the M1 Max is not less than half
https://browser.geekbench.com/v5/cpu/compare/11136167?baseli...
A15 has already been out and spoiled what's coming in M2.
Size
M1 is 119mm^2 and M2 is probably at most around 130-140mm^2 (depending on cache and RAM controllers) which is still smaller than 5800U at 180mm^2 and much smaller than Alder Lake mobile at 215mm^2.
Cache
A14 and M1 have 16mb of SLC (system level cache). A14 has 8mb L2 while M1 has 12mb. A15 doubled to 32mb SLC and jumped from 8 to 12mb L2. I suspect M2 keeps the 32mb. 12 or 16mb of cache will depend on the performance increase, but I suspect they'll stick with 12mb.
5800U has 4mb of L2 cache and 16mb of L3 cache. Alder Lake desktop has 14mb L2 cache and 30mb L3 cache (dropping to 9.5mb L2 and 20mb L3 on the 12600 and 7.5mb L2 and 18mb L3 on the leaked 12400).
Power
Peak power for M1 is around 15-20w absolute max power. A15 improved power efficiency 17% over A14, so we could actually see power usage improve. Those massive caches may slightly increase chip power, but they decrease memory accesses which radically decrease total system power. The chip running at far lower peak frequencies further helps to keep cache power consumption under control.
5800U can hit upwards of 50-60w peak power to keep the stated turbo speeds. Current Intel chips do this too and I suspect that Alder Lake won't be any different here.
CPU Performance
A15 P-core increases 8-10% (up to 37% on some things) and a lot of that was integer improvements. A15 E-core increased performance by about 30%. Their actual performance is close to Zen 2. Overall, performance in A15 was around 20% better than A14. GPU performance also jumped 32%.
M1 already handily beats 5800U in pretty much everything. I'd bet heavily that M1 beats the little Alder lake. I'm less sure about the bigger alder lake because I don't know what frequencies it will be able to sustain (I'd bet it still beats out M2 though).
GPU Performance
M1 shares SLC between the CPU and GPU. This works much like AMD's Infinity fabric to reduce necessary bandwidth and allows a pretty big increase in GPU performance over what would otherwise be possible.
AMD hit the iGPU wall a few years ago. They've slowly increased GPU resources as DDR bandwidth has increased, but that's about it. I'm really looking forward to Zen 4/5 + RDNA 2/3 + Infinity Cache, but I don't know when that will actually arrive. I've seen iGPU tests of the U770 on Alder Lake desktop and it's not even as fast as last-gen Iris Xe. The mobile variant will have 3x the GPU resources, but it still only looks to be 30% faster than last-gen and then only if it isn't bandwidth limited. This seems possible with DDR5, but I'd guess it's capped at last-gen performance on DDR4 systems.
But Intel's QA has been slipping for a long time, with buggy instructions in shipped chips. (Anybody remember transactional memory?) Have they turned that around, yet? That seems harder than getting better performance on a new process node.
I mean yeah if you're on a budget and plan to sit there with a ~230 watt all core AVX stress test benchmark running 24/7/365 for a couple of years it might matter but in that case you're probably not looking for the fastest high frequency gaming CPU in the first place. Or you're building a custom laptop and would like a long lasting battery, sure bad idea to stick a desktop gaming CPU in. Same when it comes to picking a CPU for the cheap family computer that needs to cost in total what this CPU alone does.
Edit: Holy cow, you can use ddr4!
The downside of LGA 1700 is that budget boards don't exist right now. You can pick up a B550 board any day of the week for under $100, sales as low as $60. Z690 is all $200+.
The Z690 is a nicer platform for sure, especially for peripheral connectivity.
I thought DDR5 brought a 2× increase in throughput? It didn't help that much, it seems?
Eventually ddr5 will outpace DDR4, but right now it's the awkward transition time. Not unlike early DDR4 vs. ddr3
If you have a specific task that is extremely memory bandwidth constrained, it's great. But then again, you are probably looking at something with 8 channels if it is that big of an issue to you.
What I don't understand well enough to know the impacts, is how latency plays into specific workloads.
2ch Alder lake is about 70GB/s[0]
4ch Threadripper is about 85GB/s
8ch/64c Threadripper Pro/Epyc is about 140GB/s[1]
0: https://hothardware.com/reviews/intel-12th-gen-core-alder-la...
1: https://www.anandtech.com/show/16805/amd-threadripper-pro-re...
4800MHz DDR4 even seems to be cheaper than 4800MHz DDR5.